~"U'S -/
. - l.'_._~_:.!t--·<'~':':~ ..; ,;
.. ~UCT:C:':< _ .. ..' ;',
University Free State
l,mllll~l~t~I~~~",~II~1!
Universiteit Vrystaat
INTEGRATING MICRO-FLOOD IRRIGATION WITH IN-FIELD
RAINWATER HARVESTING
By
SABELO SICELO WESLEY MAVIMBELA
A thesis submitted in accordance with the requirements for the
Philosophia Doctor Degree in the Fuculty of Natural and Agricultural
Sciences, Center of Sustainable Agriculture, Rural Development and Extension at the
University of the Free State, Bloemfontein, South Africa.
February 2012
Promoter: Prof. L. D. van Rensburg
Declaration
I declare that the thesis hereby submitted by me for the Phoilosophiae Doctor in
Sustainable Agriculture at the University of the Free State is my own
independent work and has not been previously submitted by me to another
University/Faculty. I further cede copyright of the thesis in favor of the
University of the Free State.
Signature ~ : .
Date: February 2012
Place: UFS, Bloemfontein, South Africa
Acknowledgements
Giving thanks and acknoweldgements to the many individuals and institutions that supported
this work is a humbling exercise. Iowe gratitude to the following:
• Study leader, pioneer and promoter Professor L. D. van Ransburg for his academic
assistance, and invaluable contributions from conception of the problem to the final
work of the thesis. His unreserved, timely and academic devotions towards this study.
• Professor A. H. Cloot from the Department of Mathematics in the University of the
Free State for his guidance and unreserved contributions that resulted to the
publications of two articles from this work.
• Management and technical staff of Parady's Experimental Farm, of the University of
the Free State for offering accommodation, support and assistance during the time of
field experiments.
• Strategic Academic Cluster: Water Management in Water Scarce Areas of the
University of the Free State. Bloemfontein, South Africa.
• Technical team at Kenilworth Experimental Farm for their assistance in oven-drying
maize samples from the field experiments.
• Department of Soil Crop and Climate Sciences of the University of the Free State that
made it possible for the laboratory experiments to be carried out successfully.
• Strategic Academic Cluster; Water Management in Water Scarce Areas for their
financial support during the period of study.
• Ms Liesl van der Westhuizen for her editorial assistance and constructive suggestions
during the writing of publications form this study.
• Family back home for their unreserved support and encouragement.
• To the church for their prayers and spiritual inspirations.
ii
Abstract
The mam aim of the study was to integrate micro-flood irrigation (MFI) with in-field
rainwater harvesting (IRWH). The MFI is a short furrow irrigation system that relies on small
inflow rates to mitigate the effect of dry spells in crop fields. The IRWH is an in situ based
rainwater harvesting technique that harvests rainfall in the form of runoff between crop rows
and then concentrates it in the basin area. Given the increased rainfall variability and
evaporation (Ev) in the semi arid areas of the central Free State Province of South Africa, the
merging of these two technologies is hypothesized to be able to stabilize soil water storage
during rainfall and dry spell periods in areas with access to limited irrigation water.
The developments in the study were divided into three phases. The first phase dealt with
characterization of pedological and hydraulic properties of the soils earmarked for IRWH at
the University of the Free State, 20 Km, south of Bloemfontein. These soils were represented
by the Tukulu, Sepane and Swartland soil types with the first two forms also referred as
Cutanic Luvisols and the latter as Cutanic Cambisols of the Reference Soil Group. These
soils were similar only in the orthic A- horizon. The Tukulu had developed structure only in
the prismatic C-horizon and for the Sepane it was in the pedocutanic B- and prismatic C-
horizons. The Swartland had a cambic structure in the pedocutanic B-horizon. Corresponding
hydraulic properties, soil water characteristic curve (SWCC) and hydraulic conductivity for
saturated (Ks) and unsaturated conditions (K-8) were determined using in situ and laboratory
procedures for internal drainage (ID) and evaporation (Ev) conditions. Parametric models
were used to describe SWCC and to predict K-8 relationships. Model descriptions of SWCC
were satisfactory. Predictions of K-8 were only accurate at near saturation, but HYDRUS-ID
optimization program had better predictions. Matric suction gradients corresponding to the
draining soil profile were found to fall within the matric suction range of 0 to -10 kPa.
Drainage rate of 0.001 mm hour" corresponded to drainage upper limit (DUL) and deep
drainage (DD) losses proportional to 1 % of annual rainfall over the fallow period. The
Tukulu, Sepane and Swartland soil types had respectively total DD losses of 21, 20 and 52
mm and evaporation losses of 43, 51 and 70 mm. The Ks corresponding to the C-horizons of
these soils was 9.6, 1 and 77 mm hour". During ID and Ev the K-8 functions especially for
horizons with a clay content range of 26 to 48 % dropped by several orders of magnitudes,
while SWC changed with a narrow margin. At the evaporating surface matric suction of
magni tude greater than -1500 kPa were approximated.
iii
The second phase compared four inflow rates (20,40, 80 and 160 L min-I) based on surface
and subsurface irrigation characteristics carried out on the Tukulu soil due its low DD and Ev
losses. A single irrigation on a 90 m closed ended furrow and measurements taken at every
10 m furrow distance for advance and opportunity times, stream flow depth, and SWC before
and after the irrigation. Infiltrated depths predictions from HYDRUS-2D software were
satisfactory from all inflow rates. Distribution uniformity (DU) was higher (~ 0.89) at 30 m
furrow distance from all inflow rates and the smaller inflow rate much easier to handle.
Vertical redistribution was characterized at each of the 10 m furrow distance covered with a 2
m x 2 m polythene sheet to prevent Ev. Over the 455 hours of redistribution agreement
between measured and predicted SWC from HYDRUS-2D software varied with depth and
furrow length. Low vertical redistribution (Vz) from all inflow rates was attributed to the
restrictive prismatic C-horizon. Higher rates of Vz were observed within the 0-600 mm
profile domain for the small inflow rates and at 0-850 mm for the large inflow rates.
The last phase dealt with the integration of MFI with IRWH, carried out on a 3 x 3 split plot
factorial with four blocks in a complete randomized design experiment. Each plot had five 30
m long furrows and a pair of neutron access tubes installed in each plot at the centre of the
basin and runoff area. The main treatments were runoff strip width (RSW; 1 m, 2 m and 3 m)
and water regime (WR): dryland (DL), supplemental (SPI) and full irrigation (FI). No till and
basin tillage was used to prepare the RSW and the 1 m standard basin area (BA). The BA was
further smoothed with a ridger for uniform distribution of the advance stream flow. A 120
day maturing maize variety was used. A record of rainfall and SWC was kept. The 40 L min-I
inflow rate for 15 minutes irrigation times on a fixed schedule for full and supplemental
irrigation, provided by the BEWAB+ irrigation software was used. A soil water balance
(SWB) procedure was developed to evaluate the effect of the RSW and WR treatments on the
gains and losses in soil water storage. Evapo-transpiration (ET) was partitioned into Ev and
transpiration (T) by a ~-parameter based on plant canopy area. Findings showed that SWB
components were affected by the main effect from the RSW and WR. The 1 m RSW had the
total biomass and grain yields that were respectively 21 % and 45 % higher than the 2 m
RSW, and 35 % and 89 % higher than the 3 m RSW. Total biomass and grain yields from full
and supplemental irrigation were 200 % and 76 % higher than the DL. Though tested for a
single season the combination of 1 m RSW and full irrigation produced optimum crop yields
and WUE for the newly merged MFI-IRWH water management system and is ready to be
used by small scale farmers who have access to irrigation water.
iv
Contents
Declaration ii
List of Tables xi
List of figures xiv
List of Appendices xvii
CHAPTER 1 1
INTRODUCTION 1
1.1 Motivation 1
1.2 Objectives of the study 3
1.3 Layout of the thesis 5
1.4 References 6
CHAPTER 2 8
LITERATURE REPORT 8
2.1 Introduction 8
2.2 Review on runoff farming management practices 8
2.2.1 Background 8
2.2.2 Classification of runoff farming systems 10
2.2.2.1 Micro-catchment rainwater harvesting technologies 10
2.2.2.2 Macro-catchment rainwater harvesting technologies 14
2.2.2.3 Supplemental irrigation 16
2.2.2.3.1 Irrigation methods for supplemental irrigation 16
2.3 Review on soil physical principles 18
2.3.1 Background 18
2.3.2 Soil physical properties 19
2.3.3 Soil hydraulic properties 21
2.3.3.1 Soil water content 21
2.3.3.1.1 Soil water content measurement... 22
2.3.3.2 Soil water characteristic curve 23
2.3.3.2.1 Measurement of the soil water characteristic curve 23
2.3.3.2.2 Mathematical functions of the soil water characteristic curve 25
2.3.3.3 Soil hydraulic conductivity 26
2.3.3.3.1 Determining soil hydraulic conductivity 26
2.3.3.3.2 Saturated hydraulic conductivity 26
2.3.3.3.3 Unsaturated soils hydraulic conductivity 26
2.3.4 General flow equation 27
2.4 Gap in knowledge 29
v
2.4 References 32
CHAPTER 3 42
IN SITU EVALUATION OF INTERNAL DRAINAGE IN LAYERED SOILS (TUKULU,
SEPANE AND SWARTLAND) .42
Abstract 42
3.1 Introduction 43
3.2 Material and methods 45
3.2.1 Site location and soil classification .45
3.2.2 In situ experimental set up and measurements .45
3.2.2.1 Saturated hydraulic conductivity .45
3.2.2.2 Instantaneous soil water measurement 46
3.2.3 Laboratory experimental setup and measurements .47
3.2.4. Data analysis 48
3.2.4.1 Estimation of unsaturated hydraulic conductivity .48
3.2.4.2 Statistical analysis .49
3.3 Results 49
3.3.1 Pedological properties 49
3.3.2 8-h relationship of soil horizons 53
3.3.3 K-8 relationships of soil horizons 56
3.3.4 8-T relationship of soil horizons 59
3.3.5 Total drainage and flux rates 62
3.4 Discussion 64
3.5 Conclusions 66
3.6 References 68
CHAPTER 4 73
EVALUATING MODELS FOR PREDICTING HYDRAULIC CHARECTERISTICS OF
LAYERED SOILS 73
Abstract 73
4.1 Introduction 74
4.2 Material and methods 76
4.2.1 Experimental site and location 76
4.2.2 Theory 76
4.2.3 Experimental set up and measurements 78
4.2.3.1 Soil profile classification 78
4.2.3.2 Soil particle distribution and bulk density 78
4.2.3.3 Saturated hydraulic conductivity 79
vi
4.2.3.4 Internal drainage experiment 79
4.2.3.5 Measurements of the soil water characteristics curve 80
4.2.3.5.2 Estimating unsaturated hydraulic properties for field based K-coefficient 80
4.2.4 Statistical analysis 82
4.3 Results 82
4.3.1 Soil water characteristics curve 82
4.3.2 Predicting K-coefficient from soil water characteristics curve 86
4.3.3 Parameter optimisation for HYDRUS -ID application 91
4.4 Discussion 94
4.5 Conclusions 97
4.6 References 99
CHAPTER 5 104
IN SITU EVALUATION OF EVAPORATION IN LAYERED SOILS (TUKULU, SEPANE AND
SWARTLAND) 104
Abstract 104
5.1 Introduction 105
5.2. Material and methods 107
5.2.1 Site location and soil classification 107
5.2.2 Soil water characteristic curve 107
5.2.3 Determination of actual evaporation 108
5.2.4 Data processing and analysis l 09
5.2.4.1 Soil water characteristic curve 109
5.2.4.2 Estimation of evaporative flux 109
5.2.4.3 Estimation of unsaturated hydraulic conductivity 109
5.2.4.4 Statistical analysis 110
5.3 Results 110
5.3.1 Soil water characteristic curves 110
5.3.2 Change in profile water content during evaporation 114
5.3.3 K-8 relationships of soil horizons 114
5.4 Discussion 117
5.5 Conclusions 121
5.6 References 123
CHAPTER 6 125
DETERMINING THE OPTIMUM INFLOW RATES FOR MICRO-FLOOD IRRIGATION ON
THE TUKU~LUSOIL 125
Abstract 125
vii
6. I Introduction 126
6.2 Material and methods 127
6.2.1 Site location 127
6.2.2 Experimental design and measurements 129
6.2.3 Simu1ations and inverse analyses 129
6.2.4 Experimental data analysis 131
6.3 Results 133
6.3.1 Surface flow characteristics 133
6.3.2 Subsurface soil water content.. 133
6.3.3 Subsurface soil water distribution 137
6.3.4 Irrigation infiltration functions 138
6.3.5 Average infiltration depth and measures of performance 145
6.4 Discussion 147
6.5 Conclusion 151
6.6 References 153
CHAPTER 7 156
CHARECTERISING VERTICAL REDISTRIBUTION ON IRRIGATED SHORT FURROWS IN
THE TUKULU SOIL 156
Abstract 156
7.1 Introduction 157
7.2 Material and methods 159
7.2.2 Experimental design and measurements 160
7.2.3 Description of the soil hydraulic properties 161
7.2.4 Prediction of two-di mensional redistribution 162
7.2.5 Experimental data analysis 163
7.2.6 Statistical analysis 164
7.3 Results 164
7.3.1 Parameter optimization and HYDRUS-2D inverse solution 164
7.3.2 Changes in soil water content during redistribution 167
7.3.3 Estimated effective K-coefficient during redistribution 171
7.3.4 The rate of soil water redistribution 177
7.4 Discussion 178
7.5 Conclusion 182
7.6 References 183
CHAPTER 8 186
viii
INTEGRATING MICRO-FLOOD WITH IN-FIELD RAINWATER HARVESTING :(i) SOIL
WATER BALANCE PROCEDURE AND APPLICATION 186
Abstract 186
8.1 Introduction 187
8.2 Procedures for solving IRWH soil water balance 189
8.2.1 Soil water storage 189
8.2.2 Harvested runoff 190
8.2.3 Irrigation 190
8.2.4 Deep drainage 191
8.2.5 Evapo-transpiration 191
8.2.6 Statistics 193
8.3 Results and discussions 193
8.3.1 Illustrations of the soil water balance procedure 193
8.3.2 Effect of runoff strip width and water regime on the soil water balance components 197
8.5 Conclusion 209
8.6 References 211
CHAPTER 9 214
INTEGRATING MICRO-FLOOD WITH IN-FIELD RAINWATER HARVESTING: (ii) MAIZE
YIELD AND WATER USE EFFICIENCY 214
Abstract 214
9.1 Introduction 215
9.2 Material and methods 217
9.2.1 Description of experimental site 217
9.2.2 Experimental design and layout 217
9.2.3 Planting and agronomic practices 219
9.2.4 Irrigation design and application 219
9.2.5 Field measurements 220
9.2.6 Estimation of soil water balance components 220
9.2.7 Crop yields and biomass 221
9.2.8 Water use efficiency 222
9.2.9 Statistical analysis 222
9.3 Results 222
9.4 Discussion 225
9.5 Conclusion 228
9.6 References 230
CHAPTER 10 234
ix
PERSPECTIVE ON RESEARCH 234
10.1 Introduction 234
10.2 Soil hydraulic properties 234
10.3 Micro-flood irrigation designs and evaluation 235
10.4 Integrating MFI with IRWH 236
10.5 Thesis contributions and conclusions 237
Appendices 239
x
List of Tables
Table 2.1 Rainwater harvesting practices and their management options (Rockstrom et al.,
2007) 12
Table 2.2 A summary of reviewed local water management studies related to IRWH showing
work done (+) and not done (-) on selected water balance, texture, soil hydraulic
properties (SHP), special tools or models and nature of water use efficiency (WUE) .
................................................................................................................................... 31
Table 3.l Summary of the physical and chemical characteristics of the three soil types 50
Table 3.2 8-h regression functions of soils horizons for the 0 -1000 mm suction range 54
Table 3.3 Student's t-test for differences between horizons 8-h regression coefficients 54
Table 3.4 K-8 relationships linear regression functions for the three soil types 58
Table 3.5 Homogeneity test for the three soils horizons 59
Table 3.6 Student's t-test for differences between horizons K-8 regression coefficients 59
Table 3.7 8-T relationships linear regression functions for the three soil types 61
Table 3.8 Student's t-test for the differences between horizons 8-T regression coefficients .. 61
Table 3.9 Amount of soil water that drained away from horizons 62
Table 4.1 Summary of the physical characteristics of the three soil types 81
Table 4.2 Fitting models to the hydraulic parameters of the SWCC for the Tukulu, Sepane
and Swartland form soils 84
Table 4.3 Statistical measure of fit for conductivity based parametric models on in situ K-
Coefficient for the Tukulu, Sepane and Swartland soil horizons 90
Table 4.4 Optimised parameters for the fitting of in situ K-coefficient from the
Tukulu,Sepane and Swartland soil horizons using HYDRUS -ID 92
Table 5.l 8-h regression functions of soils horizons for the 10 -100 kPa matric suction range .
................................................................................................................................. 112
Table 5.2 Student's t-test for differences between soil horizons regression coefficients 112
Table 5.3 K-8 relationships linear regression functions for the three soil types 116
Table 5.4 Homogeneity test for the K-8 relationships of the horizons of the three soils 117
Table 5.5 Student's t-test for the regression coefficient comparison between horizons 117
Table 5.6 Accumulative evaporation from the soils horizons and profile 119
Table 6.1 Summary of the soil physical of the Tukulu soil form 128
Table 6.2 Soil hydraulic characteristics of the Tukulu soil. 130
xi
Table 6.3 Stream depth and opportunity time of stream-flow along furrows under different
inflow rates 135
Table 6.4 Optimised parameters to improve model's predictions 135
Table 6.5 Statistical measure of fit from measured and predicted soil water content at three
furrow distance measuring points 137
Table 6.6 Bartlett's homogeneity test between measured and predicted infiltration functions
for the different inflow rates 138
Table 6.7 Average infiltrated depth and performance indicators for different furrow lengths
and inflow rates 146
Table 7.1 Summary of the soil physical properties of the Tukulu soil form 159
Table 7.2 Initial estimates of soil hydraulic parameters 162
Table 7.3 Optimised parameters to improve HYDRUS-2D model prediction 165
Table 7.4 Measure of dispersion between the measured and predicted soil water content (mm
mm") 170
Table 7.5 Homogeneity test for the effective K-coefficient calculated from measured and
predicted soil water contents 176
Table 7.6 Rate of vertical redistribution (Vz) expressed as a function of infiltrated depth (êz) .
................................................................................................................................. 178
Table 8.1 Regression functions of rainfall (P) and runoff (Rin) relationships from different
runoff strip widths 190
Table 8.2 Illustration of the soil water balance procedure using estimated water processes
over the production season from the basin, runoff and plot area of in-field rainwater
harvesting with a 2 m runoff strip under micro-flood supplemental irrigation ....... 195
Table 8.3 Summary of the analysis of variance depicting the effect of runoff strip width
(RSW) and water regime (WR) on selected soil water balance SWB components .
................................................................................................................................. 199
Table 8.4 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance
(SWB) during plant establishment growth stage 202
Table 8.5 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance
(SWB) during plant establishment growth stage 203
Table 8.6 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance in
the vegetative growth stage 204
Table 8.7 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance in
the reproductive growth stage 205
xii
Table 8.8 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance in
the ripening growth stage 206
Table 9.1 Summary of the physical and chemical characteristics of the Tukulu and Swartland
soil types found in the experimental area 218
Table 9.2 Intra-row plant spacing at planting time (mm) 219
Table 9.3 Summary of analysis of variance depicting the effect of runoff strip width (RSW)
and water regime (WR) on seasonal soil water components, crop yield and water use
efficiency (WUE) 223
Table 9.4 Effect of runoff strip width and water regime on seasonal water components and
crop yield 224
xiii
List of figures
Figure 2.1 A diagrammatic representation of the in-field rainwater harvesting technique
(Hensley et al., 2000) 13
Figure 2.2 A severely water stressed maize crop at tasseling following a dry spell that
persisted more than 3 weeks at Parady's experimental farm under in-field rainwater
harvesting 15
Figure 2.3 Micro-flood irrigation technologies as presented to agriculture experts in
Kouebokkeveld, South Africa (2003) 17
Figure 2.4 Application of supplemental furrow irrigation along crop row basins (a) and plants
supported by supplemental irrigation among crops damaged by insufficient rainfall
(b) (Adapted from Rocktrom et aI., 2007) 18
Figure 3.1 (a) Profile of the Tukulu and (b) mottles present in the C-horizon 51
Figure 3.2 (a) Profile of the Sepane and (b) mottles present in the B-horizon 51
Figure 3.3 (a) Profile of the Swartland and (b) dolorite intrusion present in the B-horizon 53
Figure 3.4 8-h relationships from the (a) Tukulu, (b) Sepane and (c) Swartland soil types 55
Figure 3.5 K-8-relationship from the (a) Tukulu, (b) Sepane and (c) Swartland soil types 57
Figure 3.6 8-T relationships from the (a) Tukulu, (b) Sepane and (c) Swartland soil types 60
Figure 3.7 Drainage flux from the (a) Tukulu, (b) Sepane and (c) Swartland soil types 63
Figure 4.1 Measured and fitted soil water content (SWC) and matric suction relationships for
the Tukulu (a), Sepane (b) and Swartland (c) 83
Figure 4.2 Comparison of K-coefficient from in situ and fitted by retention models for the
Tukulu A (a), B (b) and C (c) diagnostic horizons 87
Figure 4.3 Comparison of K-coefficient from in situ and fitted by retention models for the
Sepane A (a), B (b) and C (c) diagnostic horizons 88
Figure 4.4 Comparison of K-coefficient from in situ and fitted by retention models for the ..89
Figure 4.5 Comparison of in situ and fitted K-coefficient for the Tukulu (a), Sepane (b) and
Swartland (c) diagnostic horizons A(i), B(ii) and C (iii) using HYDRUS-1D
optirnised parameters 93
Figure 5.1 8-h relationship curves for the (a) Tukulu, (b) Sepane and (c) Swartland soil
forms 111
Figure 5.2 Change in soil water content during the evaporation period for the (a) Tukulu, (b)
Sepane and (c) Swartland soils 113
Figure 5.3 K-8-relationship from the (a) Tukulu, (b) Sepane and (c) Swartland soil.. 115
xiv
Figure 5.4 Evaporative flux from the (a) Tukulu, (b) Sepane and (c) Swartland soil layers. 118
Figure 6.1 (a) Profile of the Tukulu and (b) mottles present in the C-horizon 128
Figure 6.2 Fine element mesh and boundary conditions assigned inside the furrow (constant
head), zero flux at the surface and vertical sides and free drainage at the bottom of
the profile 131
Figure 6.3 Measured advance and recession times for the 20 (A), 40 (B), 80 (C) and 160 (D)
L min·l 134
Figure 6.4 Measured and predicted soil water content (SWC) for the profile depth layers (0-
300; 300-600, and 600-850 mm) before and after irrigation HYDRUS-2D software
from the 20, 40,80 and 160 L min'l inflow rates at three furrow distances 136
Figure 6.5 Subsurface soil water distribution following irrigation at 5 m, 35 m, and 55
furrowlength for the 20 L mm'l inflow rate 139
Figure 6.6 Subsurface soil water distribution following irrigation at 5 m, 55 m, and 85 m
furrow length for the 40 L mm'l inflow rate 140
Figure 6.7 Subsurface soil water distribution following irrigation at 5 m, 55 m, and 85 furrow
length for the 80 L mm' I inflow rate 14]
Figure 6.8 Subsurface soil water distribution following irrigation at 5 m, 55 m, and 85m
furrow length for the 160 L mm-I inflow rate 142
Figure 6.9 Distribution of infiltrated depth along the furrow length for the 20 (A), 40 (B), 80
(C) and 160 L min'l (D) inflow rate 143
Figure 6.10 Average furrow irrigation functions for predicted (solid line) and for measured
(broken line) corresponding to the different constants (a) adjusted to conserve mass
for the 20 (A), 40 (B), 80 (C) and 160 (D) L min-I inflow rates 144
Figure 7.1 (a) Profile of the Tukulu and (b) mottles present in the C-horizon 160
Figure 7.2 Shading of the furrow section measurement stations at 10 m intervals 161
Figure 7.3 Measured and predicted soil water content (SWC) of the soil profile during
redistribution from the 20, 40, 80 and 160 L min-I inflow rates at various furrow
distances covered by the advance stream 166
Figure 7.4 Measured and predicted soil water content (SWC) for the initial and selected time
intervals during redistribution at different furrow distances and soil profile depths
for the 20, 40,80 and 160 L min-I inflow rates 168
Figure 7.5 Effective K-coefficient from the furrow treated with 20L min-I inflow rate for the
0-600 and 0-850 mm infiltrated depths at the 5 m, 35 m and 55 m distances 172
xv
Figure 7.6 Effective K-coefficient from the 40 L min-I inflow rate at 0-600 and 0-850 mm
infiltrated depths of the 5 m, 55 m and 85 m furrow distances 173
Figure 7.7 Effective K-coefficient from the 80 L min-I inflow rate at 0-600 and 0-850 mm
infiltrated depths of the 5 m, 55 m and 85 m furrow distances 174
Figure 7.8 Effective K-coefficient from the 160 L min-I inflow rate at 0-600 and 0-850 mm
infiltrated depths of the 5 m, 55 m and 85 m furrow distances 175
Figure 7.9 Relationship between the rate of redistribution and infiltrated depth at 0-600 and
0-850 mm profile depth for the 20, 40,80 and 160 L min-I inflow rates 179
Figure 8.1 A diagrammatic representation of the in-field rainwater harvesting technique
(Hensley et al., 2000) 188
Figure 8.2 Approximated ~ parameter representing the fraction of canopy cover during the
production season for the basin area (BA) and runoff area (RA) of the dryland (A)
and irrigation (B) plots 194
Figure 9.1 Seasonal precipitation distribution and its accumulative totals with supplemental
(SPI) and full irrigation (FI) 220
xvi
List of Appendices
Appendix A Profile description of the Tukulu soil form 240
Appendix B Profile description of the Sepane soil form 241
Appendix C Profile description of the Swartland soil form 242
xvii
CHAPTER I
INTRODUCTION
1.1 Motivation
The traditional role of agriculture as a major source of food and income among developing
nations especially South Africa is threatened by growing climatic and soil related challenges.
South Africa is generally a water stress country with about 82 % of its land area classified as
arid to semi-arid. In average the country receives rainfall of 495 mm, well below the global
average of 860 per year (ARC-ISCW, 2005). Arable land constitutes of 16.7 million hectare
of which about 1.5 million is under irrigation mainly on areas with insufficient rainfall to
sustain economic crop yields. Under this circumstance, dryland crop production is critical in
improving food security and is widely practiced in the semi-arid areas with an aridity index
of 0.2 to 0.5 (UNESCO, 1977). However, increased rainfall variability and sensitivity of the
soils to degradation hazards, typical of drier regions of the world, has increased the risk of
dryland food production, especially among small scale farmers.
Food production under the difficult climate and soil conditions existing in the semi-arid areas
has been unsustainable. The majority of rainfall events occur in amounts of less than 20 mm
which are insufficient to recharge the soil profile water storage. It is estimated that most of
this rainfall (40 to 75 % of the annual rainfall) is lost to evaporation mainly from the summer
rainfall areas where 85 % of the dryland food production is practised (Bennie and Hensley,
2001). Rainfalls that significantly contribute to the total seasonal rainfall are few and occur as
outburst of heavy rainstorms that result in high runoff and evaporation losses. The former is
estimated to constitute about 6 to 30 % of annual rainfall (Bennie, Strydom and Vrey, 1998).
These unproductive losses are severe under conventional tillage systems from soils
susceptible to compaction and surface crusting, especially under the impact of rainstorms (du
Plessis and Mostert, 1965). Water induced soil erosion; dry soil water regime and
deterioration of organic matter are some of the major factors that undermine the productive
potential of semi-arid dryland food production systems.
Current and future predictions have indicated that this situation could be worse given that
rainfalls shall decline with greater inter-annual variability (Rockstrom et al., 2007). Globally,
loss of productive land under different climatic change and rainfall scenarios was estimated at
10 to 20 % with 1 to 3 billion people of the world population affected by 2080 (Fischer et aI.,
1
2002). The Sub Saharan Africa was estimated to lose about 12 % of its productive area with
the majority happening in the Sudan-Sahelian zone, which is already subject to high climate
variability and adverse crop conditions. Consistent with these predictions is the growing lack
of interest in farming especially among small scale farmers in the past decades because of
extreme climatic conditions and loss of the land's productive potential (FAO, 2009;
Rockstrom et al., 2007).
However, renewed interest on water management practices that improve rainfall capture and
water productivity has remarkably changed the situation. Popular among this management
practices is rainwater harvesting sometimes referred as runoff farming involving the
harvesting of rainfall in the form of runoff from a larger area and concentrating it to a smaller
area for immediate or later productive use (Oweis et al., 1999; Hensley et al., 2000, Prinz and
Malik, 2002; Anschutz et al., 2003). More attractive for small scale farmers is the in situ
rainwater harvesting techniques that maximise rainfall infiltration and soil water conservation
by using various forms of conservation tillage and cultural practices (Rockstrom et al., 2007).
Reinventing of small-scale water harvesting in South Africa has resulted to the development
of in-field rainwater harvesting technique (IRWH) to address the low productive potential of
layered soils predominant in the semi-arid areas of the Free State Province. Through the use
of basin and no till IRWH is reported to have reduced ex-field runoff to almost zero. Crop
yields increased from 30 to 50 % while rainfall use efficiency increased from 50 to 106 %,
respectively compared to conventional tillage system (Hensley et al., 2000). Similar results
were also reported on homogeneous soils from the Limpopo Province (Mzezewa et al., 2011).
However, concerns of high evaporation losses from the IRWH resulted to the testing of
different mulching materials by Botha (2006). Although the suppression of evaporation under
organic mulches was beneficial to crop growth but this benefit was dissipated by dry spells
that were 14 to 21 days long. Recently, the effect of intererop on IRWH conservational and
water productive potential was investigated and the results on water use efficiently that is
dependent on evaporation was inconclusive (Mzezewa and van Rensburg, 2011).
Integrating of in situ rainwater harvesting with some form of supplemental irrigation to
mitigate the effect of persistent dry spells on crop growth and yields is encouraged m
literature (Oweis et al., 1999; Walker, 2003; Rodrfguez et al., 2004; World bank, 2005;
Rockstrom, 2000; Moult et al., 2009). Increased restriction on available water for irrigation
requires that any access be efficiently used through the choice of efficient method and
2
application regime. Considering the poor resource base of small scale farmers, Austin (2003)
developed micro-flood irrigation (MFI) to enhance water productivity and efficiency from
soils with an inherent dry water regime. Micro-flood irrigation is mainly driven by gravity
and use small inflow rates on short furrows to obtain distribution uniformity as high as 95%,
suggesting its competitiveness to other piped irrigation system. Integration of in situ
rainwater harvesting with some form of micro-flood irrigation has been carried with great
success in East African and South Asian countries (Awulachew et al., 2005; World Bank,
2005; Rockstrom et al., 2007). Given that IRWH and MFI involves runoff management but
of different forms to improve soil water storage and productivity, their integrating could be
very instrumental in advancing food security.
1.2 Objectives of the study
In the light of this background the main objective of the study's overall hypothesis was that
the integration of MFI with IRWH would enhance water productivity. To test this hypothesis
a series of studies were carried with each having its own set of specific objectives as outlined
below.
Study 1 (Chapter 2): This study was titled: "Literature report". The primary objective was to
study the body of knowledge pertaining to the development and application of runoff farming
technologies with specific attention given to in situ rainwater harvesting and flood irrigation
methods.
Study 2 (Chapter 3): This study was titled: "In-situ evaluation of internal drainage in layered
soils (Tukulu, Sepane and Swartland)". The specific objectives were:
(i) to describe the pedological properties that relate to the presence of layering on
the three soil types, and
(ii) to determine the soil water release, unsaturated hydraulic conductivity and
drainage-time functions that characterised the internal drainage outcomes of
layered soils.
Study 3 (Chapter 4): This study was titled: "Evaluating models for predicting hydraulic
characteristics of layered soils". The specific objectives were:
(i) to characterise the soil water characteristic curve of the respective horizons of
the three soil profiles,
3
(ii) to validate the conductivity functions based on the soil water characteristic
parameters for the estimation of in situ unsaturated hydraulic conductivity,
and
(iii) to estimate 111 situ unsaturated hydraulic conductivity from optimised soil
water characteristic curve based parameters.
Study 4 (Chapter 5): This study was titled: "In situ evaluation of evaporation in layered
soils". The specific objectives were:
(i) to describe the amount of soil water available to evaporation by the soil water
release curve;
(ii) to describe the delivery rate of soil water to the evaporating site by the
unsaturated hydraulic conductivity function of soil profile layers; and
(iii) to describe the evaporation flux and accumulative evaporation of the three
soils with respect to their hydraulic characteristics.
Study 5 (Chapter 6): This study was titled: "Determining the optimum inflow rates for micro-
flood irrigation on the Tukulu soil". The specific objectives were:
(i) to characterise the surface water distribution during the advance-supply phase.
Secondly,
(ii) to describe the subsurface water distribution during irrigation with the aid of
the HYDRUS-2D software, and
(iii) to evaluate the efficiencies from the irrigation outcomes.
Study 6 (Chapter 7): This study was titled: "Characterising vertical redistribution on irrigated
short furrows in the Tukulu soil". The specific objectives were:
(i) to estimate the soil hydraulic parameters of the Tukulu soil that describes soil
water movement during redistribution using the Levenberg-Marquardt
parameter optimization algorithm,
(ii) to compare the HYDRUS-2D models predictions with field measured soil
water content and calculated effective K-coefficient, and
(iii) to estimate the relationship between the rate of redistribution and infiltrated
depth from furrows under different soil water regimes.
4
Study 7 (Chapter 8): This study was titled: "Integrating micro-flood with in-field rainwater
harvesting: (i) Soil water balance procedure and application". The specific objectives were:
(i) to develop a procedure for quantifying SWB components for the integrated
MFI-IRWH management practice, and
(ii) to determine the effect of runoff strip width (RSW) and water regime (WR) on
the SWB components over the production season.
Study 8 (Chapter 9):"This study was titled: Integrating micro-flood with in-field rainwater
harvesting: (ii) maize yields and water efficiency".The specific objective was to determine
the effect of runoff strip width (RSW) and water regime (WR) on maize production.
1.3 Layout of the thesis
This thesis consists of 10 chapters. Chapter one deals with the motivation and objectives of
the study while literature review is reported in chapter 2. Individual chapters (3, 4, 5, 6, 7, 8,
and 9) consist of abstracts, introduction, material and methods, results, discussions and
references. Chapters 3 to 9 were prepared in article format with four chapters already sent to
different Journals for publications. These chapters also include acknowledgements. Chapter
10 provide a perspective on the thesis major findings, contributions and conclusions.
5
1.4 References
Anschutz, J., Kome A., Nederlof, M., de Neef, R., van de Van, T., 2003. Water harvesting
and Soil water retention. Sec ed. Agromisa Foundation. Wageningen, Netherlands.
ARC-ASCW, 2005.0verview of agricultural natural resources of South Africa. ARC-ASCW.
Report No. GW/N2004/13. (CD-ROM) ARC-In titute for Soil, Climate and Water,
Pretoria, South Africa.
Austin, C., 2003. Micro flood, a new way of applying waters. http://waterright.com.
22/09/2011, 10.00 a.m (LT).
Awulachew, S. B., Merrey, DJ., Kamara, A. B., Koppen Van B., de Vries P. Boelee E.,
2005. Experiences and opportunities for promoting small scale/micro irrigation and
rainwater harvesting for food security in Ethiopia.
Bennie, A. T. P. and Hensley, M., 2001. Maximising precipitation utilization in dryland
agriculture in South Africa- a review. J. Hydrol. 241: 124-139.
Bennie, A. T. P., Strydom, M. G. and Vrey, H. S., 1998. The use of computermodels or
agricultural water management on ecotope level. Water Re earch Commission report,
TT 102/98, Pretoria, South Africa.
Botha, J. J., 2006. Evaluation of maize and unflower production in a semi-arid area using In-
field rainwater harvesting. Ph.D. (Agric) Di ertation, University of the Free State,
Bloemfontein, South Africa.
du Plessis, M. C. F. and Mostert, J.W.c., 1965. Runoff and oil loss at the Agricultural
Research Centre at Glen.S, Afr. J. Agric. Sci., 8: 1051-1060.
FAO, 2009. How to feed the world in 2050.www.fao.org., 28/01/2012, 10.30 am (LT).
Fischer, G., M. Shah, and van Velthuizen, H., 2002.Climate Change and Agricultural
Vulnerability. Special report forthe UN World Summit on Sustainable Development,
26 August-4 September, Johannesburg. Laxenburg, Austria: International Institute for
Applied Sy terns Analysis.
Fischer G., van Velthuizen H, Shah, M and Nachtergaele, F., 2002. Global Agro-ecological
Assessment for Agriculture in the 21 Century. llASA Research Report. nASA,
Laxenburg, Austria.
Hensley, M., Botha, J. J., Ander on, J. J., van Staden, P. P. and Du Toit, A., 2000. Optimising
rainfall use efficiency for developing farmers with limited access to irrigation water.
Water Research Commission Report no. 878/1/00, Pretoria, South Africa.
6
Moult N. c., Lecler N. 1., and Smithers J. C., (2009). A catchment-scale irrigation systems
model for sugarcane; Model application. WaterSA, 35: 29-36.
Mzezewa, J. and van Rensburg, L. D., 2011. Effects of tillage on runoff from a bare clayey
soil on a semi-arid ecotope in the Limpopo Province of South Africa. Water SA, 37;
1-8.
Mzezewa, J., Gwata, E.T. and van Rensburg, L. D., 2011. Yield and seasonal water
productivity of sunflower as affected by tillage and cropping systems under dryland
conditions in the Limpopo Province of South Africa. Agricultural water Management,
98: 1641-1648.
Oweis, T., Hachum, A. and Kijne, J., 1999. Water harvesting and supplementary irrigation
for improved water use efficiency in dry areas. !WMl Contribution No. 7, System
Wide Initiative on Water Management (SWIM).
Prinz, D., and Malik A. H., 2002.Runoff farming. Institute of Water Resources Management,
Hydraulic and Rural Engineering.Department of Rural Engineering, University of
Karlsruhe. Germany.
Rockstrom, J., Hatibu, N., Oweis, T. Y. and Wani, S., 2007. Managing water in rain-fed and
Designs. Guide and technical document. Department of Biological and irrigation
Engineering, Utah State University agriculture, Part 4 In: Unlocking the potential of
rain-fed agriculture, !WMl, Colombo, Sri Lanka.
Rockstrom, J., (2000). Water resources management in smallholder farms in Eastern and
Southern Africa: An overview. Phys. Chemo Earth, 25: 275- 283.
Rodriguez, J. A, Diaz, A., Reyes J. A. and Pujols R., 2004. Comparison between surge
irrigation and conventional furrow irrigation for covered black tobacco cultivation in
a Ferralsol soil. Span 1. Agric. Res., 2: 445-458.
TheWorld Bank, 2005. Shaping the future of water for agriculture; A source book for
investment in agricultural water management. Rural and Agricultural Development,
Washington D. C. USA.
UNESCO, 1977. World map of desertification. United Nations Conference and
Desertification Report 74/2. United Nations, New York, USA.
Walker W. R., 2003. SIRMOD III; Surface Irrigation Simulation, EvaluationWosten, J.H.M.
and van Genuchten, T. H. M.,1988. Using texture and other soil properties to predict
the unsaturated soil hydraulic functions. Division S-6, Soil water management and
conservation. Soil Sci. Soc. Am. J., 52: 1762-1770.
7
CHAPTER2
LITERATURE REPORT
2.1 Introduction
Development and application of soil water management practices and soil physical principles
for crop production in semi-arid and arid areas are the two aspects of literature study covered
in this study. The aim was to acknowledge available literature and to develop a framework of
body of knowledge upon which the subsequent research studies reported in this thesis shall be
developed upon. In this regard the literature study is presented in three sections. The first
section reviews the water management practices pertaining to runoff farming practices; the
section pays attention to the soil physics theories and concepts applicable in rainwater
harvesting water management systems, and the third section acknowledging gab in
knowledge in the different areas pertinent to this research.
2.2 Review on runoff farming management practices
2.2.1 Background
Runoff farming embraces water management practices that aims at harvesting rainwater in
the form of runoff from a relatively larger area and concentrated it to a smaller area for
immediate or later productive use (Owes et al., 1999, Prinz and Malik, 2002, Anschutz et al.,
2003). The aspect of later productive use mainly refers to the provision of irrigation to
supplement rainfall during times when it is insufficient or persistent dry spells. Evidence of
rainwater harvesting that supported ancient farming communities and desert dwellers could
still be traced across Australia, south western United States, Middle East and Africa
(Rockstrom et al., 2007).
Runoff farming also to be referred as rainwater harvesting in this study is traditionally
practiced in water scarce areas. It is widely practiced over 110 countries that roughly contain
about 40 % of the world population found across the arid and semi arid regions of the world
(Critchley and Siegert, 1991; Rockstrom et al., 2007) where rainfall is low, unevenly
distributed and highly r~ndom. Annual rainfall is concentrated to few but heavy intensity
rainstorms that results to high runoff and evaporation losses. About 6 to 30 % and 60 to 75 %
of annual rainfall is reported to be lost as runoff and evaporation, respectively (Bennie et al.,
1994). Consequently, soil water induced erosion; compaction and dry soil water regime are
8
some of the common factors that reduce semi-arid soils productive potential. However,
rainwater harvesting has provided management practices that control runoff and substitute a
significant portion of evaporation for transpiration. This is achieved by supplementing the
water shortages in the cultivated area by runoff harvested from the catchment area.
Rainfall characteristics, topography and soil hydraulic properties determine the effectiveness
of rainwater harvesting. A wide range of rainwater harvesting technologies is suitable for
areas with annual rainfalls falling within the range 300-700 mm. Some specialized
technologies are adapted to very low rainfall areas as low as 100 mm per annum. Small
catchments are recommended in areas with uneven slopes of greater than 5 %, increased
rainfall seasonality and high runoff coefficient as is the case with clay soils (Owes et al.,
1999, Prinz and Malik, 2002; Anschiitz et al., 2003; Hillel, 2004; Ali et al., 2007). However, in
years when rainfalls are low and erratic small catchments could not be able to store adequate
water to sustain the crop (Rockstrom et al., 2007). Catchments with low runoff turnover
could either be physically treated by surface clearance and compaction or chemically treated
with sodium salts, latex, silicone and asphalt paraffin wax as well as oil emulsions. In certain
cases plastic sheets could be used to optimize harvested runoff per unit area (Hillel, 2004).
The benefits of rainwater harvesting are many and multifaceted. Crop responses to improved
soil water storage and availability has been remarkable (Hensley et al., 2000; World Bank,
2005; Rockstrom, 2000; Rockstrom et al., 2007). Average grain yields under dryland
conditions are about 1500kg ha-I but yields up to 1800 kg ha-I (Botha 2006) and 3100 kg ha-I
(Rosegrant et al., 2002) has been reported from in situ and supplemental irrigation rainwater
management practices. Each 1 % growth in agricultural yields brings about an estimated 0.5
to 0.7 % reduction in number of poor people (World Bank, 2005). Evidence has been shown
that costs-benefit ratio for investing in rainwater harvesting techniques has been high
(Kundhlande et al., 2004; Joshi et al., 2005). In many instances soils rendered unproductive
because of surface desiccation and high salt accumulation as been reclaimed and restored
through rainwater harvesting strategies (Hensley et al., 2000; Awulachew et al., 2005).
However, the expansion of rainwater harvesting for food production should be carried out
judiciously. Runoff also plays a role in partitioning of rainwater in the landscape. It is
responsible for recharging seasonal flows in the major drainage river basins, natural wetlands
and underground waters. Current records indicate that over 75 % of major river basins are
already utilized for human needs with only 25 % available for ecological functions (FAO,
9
2009). Underground water resources are also reported to be approaching depletion in a couple
of decades if abstraction is to be continued at current rates (FAO, 2009). Under these
circumstances, investment in rainwater harvesting should assume a more integrated approach
if food production is to be improved or maintained under the growing climate and soil
limitations.
2.2.2 Classification of runoff farming systems
Different approaches are mainly used to classify runoff farming systems. These approaches
mainly consider the designation of catchment, storage method and nature of harvested runoff.
Generally runoff could be harvested from external and in situ or infield catchments as well as
from roof tops (Pacey and Cullis, ]986). Nasr (1999) classified rainwater harvesting
technologies according to their ability to harvest and store runoff in the soil profile of
cultivated fields or one that harvest and store runoff offsite the cropped area in the form of
surface reservoirs and darns mainly for supplemental irrigation. A different classification is
that provided by Critchely and Siegert (1991) that distinguish rainwater harvesting from flood
water harvesting. The latter depicts the capture of surface runoff within or diverted from the
seasonal streambed towards the cropped area to be stored in the soil profile. Percolation
dams, earth bunds and liman terraces are used in flood water harvesting in areas with annual
rainfalls ranging from 100 to 600 mm on catchment: cropped area ratio of 100: 1 to 10000: 1
depending rainfall seasonality (Prinz, 1996). Rockstrom et al., (2007) used the water
management aspect to improve water availability and conservation to classify runoff farming
practices as shown in Table 2.1. This approach also considered evaporation suppressing and
integrated soil, crop and water management practices as components of rainwater harvesting
strategies. A simple approach, adopted for discussion purposes in this study, is the one
proposed by Prinz, (1996) that classify rainwater harvesting technologies according to the
size of the catchments; micro and macro-catchment systems being the main categories (Prinz,
1996). Oweis et al., (1999) modified this approach by including a mini-catchment.
2.2.2.1 Micro-catchment rainwater harvesting technologies
The cultivated area could range from 0.08 m2 to 250 m2 for micro-catchments with
catchment: cropped area ratio ranging from 1:1 to 20: 1. Mini-catchments are adaptable to a
wide range of cultivation methods including row, contour and basin cropping systems such as
those used in tree and agro-forestry plantations. In cultivated fields this class of rainwater
harvesting rely on in-situ soil water management practices to capture and store runoff in the
10
soil profile. Runoff produced between cropped rows is not allowed to develop to ex-field
runoff through the use of contour farming surface structures that maximise infiltration.
Example of free standing in-field runoff harvesting surface structures includes among others:
bunds or terraces, pitting, furrows, ridges and basins of which most are effective when
developed along the contour (Prinz, 1996 and Oweis et al., 1999). Common micro-catchment
systems include the semi circular bunds and the eye brow terraces practiced in the Israel and
Negev desert, and the Meskat and Negarin types which rely on earth contour bands to hold
and distribute runoff along cultivated slopes. Generally regarded as labour intensive
preparations of micro-catchments varies from the use of hand tools to highly specialized
mechanical ploughs such as the 'wavy' dolphin plough used to prepare the vallerani micro
trenches to common basin and ridging implements. Reduction in runoff and soil erosion is
reported to have ranged from 50 % to almost negligible levels (Hensley et al., 2000;
Anschutz et al., 2003; Joseph, 2007). Because of their small catchment size they could be
easily integrated with other cultural practices that reduce evaporation losses similar to those
suggested in Table 2.1. Also the use of no-till in the catchment area to enhance in-field runoff
(Hensley et al., 2000) and ripping or sub-soiling in the cultivated area (Rockstrom et al.,
2007) to encourage deep soil profile storage does improve the effectiveness of micro-
catchment water management systems.
A classical example of an in situ water management strategy is the in-field rainfall harvesting
technique (IRWH) developed by Hensley et al. (2000). The developers preferred to use the
term mini-catchment as equivalent to IRWH even though the catchment: cropped area ratio
falls within the micro-catchments. No-till was used between the 2 m crop rows to encourage
in-field runoff production, while basin tillage along rows retained and facilitated deep
infiltration. The IRWH integrated the basin till with mulch as an evaporation management
strategy according to Rockstrom et al. (2007) classification. To minimise deep drainage
losses the IRWH was developed for the layered and duplex soils predominant in the semi-
arid areas of the Free State Province of South Africa. These soils have a clay rich layer of not
less 40 % either in the B or C horizons. Runoff coefficient from a typical clay soil was found
to be about 0.47 (Botha et al., 2003). A linear function representing the relationship between
runoff efficiency (IR) and rainfall (P) from data for over a 3 year period on clay soils was
expressed (Botha et al., 2003) as:
IR = -0.879 + (0.474) x P (r2 = 0.64) (2.1)
11
Table 2.1 Rainwater harvesting practices and their management options (Rockstrom et al.,
2007).
Rainwater harvesting
Aim strategy Purpose Management option
External water Mitigate dry spells, Surface micro-dams,
Harvesting systems protect springs, subsurface tanks
recharge groundwater, farm ponds,
enable off-season percolation of dams
irrigation, permit and tanks, diversion
multiple uses of water and recharging structures
Conserve rainwater Bunds, ridges,
through runoff to broad-beds and
Increase plant In situ water cropped area or furrows, micro-basins,
water availability harvesting systems, other use runoff strips,
soil and water Maximise rainfall Terracing,
conservation infiltration contour cultivation,
Conservation agriculture
dead furrows
staggered trenches
Evaporation Reduce non-productive Dry planting,
management evaporative mulching,
conservation agriculture,
intereropping, wind breaks
Agro- forestry,
early plant vigour and
vegetative bunds.
Conservation agriculture,
dry planting (early),
Increase plant water Integrated Increase proportion improved crop varieties,
uptake soil, crop and water of water balance optimum crop geometry,
management flowing as productive soil fertility management,
transpiration optimum crop rotation,
intereropping, pest control
and organic matter
management
12
1m 2m
Runoff water accumulates
in basins and percolates
beyond the evaporation zone
Figure 2.1 A diagrammatic representation of the in-field rainwater harvesting technique
(Hen ley et al., 2000)
Where -0.879 could be inferred to depict the combined effect of the soil-water system
interaction on clays attributed to cracking and swelling tendencies, which reduced IR to about
0.47. This value is relatively high compared to the 0.2 to 0.3 recorded from other studies on
micro-catchments with slopes of le than 10 % (Oweis et al., 1999). However higher runoff
coefficients of 0.5 in rocky or stony soils were achievable (Anschutz, et al., 2003). Oweis et
al. (1999) articulated the effect of cropped area (CA) and runoff area (C) , runoff coefficient
(Re) and rainfall (P) on runoff efficiency or depth (IR) to have the function;
IR = Cx P x Re (2.2)
CA
The above analogy depicts the inverse relationship of cropped area on runoff efficiency and
storage. Although increasing the runoff area (C) in drier areas improves accumulative runoff
but issues of prolonged surface ponding along the drainage cropped area could exacerbate
evaporation and risk of overtopping. Inconsistencie on slope along large runoff catchments
also contribute to lower runoff coefficient (Bruggerman and Oweis, 2001, Anschutz et al.,
2003). Similar conclusions were made when 20 m length runoff catchments yielded to erratic
and inconclusive runoff efficiency compared to 2 m length catchments (Hen ley et al., 2000).
It may be tedious to measure runoff coefficients on narrow runoff catchments, but higher
flexibility, effective erosion control and runoff turnover per unit area as well as the benefit of
13
crop canopy cover on the basin area adds to the comparative advantage of micro-catchments
systems (Prinz, 1996; Anschutz et aI., 2003).
Regardless of any C: CA area ratio crop yields per unit hectare in most cases offsets the loss
in productive land as a result of interspacing the cultivated field with catchment runoff area
(Oweis et al., 1999; Hensley et al., 2000: Rockstroom, 2000). For the traditional 2: 1 C:CA
area ratio for the IRWH crop yields and water use efficiency compared to conventional tillage
system was consistently higher than conventional tillage systems (Hensley et aI., 2000; Botha
et al., 2006; Joseph, 2007; Mzezewa et al., 2011; Mzezewa and van Rensburg, 2011).
However, the major concern from all these studies was the negative impact long dry spells of
about 14 to 21 days had on soil water storage even when the basin area is covered with mulch
(Botha, 2006). Figure 2.2 illustrates the severity of water stress on maize plants at tasseling
under IRWH on a 10 hectare field following a more than 3 weeks dry spell. Other researchers
have reported that different row spacing, does influence soil water storage and crop growth
(Sangoi et al., 2001; Eberbatch and Pala, 2005; Acciaresi and Zulunga, 2006; Onyango,
2009). It would be therefore reasonable to determine how different RSW compared with the
traditional 2 RSW with respect to soil water storage and crop yields. In addition, the most
effective way to break yield reducing dry spells is through supplemental irrigation (Zhang et
al., 1998; World Bank, 2005; Rockstrom et al., 2007).
2.2.2.2 Macro-catchment rainwater harvesting technologies
Macro-catchment rainwater harvesting is often simplified as the general collection of runoff
from external catchments for immediate or later crop water use. Runoff from grazing lands
and long hill slopes draining into secondary water ways is harnessed for macro-catchment
rainwater harvesting. Traditionally, runoff from drainage lines is diverted towards and
distributed in pastures or cropped fields on low lying lands using wadi structures, stone or
earth contour bunds. Water harvesting practices found in macro-catchments systems include
among others staggered liman terraces, trapezoidal bunds and semicircular hoops as well as
the stone dams, deep trenches and hillside conduit systems. Catchments of area size ranging
from 1000 m2 to 200 ha with inclination of 5 to 60 % are appropriate for macro-catchments
runoff harvesting. Cultivated areas with C: CA area ratio ranging from 10:1 to 100 could well
be supported by macro-catchment rainwater harvesting designs (Prinz, 1996). Capacity of
this system largely depends on rainfall seasonality which could vary between 100 to 1000
mm (Prinz, 1996).
14
Figure 2.2 A severely water stressed maize crop at tasseling following a dry spell that
persisted more than 3 weeks at Parady's experimental farm under in-field rainwater
harvesting.
Macro-catchments runoff harvesting is usually elevated into flood water harvesting in areas
receiving seasonal rainfalls concentrated into one or two rainstorms. Ephemeral floods from
tream beds and wetlands are harnessed either by damming stream bed to encourage
subsurface flow or diverting surface floods by wadis towards neighbouring cultivated fields
(Prinz, J 996). Size of external catchments ranges from 200 ha to 50 km2 with the runoff to
cropped area ratio varying from 100: 1 to 10000: 1 (Prinz, 1996; Rockstrom, 2000).Spate
irrigation practiced in Eritrea and Ethiopia is a typical example of flood water harvesting
where runoff from the highland slopes is diverted by earth embankments into low lying fields
with levelled basins. Provision of storage structures for seasonal floods provide a much more
controlled water supply to crop plants for greater part of the growing sea on. Earth and
concrete dams are widely used to store floods from external catchments depending on capital
investment and scale of crop production (Prinz. 1996; Oweis et aI., 1999; Rockstrom, 2000).
For example, a much smaller pond with carrying capacity of 180-185 m3 could support
supplemental irrigation on fields ranging from 500-1000 m2, while dams with maximum
storage of 50 million m3 could support supplemental irrigation for an area of about 3000 ha.
Worldwide, the water harvested from mam river streams made available to agriculture
supports about 28 % of the croplands under irrigation (Rockstrom et al., 2007). This
15
contributes about 46 % of the global agricultural economic output (FAO, 2009, 2007). In
South Africa, about 1.5 million hectare is irrigated and the agriculture sector is by far the
biggest consumer of runoff water (Annandale et al., 2011). The agriculture sector used about
7900 million rrr' water compared to the 12900 m3 shared by the rest of the sector in the year
2000. Greater variability on seasonal flows from the major river basins, because of climatic
change and growing pressure for the sector to relinquish more water to others sectors, has
increased the need for supplemental or deficit irrigation and efficient irrigation methods.
2.2.2.3 Supplemental irrigation
Supplemental irrigation is a function of rainwater harvested mainly from macro-catchments
and stored outside the cultivated area to bridge dry spells. It seeks not to alleviate crops from
water stress, but to provide sufficient water for the crop to survive the dry period.
Supplementing seasonal rainfall with about 50 to 200 mm of water by irrigation has shown to
be sufficient to prevent severe yield losses from drier years (Oweis, 1997; World Bank,
2005). The primary goal for supplemental irrigation is to manage a deficit soil water regime
that is sufficient to sustain the crop, but low enough to maximise rainfall capture. This brings
into light the concept of profile available water (PAW) that was further defined in South
Africa to be the difference between the drained upper limit (DUL) and the lower limit (LL).
The DUL is directly correlated to the water content of the root zone where the drainage rate
reaches very low values after saturation of the profile (Ratliff et al., 1983). Bennie et al.
(1991) defined the LL as a function of (i) evaporative demand (ii) crop factors (canopy and
roots) (iii) profile water supply rate (PWSR). Since the PWSR is dependent on soil water
content and evaporative demand, plant water stress index is proportional to the gradient of
PWSR to atmospheric evaporative demand. In this regard the soil profile hydraulic
properties, soil water retention and hydraulic conductivity, become important parameters in
the management of supplemental irrigation at different crop growth stages. Higher crop water
requirements at silking and fruiting stage could trigger severe crop yields losses if dry spells
are not properly supplemented by irrigation (Anrade et al., 1999; Schneekloth and Andeles,
2009).
2.2.2.3.1 Irrigation methods for supplemental irrigation
Advances in technology have resulted to the development of more sophisticated and efficient
irrigation methods, but technological transfer to small scale farmers has been slow.
Therefore, flood systems especially furrow irrigation is still popular among developing
16
communi tie who mainly grow table foods in row cropping ystems (Walker and Skorgeboe,
1987; Crosby et al., 2000). Furrow irrigation efficiencies are estimated to be 25 to 60 %
compared to the sprinkler (60 to 95 %) and drip (80 to 95 %) irrigation system (Waskom,
1994). However, the adoption of blocked ended short furrows with smaller inflow rates has
minimized deep drainage and tail end runoff lo ses (Walker, 2003).
A classical example of a hort furrow irrigation is the micro-flood irrigation (MFI) developed
by Austin (2003). The MFI (Figure 2.3) uses small inflow rates as low as 20 L min-I driven
by gravity from light plastic pipes to irrigate short furrow runs. Distribution uniformity of up
to 95% is competitive to sprinkler or drip irrigation was reported by the developer.
Figure 2.3 Micro-flood irrigation technologies as presented to agriculture experts in
Kouebokkeveld, South Africa (2003).
A nearly uniform opportunity time between the furrow inlet and end was the justification.
Under conventionally long furrow, the same level of performance is attempted through the
use of very large inflow rates controlled by some automated valves (Maskom, 1994, Mailhol
et al., 1999; Walker, 2003). Short furrow irrigation has been successfully integrated with in-
situ rainwater harvesting systems such as those that use basins, furrows and contour ridges.
Figure 2.4 shows the usefulness and compatibility of supplemental furrow irrigation when
integrated with in situ rainwater harve ting techniques.
Based on the evidence picked up in this study the water storage of in situ based rainwater
harvesting strategies including IRWH do not have the capacity to withstand the drying effect
of persistent dry spells. Application of supplemental water during dry spells or when access is
17
very restricted during the critical crop growth tages could stabilise water availability and
crop yields. Since its inception the IRWH has never been integrated with
supplementalirrigation. Increased rainfall variability and severe evaporation occurring in the
mid summer month, November to January (Botha et al., 2003), when crop water requirement
i critical for optimum yields could be uggesting the need for integrating uppiemental
irrigation with IRWH. However, the irrigation method should have a high application
efficiency and compatible with the basin area runoff management structures.
(a) (bl
Figure 2.4 Application of supplemental furrow irrigation along crop row basins (a) and plants
supported by supplemental irrigation among crops damaged by insufficient rainfall (b)
(Adapted from Rocktrom et al., 2007).
2.3 Review on soil physical principles
2.3.1 Background
Efficient use of soil water re ources shall be expected to be improved if the world food
production levels are to be improved or sustained. In the light of increa ed climatic variability
and declining productive potential of cultivated soils, water remains the mo t limiting
resource in crop production among the arid and semi arid zones. Therefore the adoption of
innovative water management practices based on scientific concepts carries the prospects of
enhancing water u e efficiency, especially among the mall scale farmers of developing
nations.
The soil-plant-atmosphere continuum (SPAC) reflect the hydrological processes that
characterize the inflow and outflow of water from the soil, plants and atmosphere
environment. Given that the soil erve as an immediate reservoir for water, its ability to store
18
and release water is of critical importance to crop production. In this regard the conservation
of mass in the SPAC could be defined with respect to the soil water balance (SWB).
Hydrological processes affecting the changes in soil water content (SWC) could also be
mathematically described as
~swc = (P + I) ± R ± DD - (Ev + T) (3.1)
Where L1SWCis the change in soil water content of the soil profile (mm), P is precipitation
or rainfall amount (mm), I being irrigation amount (mm), R is runoff as a loss from the field
(ex-field) (-) or harvested as in-field (+), DD is Deed drainage below the soil profile (-) or
capillary rise from a water table and Ev and T being water evaporated from the soil surface
(mm) and water transpired by crop plants (mm) respectively. Transpiration is regarded as a
productive loss since it relates to crop growth and photosynthesis.
The primary goal for any food production system is to minimize the unproductive soil water
losses from DD, ex-field runoff and Ev while maximizing T for optimum plant growth and
yields. This is a challenge especially in arid and semi-arid areas where Ev exceeds P by flow.
Various soil water management strategies including rainwater harvesting and conservation
practices have been developed to improve water availability and to suppress Ev. However
because of differences in soil physical properties and climatic conditions there is no one
water management practice that is suitable to all soils. To make an informed choice
reasonable knowledge on soil physical concepts affecting water movement within the SPAC
is necessary. Soil physics is defined as that branch of soil science that is concerned with the
application of physical principles to characterize soil properties and to understand the variety
of dynamic processes occurring in soils (Dani and Wraith, 2002).
2.3.2 Soil physical properties
Soil could be defined as an unconsolidated mineral or organic material on the immediate
surface of the earth and serve as a natural medium for plant growth (Hillel, 2004). As a
product of weathered rock material the unconsolidated mineral formation assumes a
particulate, porous and disperse system that readily mix with air, water and organic matter.
An average soil is accepted to have a composition of 45 % mineral matter, 5 % organic
matter, air and water having an equal relative proportion of 25 % of the 50 % pore space
(Hillel, 2004). However, the reality is that soils differ widely from this idealistic form,
especially in arid and semi-arid climates.
19
Mineral matter of varying particle shape and size distribution or texture forms the bulk of the
soil matrix and is a permanent property soils. Soil texture constitutes of the sand, silt and clay
fractions that are generally classified to fall within the respective size range of 0.05 to 2 mm,
0.002 to 0.05 mm and less than 0.002 mm (Hillel, 2004). Particle geometry and texture is
recognized for determining the solid interfacial surface area, pore volume and pore size
distribution that are of significant importance in soil water studies. Through the pore space
water and air is drawn and transmitted (Hillel, 2004).Since all pores are connected with one
another and with the atmosphere, the air phase usually does not affect flow within the soil
matrix. The pore space or porosity is generally classified into micro-, meso- and macro-pores
classes with respective pore diameter ranging from less than 0.001 mm, to 1 mm and greater
than 1 mm (Luxmoore, 1981) Other porosity classes were given by Clothier and White
(1981), Wilson and Luxmoore (1988) and Anderson et al. (1990). Clay has the largest
specific surface area and constitutes primarily of the micro-pores appreciated for their high
ionic adsorption and capiUary activity (Dane and Hopmans, 2002). Soils with high clay
content have been found to have a higher adsorption and capillarity; a physical relationship
that has widely been used to estimate the soil pore size distribution and other soil water
related functions (Bennie, 1991; Simunek et al., 2008; Vereecken et al., 2010; Bouma, 2010).
Soils could assume various textural classes depending on the composition of the different
mineral size particles (Hillel, 2004).
Under in situ conditions the different soil textures in a vertical soil profile could be expressed
in the form horizons or layers parallel to the soil surface. Sequence and composition of the
horizons is mainly a product of pedological evolution. In this regard soil structure, referring
to the way the mineral particles are packed aggregated in the bulk of soil, is of vital
importance in determining the relative proportion of water and air filled pores in the soil
profile. The pore spaces or macro-pores between the soil aggregates or peds are usually
referred to as structural pores (Kutilek, 2004) or drain able pores (Chimungu, 2009). Since
these pores are the first to drain when a]] pores are filled with water they contribute to the
proportion of air fi11ed pores. According to Hillel (2004) soil structure could be of single
grain; referring to loose and unattached particles typical of windblown sands (Hensley et al.,
2007); massive structure referring to heavy or tightly packed structures as in the case of clay
soils; and aggregated referring to structure associated with small clods or peds typical of
loamy soils. Increasing organic matter and earth worm activity is reported to have improved
aggregation and structural pores (VandenBygaart et al., 2000). Loose structure is
20
acknowledged for its large volume of macro-pores that readily draws and release water,
hence assuming a drier soil water regime (Rhoads, and Sukhodolo 1991). However, the
proportion of macro-pores could be reduced through compaction and surface sealing by
tillage or raindrop impact on course and fine grained soils in particular (Moolman, 1981;
Ahuja et aI., 1998). On the other hand heavy soils of massive structure have low volume of
macro-pores and thus resist water movement although it could open cracks when dry that
promote rapid initial water intake or preferential flow (Lin et aI., 1998). In layered soils the
change in structure and texture between horizons could be of smooth or abrupt transition
(Hensley et al., 2007). The layering phenomenon depending on sequence could impair
vertical distribution if the underlying horizon is more restrictive (Hensley et aI., 2000) or
promote unstable flow if the bottom layer is courser than the overlying layer (Jury et aI.,
2002).
2.3.3 Soil hydraulic properties
Soil hydraulic properties could be simplified to mean the characteristics that describe soil
water systems. Characteristics of importance in cultivated soils include soil water content
(SWC), flux rates, the relationship between SWC and matric potential and the hydraulic
conductivity. These properties could be studied using either the single or non-single porosity
models. The single porosity model considers the soil system on at microscopic scale not
affected by the bulk of soil water movement including the activity of macro-pores, root
channels and fractures (Schroder et al., 2008). In contrast is the non single-porosity media
approach that considers the bulk soil water movement as a collective product of textural and
structural porosity along with its spatial variability (Simunek et aI., 2003). It also considers
the representative elementary volume (REV) more meaningful to describe soil physical
properties than a single void flow path.
2.3.3.1 Soil water content
Soil water content in the pore space is highly variable and the most limiting factor in food
production in drier areas. It could be expressed on mass basis as gravimetric water content; in
volume basis as volumetric water content; in depth of equivalent as soil water storage; and in
degree of saturation as the ratio of volume of water to the volume of porosity. Soil
mechanical properties including strength, compatibility and penetrability (Dane and Klute,
1977) as well as bulk density, the ratio of mass of oven dry soil to its bulk volume, and
swelling clays may change as a result of changes in SWC (Dane and Hopmans, 2002). Soil
21
water content is primarily recharged by rainfaJJ and in very limited cases by irrigation. In
drier areas SWC is subject to depletion by either evaporation or transpiration to below the
accessible levels to crop plants referred by Bennie (1991) as the lower limit (LL) of plant
available water (PAW Ld. Following saturation SWC immediately drains away through the
macro-pores, but stabilizes once the drainable pores have approached a near equilibrium state
usually referred as field capacity (HiJJel, 2004) or the drained upper limit (Bennie et al.,
1991). Work by HensJey and Jager (1982) successfuJJy estimated LL from plant parameters
while DUL was estimated from silt plus clay content (Bennie et al., 1988). The difference
between DUL and LL represents the plant available water (PAW) (Bennie, 1995).
Pedological features provide evidence of the soils water regime (Hensley et al., 2007) and the
relationship between these pedological features with DUL are yet to be defined especiaJJy in
layered soils.
2.3.3.1.1 Soil water content measurement
Soil water content measurement has rapidly evolved in recent times from the destructive
techniques to the modern non invasive soil water sensor technologies. The gravimetric SWC
measurement is one of the traditional methods that express gravimetric water content as the
mass of water relative to the mass of oven-dry soil. This method is still the most relied upon
and widely used to calibrate newer techniques. The neutron scattering method (NSM) is
widely applied for field measurement and is non destructive once access tubes are installed. It
relies on a radioactive source of high energy-epithermal neutrons to indirectly measure water
content from the concentration of hydrogen atoms in the soil (Fare and Polyakov, 2006).
Although with effective site calibration the NSM offers accurate results the margin of error
often increases near the soil surface (Leib et al., 2003; Hillel, 2004). Dielectric sensors are
also becoming popular for laboratory and field soil water studies. Common are the time
domain reflectometry (TDR), capacitance (TDR) and the Doppler Flow Meter (DFM)
sensors. Evidence from various studies has shown that all these sensors are automated, easy
to operate and provide accurate measurements when properly calibrated (Nhlabatsi, 2010;
Chimungu, 2009). However, the DFM capacitance sensors occur in various lengths up to 1.8
m making them more appropriate for measuring the soil profile at varying depths. Other
methods include the gamma ray attenuation method that could measure in addition to SWC,
bulk density and X-ray computed tomography, ground penetrating radar, and nuclear
magnetic resonance (Leib et al., 2003).
22
2.3.3.2 Soil water characteristic curve
Given that water in soils loses its pure and free status, the relationship between SWC (8) and
matric potential (h) is a very important soil hydraulic property. This relationship could be
expressed as the soil water characterize curve (SWCC) sometimes refers as the soil water
retention curve (SWRC) (Dane and Hopmans, 2002). The matric suction relates to the joint
effect of capillarity and adsorptive potential affecting water in the pore space that makes
SWCC a physical property that is a function of pore size distribution (Bohne, 2003; Hillel,
2004). This could be explained by the fact that large pores get filled with water at near
saturation and because of their large pore radius there is minimum resistance offered by the
matric suction; hence they readily release water under gravitational influence. However, the
remaining water in the pore space with relatively smaller radius experiences an increase in
matric suction that subsequently reduces the outflow rate from the soil matrix. It is therefore
not surprising that the SWCC is non linear and the most studied physical property (Gardner,
1956; van Genuchten, 1980; Dane and Hopmans, 2002; Haverkamp et al., 2005). In
cultivated soils the matric suction of importance ranges from 0 kPa to near oven dry at -1500
kPa. Most crop plants approach wilting point when soil water content becomes tightly held at
matric suction near -1500 kPa.
2.3.3.2.1 Measurement of the soil water characteristic curve
Although there is not yet a universal theory to describe or predict 8-h relationship from soil
properties, various methods have been developed and a comprehensive review was done by
Dane and Hopmans, (2002). The SWCC could be measured using direct and indirect
methods.
Direct methods: These are among the widely used techniques including the hanging water
column (Haines, 1930), the pressure plate (Richards, 1941) techniques for laboratory
procedures and transducer tensiometers for in situ conditions. The hanging water column
(HWC) also referred as the Haines apparatus. It constitute of initially saturated undisturbed
soil core samples exposed to various lengths of hanging water column capable of inducing
soil water release by pores at matric suctions less than -10kPa. Limitations on the length of
hanging column and possible cavitations (Jury et aI., 1991) have confined the HWC to near
saturation matric suctions. Evidence has shown that structural or drain able pores of diameter
larger than 1 mm (Luxmoore, 1981) drains at matric suction range of less than -30 kPa (Jury,
1991). Nhlabatsi (2010) and Chimungu (2009) recorded drainage curves that approached
23
DUL at matric suctions less than -10 kPa from different clay layered soils for all horizons. In
this regard the use of the HWC to describe the matric suction range within which DUL is
attained would provide a more practical approach as an entry point for small scale farmers in
applying scientific concepts in soil water management practices.
However, to cover a wider matric suction range for the SWCC the pressure plate is
appropriate for a suction range of up to -1500 kPa (Klute, 1986; Jury et al., 1991; Bohne,
2005). In this method volumetric water content is monitored as pressure head is increased in
a step wise manner until the desired matric suction range is realized. An accuracy of this
procedure was observed within a coefficient of variation of 1 to 2% (Reeve and Carter,
1991). However Bittelli and Flury (2009) highlighted some errors associated with the
pressure plate technique including the loss of accuracy at lower matric suction due to the
clogging of pressure plates by soils particles or algae growth. To this effect the use of the dew
point method is recommended for use in collaboration with the pressure plate technique. A
much faster method was the controlled outflow method (Lorentz et al., 1993) but due to its
hysteresis challenges it was replaced by the flow pump method that was first used by
Znidarcic et al. (1991). The SWCC was completed in 3 to 4 days something impossible with
the traditional methods, with reproducible results and least hysteresis. However when
retention curves were compared with the traditional methods a good agreement was observed
(Wildenschild, 1996) suggesting that the investment cost implicated on the new methods
were not justifiable with respect to the accuracy of the SWCc.
Tensiometes equipped with electric transducers is the most appropriate method to determine
the SWCC under in situ conditions. Installed at every horizon it effectively works alongside
with SWC measuring sensors such as the TDR or capacitance probes. Apart of the heavy
tensiometry instrumentation required the challenges that come along with controlling weather
elements and achieving effective soil contact with the porous ceramic cups are limiting the
use of this method (Hillel, 2004).
Indirect methods: Alternative to the direct methods is the pedo-transfer and inverse methods.
The pedo-transfer approach involves estimation of retention characteristics from physical soil
properties such as texture, bulk density, organic matter and clay mineral formation using
regression equations (Rawls et al., 1992). Particle size and pore size distribution (Wosten and
van Genuchten, 1988) have been successfully used to estimate retention characteristics. Apart
24
from the promising results of these methods, strong empiricism and secondary information
required for the soil in question may limit their application.
On the other hand, the inverse numerical method of estimating retention characteristics from
the Richard's flow equation is solved by traditional finite element routine (Kool et al., 1987).
Flow controlled variables from an experiment with known initial estimates is fitted on a flow
equation and numerically solved using a parameterized hydraulic function. Through inversion
the unknown parameters of the retention curve are estimated and optimized using an
objective function. Although this procedure relies on mathematical computations and
parameters assumptions there is evidence that the estimated retention curves are reasonable
(Eching and Hopmans, 1993).
2.3.3.2.2 Mathematical functions of the soil water characteristic curve
To develop the few discrete points measured during the desorption experiment into a
complete SWCC a mathematical expression is usually required for srnoothening and
interpolation (Bohne, 2005). Governed by the Poiseuille's and Darcian flow theory, a number
of these expressions were developed. Earlier models are those of Burdine (1953), Brooks-
Corey (1964) and Mualem (1976). The latter was then transformed to closed form van
Genuchten (1980) that has become the widely used function assuming the following
expression:
(3.2)
Where 8s and 8r are the respective saturated and residual values of the volumetric water
content, 8 (mm mm"), h is the matric suction (mm), while a and n are the respective shape
and pore size distribution parameters, and the condition m= 1-1/n should be satisfied.
Although the van Genuchten (1980) model is popular the goodness of the interpolation is
determined by the ability of the model to fit the measured data. Given that different models
assumes a different air entry point it would be therefore appropriate to compare how best the
available models fit the measured data before selecting a model for any other reason.
25
2.3.3.3 Soil hydraulic conductivity
Nevertheless estimation of either saturated or unsaturated hydraulic conductivity is regarded
as one of the time consuming and yet sensitive exercises that requires both expertise and
specialized equipment. Different approaches using direct and indirect methods including field
and laboratory scale measurements have been developed
2.3.3.3.1 Determining soil hydraulic conductivity
Soil hydraulic conductivity could be viewed according to the Darcian law principle
(Narasimhan, 2004) and refer to as the ratio of flux rate to the hydraulic gradient operating
across the flow domain. While saturated flows are limited to large draining pores, flow under
transient conditions continues following de-saturation and may persist much longer on finely
textured or compacted soils.
2.3.3.3.2 Saturated hydraulic conductivity
Saturated hydraulic conductivity (Ks) reflects the permeability constant of the soil when all
pores are filled with water, a situation that occurs directly after rainfall or irrigation. In the
absence of significant matric suctions at near saturation, Ks is representative of the maximum
flux rate the soil could assume under steady state conditions. Its dependence to pore size
distribution, suggest that each soil shall bear a different Ks value and this could also vary
under different tillage management practices. To this effect in situ methods for determining
Ks provide meaningful estimates. The sprinkler method or infiltrometer is one common
method, but due to kinetic energy of dropslets causing compaction (Hillel, 2004) of the
surface, it is more suitable for rainfall simulation than for Ks determination. The double ring
method (Haise et aI., 1956) was developed for field conditions and is adapted to measure
saturated and unsaturated hydraulic conductivity on both homogenous and layered soils. To
accommodate spatial variability the selected size and number of plots is of importance
(Wildenschild, 1996). Described in details by Haise et al. (1956) the double ring method is an
inverse technique that with the necessary instruments and apparatus the entire set of hydraulic
parameters could be measured simultaneously.
2.3.3.3.3 Unsaturated soils hydraulic conductivity
Hydraulic conductivity (K) under unsaturated conditions loses its constant status it assumed
under saturated flows and become a functional coefficient expressed with respect to either
26
matric suction (h) or volumetric soil water content (8). For either case K estimates are largely
based on laboratory experiments given the difficulty to control field experiments from
environmental elements such as relative humidity. A total of up to 14 different techniques are
available to measure K and were summarized by Benson and Gribb (1997). Common among
this group are the dried monoliths evaporation method (Wind, 1955), Outflow Methods
(Gardner, 1961), Bruce-Klute Method (Klute, 1972), and the Sorptivity Method (Dirksen,
1979). However, difficulties to attain equilibration and need for independent retention data
make these procedures cumbersome. More so, laboratory soil columns fail to represent field
conditions especially for layer soils with abrupt transitions. The instantaneous profile method
(lPM) was traditionally developed for laboratory conditions (S:lawinski et al., 2004) but has
become a reliable in situ procedure especially if the experiment is insulated from weather
elements and water content and tensiometry simultaneous measurements are feasible (Hillel
et al., 1972). Alternative to the lPM is the inverse method that relies on mathematical models
related to those established to characterise the SWCC to predict the K-coefficient. Given that
SWCC is the much easier hydraulic property to measure, the early models used the Burdine
theory (Millington-Quirk, 1961) based on SWCC to predict the K-coefficient. Follow up
work by Brooks and Corey (1964), Mualem, (1976) and van Genuchten (1980) as well as the
Kosugi (1996) contributed remarkably to the predicting of K from SWCC parameters.
Despite their popularity overseas they have not been fully exploited under local soil
conditions especially their ability to predict K under in situ conditions.
2.3.4 General flow equation
The knowledge of SWC, SWCC and K functions needs to be known before any physical
quantification of soil water movement and storage could be determined. Soil as a multi-
porosity system requires hydraulic functions that captures mass and momentum balance in a
dimensionless field for steady and transient state conditions. Based on the conceptual
representation of soil pores as equivalent capillaries, the first mathematical expression of soil
water movement was that of Darcy (1856) which for a three dimensional space (\1) vector
under saturated soils was able to conserved mass by the differential expression:
q = -K\1H (3.3)
Where q is volume flux (cm sol), K is hydraulic conductivity coefficient (cm s -I). \1 is the
three-dimensional space vector that embraces ax, ay and az, \1H is the hydraulic head
27
gradient in three dimensional space. This equation was extended for unsaturated conditions
by Richard (1931) with the provision that K becomes a function of matric suction, K= K (h),
assuming the expression:
q = -K(h)\1H (3.4)
Where \1H is the hydraulic head gradient with reference to matric suction (h) and
gravitational potential (z). The Eq. 3.4 can also express K in terms of volumetric water
content, K(S). Richard (1931) also integrated the dynamic aspect of flow on the Darcian mass
balance equation that resulted in the general flow equation where saturated and unsaturated
conditions could be respectively expressed as:
~~= \1. K\1H (3.5)
:: = \1. (K(h)\1H) (3.6)
The Richard's equations (RE) constitutes two independent variables time (t) and spatial
vector (x,y,z) and two dependent variables; volumetric water content (S) and hydraulic head
(dH), which may include both matrix suction head (h) and gravitational head (z) components
as defined in Eq. (3.4). Various forms of the RE are available and could be solved by
analytical or numerically schemes once their hydraulic constitutive relations and boundary
conditions are defined (Barari et al., 2009). Because of the nonlinearity of the hydraulic
functions used as inputs to the RE various physical models have been developed with
efficient numerical schemes to solve the RE for diverse flow problems involving the
components of SPAC.
One of the attractive physical models is the HYDRUS-ID and HYDRUS-2D computer codes
that is equipped with inverse optimization algorithm that allow inversion of the RE for
various flow boundary conditions such as those involving deep drainage, redistribution and
evaporation (Simunek et al., 2009). Development of optimized physical parameters for some
of the local soils earmarked for rainwater harvesting could be a more convenient approach to
accelerate the up scalling of water harvesting tillage practices. Although HYDRUS codes is
yet to be equipped to simulate surface runoff and to partition total crop water use according
evaporation and transpiration; its application in soil water balance studies including drip and
surface irrigation has been very attractive (Abbasi et al., 2003; Kandelous and Simunek,
2010). However, evidence shows that simple approaches to characterise surface runoff either
28
from rainfall (Morin and Benyamini, 1977) or during surface irrigation (Mateos and
Oyonarte, 2005) and soil water balance including the partitioning of evapo-transpiration into
separate constitutes (Stroosnijder, 1987; Zhang et al., 1998) produce reasonable estimates.
2.4 Gap in knowledge
Evidence from the previous review sections (2.2 to 2.2.2.3.1) points out that rainwater
harvesting water management system in practice do improve rainfall water productivity and
that a number of alternative strategies that are cost effective and easy to manage were
available. In-field rainwater harvesting (IRWH) is one of the attractive strategies that have
been developed to improved water productivity in the dryland crop production areas in the
semi-arid areas of the Free State Province of South Africa. To improve the productive scope
of IRWH in the light of increased concerns on rainfall variability and severe evaporation in
the area a series of studies with IRWH being the focus point have been investigated. Some of
these works are summarized in Table 2.2. The soil water balance components, physical and
pedological features, hydraulic processes and soil physical technologies were illustrated in
almost all these studies the concept of IRWH was well developed and substantiated. Soil
water balance components studied were comprehensively addressed. However the aspect of
redistribution on unsaturated soil profiles was not addressed by all these studies. This
component predominant in semi-arid areas where evaporation is severe and rainfalls are
erratic and too low to wet the deeper profile layers (Hillel, 2004). In addition, although
evaporation has been extensively studied by Hensley et al. (2000); Botha (2006) and
Nhlabatsi (2010), the relationship between the soil hydraulic properties and the progression
of the drying front into deeper layers of the soil profile on different soil types has not been
compared. Evidence shows that soils from semi-arid areas are predominately unsaturated and
that most evaporation occurring under these conditions is controlled by the soil hydraulic
properties (Wilson et al., 1997; Peters and Durner, 2008). Reliance on the arithmetic mean
(Yeh et al., 2005) to determine the soil water balance for the plot is another concern that
needs to be addressed given the proportional differences in the surface area demarcated for
the runoff and basin area. Computer based technologies are fairly used among these studies
including those that predicted runoff and yields such as Putu Run (Walker et al., 2003) and
CYP-SA models. Given the increasing difficulties to measure soil hydraulic properties under
in-situ conditions the introduction of the HYDRUS-ID and -2D codes (Simunek et al., 2009)
to predict these properties from soil physical properties that are easy to measure would be
beneficial to IRWH water management systems.
29
According to the design of IRWH all these studies used the 2 m RSW as a standard
catchment size and questions on how this RSW affected soil water storage with respect to
crop suppression evaporation through their canopies has never been brought into perspective.
The importance of crop canopy cover on suppressing evaporation is well acknowledged in
literature (Grena and Hess, 1994; Eberbatch and Pala, 2005; Magbool et al., 2006).
Irrigation according to the developers of IRWH has been out of question given the
predominance of rain-fed agriculture in the Province. It is not surprising that that none of the
listed studies included irrigation. However, where limited accessible irrigation water is
available it could make a significant different in stabilising yields (World Bank, 2005) in the
face of persistent dry spells common in the area (Bennie and Hensley, 2001). Applying water
where the plant grows and at low cost as possible are the important factors of selecting the
irrigation method. The micro-flood irrigation appears to satisfy these criteria but its
compatibility with IRWH is yet to be tested.
Water use efficiency is one of the criteria for evaluating the productivity of water
management practices. Among the few that applied this criteria only one used the
transpiration based water use efficiency. This was more meaningful than the ET based
efficiency since it provides a measure of the crop performance under a given set of
management practices (Stroosnijder, 1987). The procedure used to partition ET into E and T
used by Hensley et al. (2000) relied on the Tanner and Sinclair (1983) that used a crop
specific transpiration efficiency that was inherently insensitive to the climatic variability and
water stress (Morgan et al., 2003). Therefore, an alternative procedure of practical relevance
needs to be proposed mainly within the framework of integrating micro-flood irrigation with
IRWH.
30
Table 2.2 A summary of reviewed local water management studies related to IRWH showing work done (+) and not done (-) on selected water
balance, texture, soil hydraulic properties (SHP), special tools or models and nature of water use efficiency (WUE).
Soil balance com~onents (mm) Soil texture (%) SHP Special tools/ WUE
Author (s) Cro~ RSW I R DD RED RFF ET T E Clay Silt cla~ SWCC K(9) models ET Tor Ev
Hensley et al. (2000) Maize 2 - + + + + + + + + + CYP-SA + +
Botha, (2006) Maize 2 + + + + + + + + + CYP-SA, + +
PUTU-RUN
Mandiringana Maize 2 +
et al. (2003)
Joseph, (2007) Maize 2 + + + + + + + + + + + Tanner and + +
Sinclair (1983
Fraenkel, (2008) - + + + + + + Pedo-transfer
Bothma, (2009) Maize 2 + + + + + + + + Ridger plough
Chimungu, (2009) + + + + + + RECT
Nhabatsi, ( 20 I0) + + + + + + + + Lysimeters,
Sensors
Mzezewa et al.
(2010) Sunflower 2 + + + + + +
cowpeas
Mzezewa and 2 + + + + + Hofrey rainfall
van Rensbaurg
( 2010) simulator
31
2.4 References
Abbasi, F., Jacques, D. Simunek, J. and van Genuchten, M.Th., 2003. Inverse estimation of
soil hydraulic and solute transport parameters from transient field experiments:
Heterogeneous soil. Am. Soc. Of Agric. Eng., 46: 1097-1111.
Acciaresi, H. A, and Zuluaga, M. S., 2006. Effect of plant row spacing and herbicide use on
weed above ground biomass and corn grain yield. Planta Daninha, 24: 287-293.
Ahuja L. R, Fiedler F, Dunn GH, Benjamin JG., 1998. Changes in soil water retention curves
due to tillage and natural reconsolidation. Soil Science Society of America Journal,
62: 1228-1233.
Ali, A., Oweis T., Rashid, M., EI-Nagger, S., Abdul-Ail, A., 2007. Water harvesting options
in the drylands at spatial scales. Land Use and Water Research, 7: 1-13.
Anderson, S. H., Gantzer, C. J., Brown, J. R., 1990. Soil physical properties after 100 years
of continuous cultivation. J. of Soil Water Conserv., 45: 117-121.
Andrade, E.T., de Correa, P. c., Martins, J. H., Alvarenga, E.M., 1999. Avaliacao de dano
mecánico em sementes de feijao por meio de condutividade eletrica. Revista
Brasileira de Engenharia Agrfcola e Ambiental, 3:54-60.
Anschutz, J., Kome, A., Nederlof, M., de Neef, R. and van de Van, T., 2003. Water
harvesting and Soil water retention. Sec ed. Agromisa Foundation. Wageningen,
Netherlands.
Annandale, J. G., Stirzaker, R. J., Singles, A., van der Laan, M. and Laker, M. C., 2011.
Irrigation scheduling research: South African experiences and future prospects. Water
SA, 37: 751-762.
Austin, c., 2003. Micro flood, a new way of applying waters. http://waterright.com.a. 22/
09/2011, 10.00 a.m (LT).
Awulachew, S. B., Merrey, D. J., Kamara, A. B., Koppen Van, B., de Vries, P. and Boelee,
E., 2005. Experiences and opportunities for promoting small scale/micro irrigation
and rainwater harvesting for food security in Ethiopia. International Water
Management Institute, Working Paper 98, Addis Ababa, Ethiopia.
Barari, A., Janalizadeh, A., Farrokhzad, F. and Ganji, D.D., 2009. Application of homotopy
perturbation method and variational iteration method to anonlinear diffusion equation
with a reaction term. J. Appl. Functional/snal., 4: 300-311.
32
Bennie, A. T. P, Coetzee, M. J. and van Antewerpen, R., 1988. A water balance model for
irrigation based on the water suppy rate from the soil profile and crop water demand.
WRC Report No. 144/1/88. Water Research Commision, Pretoria, South Africa.
Bennie A. T. P., 1990. Sound water management concepts and their applications at farm
level. Department of Soil Science, University of the Free State, Bloemfontein. South
Africa.
Bennie, A. T. P., 1991. Managing the profile available water capacity of soils for irrigation of
annual crops in arid regions. Department of Soil Crop Climate Science, University of
the Free State, Bloemfontein, South Africa.
Bennie A. T. P., 1994. Managing the profile available water capacity for irrigation of annual
crops in arid regions. Department of Soil Crop Climate Science, University of the
Free State. Bloemfontein, South Africa.
Bennie, A.T. P., Hoffman, J. E., Coetzee, M. J. and Vrey, H. S., 1994. Storage and utilization
of rainwater in soils for stabilizing crop production in semi arid areas (Afr.). Report
No. 227/1/94. Water Research Commission, Pretoria. South Africa.
Bennie, A. T. P. and Hensley, M., 2001. Maximising precipitation utilization m dryland
agriculture in South Africa- a review. 1. Hydrol., 241: 124-139.
Benson, C. H. and Gribb, M. M., 1997, Measuring unsaturated hydraulic conductivity in the
laboratory and field, ASCE GSP, 68: 113-168.
Bello, W. B., 2008. The effect of rain-fed and supplemetal irrigation on the yield and yield
components of maize in Mekelle, Ethopia. Ethopian J. of Env.Studies and
management, 2: 1- 7.
Bittelli, M. and Flury, M., 2009. Errors in water retention curves determined with pressure
plates. SSAJ, 73: 1453-1460.
Botha, J. J., 2006. Evaluation of maize and sunflower production in a semi-arid area using In-
field rainwater harvesting. Ph.D. (Agric) Dissertation, University of the Free State,
Bloemfontein, South Africa.
Bohne, K., 2005. An introduction into applied soil hydrology. Lecture notes. Catena Verlag
GMBH, Reiskirchen, Germany.
Bouma, J., 2010. Functional characterization of soil structure field descriptions. 19th World
Congress of Soil Science, Soil Solutions for a Changing World. University of The
Netherlands, Wageningen, Netherlands.
Brooks, R. H., and Corey, A. T., 1964. Hydraulic properties of porous media. Hydrology
paper no.3. Civil Engineering Dep. Colarado State University, Fort Collins, USA.
33
Bruggerman, A. and Oweis T., 2001. Water Resource Conservation and Management for
agricultural production in dry areas. Project 3.1. www.icarda.org. 24/0112012, 11.00
a.m (LT).
Burdine, N. T., 1953. Relative permeability calculation from pore size distribution data.
Trans. Am. Inst. Min. Eng., 198: 71-78.
Chimungu, J. G., 2009. Comparison of field and laboratory measured hydraulic properties of
selected diagnostic soil horizons. M.sc. (Agric) Dissertation, Univer ity of the Free
State Bloemfontein, South Africa
Clothier, B. E., and White, I., 1981.Measurement of sorptivity and soilwater diffusivity in the
field. Soil Sci. Soc. Am. J., 4: 241-245.
Critchley, W. and Siegert, K., 1991. Water harvesting: a manual for the design and
construction of water harvesting scheme for plant production. AGL Mise. Pap. 17.
FAO, Rome.
Crosby, C. T., de Lange, M., Stimie, C. M. and Stoep, I., 2000. A review of planning and
design procedures applicable to small cale farmer irrigation projects. WRC No 578:
2.Darcy, H., 1856. Les Fontaines Publiques de la Ville de Dijon (Fr), Dalmont, Paris.
Dane, J. H., and Klute, A., 1977. Salt effects on the hydraulic properties of a swelling soil.
Soil Sci. Soc. Am. J., 41:1043-1049.
Dane, J. H. and Hopmans, J. W., 2002. Soil water retention and storage - Introduction. In:
Methods of Soil Analy is. Part 4. Phy ical Methods (l.H. Dane and G.c. Topp, Eds.).
Soil Science Society of America Book Series No. 5: 671-674
Darcy, H., 1856. Les Fontaines Publiques dela Ville de Dijon, Dalmont, Paris France (Fr).
Eberbatch, P. and Pala, M., 2005. Crop row pacing and its influence on the partitioning of
evapo-tran piration by winter- grown wheat in Northern Syria. Plant and soil, 268;
195-208.
Eching, S.O. and Hopmans, l.W., 1993. Inverse solution of unsaturated soil hydraulic
functions from transient outflow and oil water pressure data. Land, Air and Water
Resources Paper NO.I00021. University of California, Davis, USA.
Eching, S.O., Hopmans, l.W. and Wendroth, 0., 1994. Un aturated hydraulic conductivity
from transient multistep outflow and soil water pressure data. Soil Sci. Soc. Am. J.
58:687-695.
FAO, 2009. How to feed the world in 2050.www.fao.org., 28/0112012, 10.30 am (LT).
Fischer, G., M. Shah, and van Velthuizen, H., 2002. Climate Change and Agricultural
34
Vulnerability. Special report for the UN World Summit on Sustainable Development,
26 August-4 September, Johannesburg. Laxenburg, Austria
Fares, A. and Polyakov, V.O., 2006. Advances in Crop Water Management Using Capacitive
Water Sensors. In D. Sparks (ed.), Advances in Agronomy.VoI.90, Academic Press.
Gardner, W. R., 1956. Calculation of capillary conductivity from pressure plate outflow data.
Soil Sci. Soc. Amer. Proc., 20: 317-320.
Grena, A. K. and Hess, T. M., 1994. Water balance and water use of pearl millet-cow pea
intercrops in north east Nigeria. Agricultural Water Management, 26: 169-185.
Haise, H. R., Donnan, J.T., Phelan, L.F. and Shockley, D.G., 1956. The use of cylinder
infiltrometers to determined the intake characteristics of irrigated soil. USDA-ARS
and USDA-SCS, ARS 41-7. Gov. Print. Office, Washington, De.
Hanks, R. 1., 1974. Model for predicting plant yield as influenced by water use. Agron. 1.,66:
660-665.
Haise, H. R., Donnan, J.T., Phelan, L.F. and Shockley, D.G., 1956. The use of cylinder
infiltrometers to determined the intake characteristics of irrigated soiI.USDA-ARS
and USDA-SCS, ARS 41-7. Gov. Print. Office, Washington, D.e., USA.
Haines, W. B., 1930. The hysteresis effect in capillary properties and the modes of moisture
distribution associated therewith. J. Agric. Sci., 20:96-105.
Haverkamp, R., Debionne, D., VialIet, P., Angulo-Jararnillo, R. and de Condapa D., 2005.
Soil properties and moisture movement in the unsaturatedzone, in The Handbook of
Groundwater Engineering, CRC Press, Boca Raton, Fla.
Hensley, M. and de Jager, J. M., 1982. The determination of the profile available water
capacities of soils. Department of Soil Science. University of Fort Hare, Alice, South
Africa.
Hensley, M., Botha, J. J., Anderson, J. J., van Staden, P. P. and Du Toit, A., 2000. Optimising
rainfall use efficiency for developing farmers with limited access to irrigation water.
Water Research Commission Report no. 878/1/00, Pretoria, South Africa
Hensley, M., Le Roux, P., Gutter, 1. and Zerizghy, M.G., 2007. A procedure for an improved
soil survey technique for delineating land suitable for rainwater harvesting. WRC
Report No. TT 311/07, Water Research Commission, Pretoria, South Africa.
Hillel, D., Krentos, V. D., and Stylianou, Y., 1972. Procedure and test of an internal drainage
method for measuring soil hydraulic characteristics in situ. Soil Sci., 114:395-400.
Hillel, D. H., 2004. Environmental soil physics. Academic Press, New York, USA.
35
Joseph, F., 2007. Maize response to in-field rainwater harvesting on the Fort Hare/Oakleaf
Ecotope. MSc (Agric) Dissertation, University of the Free State, Bloemfontein, South
Africa.
Joshi, P. K, Jha, A. K, Wani, S. P, Joshi, L. and Shiyani. R. L., 2005. Meta-analysis to
assess impact of watershed program and people's participation. Comprehensive
Assessment of Water Management in Agriculture Research Report 8. !WMl,
Colombo, Sri lanka.
Jury, W., Gardener, W. R., and Gardner, W. H., 1991. Soil Physics (5th Ed.). John Wiley and
Sons, New York, 1991.
Kool, J. B., Parker, J. C. and van Genuchten, M. Th., 1987. Parameter estimation for
unsaturated flow and transport models, a review. J. Hydrol., 91 :255-293.
Klute, A., 1986. Water retention: Laboratory methods. In A. Klute (ed.). Methods of soil
analysis. Part 1, 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI.
Kutilek, M., 2004. Soil water in the system of hydropedology. Nad Patankou 34, 160 00
Prague 6, Czech Republic.
Kandelous, M.M. and Simunek, J., 2010. Numerical simulations of water movement in a
subsurface drip irrigation system under field and laboratory conditions using
HYDRUS 2-D. Agricultural Water Management, 97:1070-1076.
Leib, B.G., Jay, J., Matthews, D. and Gary, R., 2003. Field evaluation and performance
comparison of soil moisture sensors. J. Soil Sci., 168: 396-408.
Lin, H.S., McInnes, K·J., Wilding, L.P., Hallmark, C.T., 1998. Macroporosityand initial
moisture effects on infiltration rates in vertisols and verticintergrades. Soil Science,
163: 2-8.
Lorentz, S. A., DurnFord, D. S. and Corey, A., 1993. Liquid Retention Measurement on
Porous Media Using a Controlled Outflow Cell.Proceedings of American Society of
Agronomy-Crop Science Society of America-Soil Science Society of America 1991
Annual meeting, Denver, Colorado, USA.
Luxmoore, R., 1981. Micro-, Meso- and macroporosity of soil.Soil Sci. Sc. Am. 1., 45: 241-
285.
Mailhol, J. c., Priol, M., Benali, M., 1999. A furrow irrigation model to improve irrigation
practices in the Gharb valley of Morocco. Agricultural Water Management, 42: 65-
80.
36
Maqbool, M. M., Tanveer, A., Ata, Z. and Ahmad, R., 2006. Growth and yield of maize (Zea
Mays L.) as affected by row spacing and weed competition durations. Pak. J. Bot., 38;
1227-1236.
Mateos, L. and Oyonarte, N.A., 2005. A spreadsheet model to evaluate sloping furrow
irrigation accounting for infiltration variability. Agricultural Water Management, 76:
62-75.
Millington, R. J. and Quirk, J. P., 1961. Permeability of porous solids.Trans. Faraday
Society, 57: 1200-1207.
Morgan, c.L.S., Norman, J. M. and Lowery, B., 2003. Estimating plant available water
across a field with an inverse yield model. Division S-6; Soil and Water Management
and Conservation. Soil Sci. Soc. Am. J., 67: 620-629.
Moult, N. G., Lecler N. L. and Smithers 1. C., 2009. A catchment-scale irrigation systems
model for sugarcane; Model application. Water SA, 35: 29-36.
Morin, 1. and Benyarnini, Y., 1977. Rainfall infiltration into bare soils. Water Resour. Res.,
13: 813-817.
Mualem, Y., 1976. A new model for predicting the hydraulic conductivity of unsaturated
porous media. Water Resour. Res., 12: 513-522, 1976.
Mzezewa, J. and van Rensburg, L.D., 2011. Effects of tillage on runoff from a bare clayey
soil on a semi-arid ecotope in the Limpopo Province of South Africa. Water SA., 37;
1-8.
Mzezewa, J., Gwata, E.T. and van Rensburg, L. D., 2011. Yield and seasonal water
productivity of sunflower as affected by tillage and cropping systems under dryland
conditions in the Limpopo Province of South Africa. Agricultural water Management,
98; 1641-1648.
Narasimhan, T. N., 2004. Darcy's law and unsaturated flow. Vadose Zone J., 3: 2059.
Nasr, 1999. Assessing desertification and water harvesting in the Middle East and North
Africa: No.lO, Policy Implications ZET- Discussions.Papers on Development Policy.
Bonn, ZEF Bonn, Zentrum fur Entwicklungsforschung, Center for Development
Research, Univestat Bonn. Balkema. Rotterdam.
Nhlabatsi, N. N., 2010. Soil water evaporation studies on the Glen/Boheim ecotope. Ph D.
Thesis, University of the Free State, Bloemfontein, South Africa.
Onyango, O. c., 2009. Decreased row spacing as an option for decreasing maize (Zea may
L.) yield in Trans Nzoi district, Kenya. J. of Plant Breeding and Crop Science, 1: 281-
283.
37
Oweis, T., Hachum, A. and Kijne. J., 1999. Water harvesting and supplementary irrigation
for improved water use efficiency in dry areas. IWMI Contribution (No. 7), System
Wide Initiative on Water Management (SWIM). Colombo, Sri Lanka.
Pacey, A., and Cullis A., 1986. Rainwater harvesting: the collection of rainfall and runoff in
rural areas. Intermediate Technology Publ, London. England.
Peters, A. and Durner, W., 2008. Simplified Evaporation Method for Determining Soil
Hydraulic Properties. Journal of Hydrology, 356: 147- 162.
Prinz, D., 1996. Water harvesting: Past and Future. In Pereira, L.S., (ed), Sustainability of
Irrigated Agriculture. Proceedings, NATO Advanced Research Workshop. Vimeiro.
Prinz, D., and Malik A. H., 2002. Runoff farming. Institute of Water Resources Management,
Hydraulic and Rural Engineering.Department of Rural Engineering, University of
Karlsruhe. German y.
Ratliff, L. F., Ritchie, 1. T. and Cassel, D. K., 1983. Field-measured limits of soil water
availability as related to laboratory-measured properties. Soil Sci. Soc. Am. 1., 47:770-
775.
Rawls, W. J., Ahuja, L. R. and Brakensiek, D. L., 1992. Estimating soil hydraulic properties
from soils data. In M.Th. Van Genuchten et al. (ed.). Indirect methods for estimating
the hydraulic properties of unsaturated soils. Univ. of California, Riverside, USA.
Reeve, M. J. and Carter A. D., 1991. Water release characteristic. In K.A. Smith and
C.E.Mullins (ed.) Soil analysis: Physical methods. Marcel Dekker, New York. USA.
Rhoads, B. L. and Sukhodolo, A. N., 1991. Field investigation of three-dimensional flow
structure at confluences: 1. Thermal mixing and time-averaged velocities. Water
resources research, 37:2393-2410.
Richards, L. A., 1941. A pressure-membrane extraction apparatus for soil solution. Soil
Science, 51: 377-386.
Richards, L. A., 1931. Capillary conduction of liquids through porous mediums. Physics, 1:
318-333.
Rockstrom, J., 2000. Water resources management in smallholder farms III Eastern and
Southern Africa: An overview. Phys. Chemo Earth, 25: 275- 283.
Rockstrom, 1., Hatibu, N., Oweis, T. Y. and Wani, S., 2007. Managing water in rain-fed
agriculture, Part 4 In: Unlocking the potential of rain-fed agriculture, IWMI,
Colombo, Sri Lanka.
38
Rodriguez, J. A, Dlaz, A., Reyes J. A. and Pujols, R., 2004. Comparison between surge
irrigation and conventional furrow irrigation for covered black tobacco cultivation in
a Ferralsol soil. Span J. Agric. Res., 2: 445-458.
Rosegrant, M., Ximing, C. Cline, S. and Nakagawa, N., 2002. The role of rainfed agriculture
in the future of global food production. EPTD Discussion Paper 90. International
Food Policy Research Institute, Environment and Production Technology Division,
Washington DC,USA.
Sangoi, L., Ender, M., Guidolin, A. F., de Almeida, M. L. and Heberle, P. c. 2001. Influence
of row spacing reduction on maize grain yield in regions with a short summer.
Pesq.agropes bras, 36:861-869.
Schroder, T., Javaux, M., Vanderborgh, J., Korfgen. B., Vereecken, H., 2008. Three
dimensional modelling of soil-plant interactions: consistent coupling of soil and plant
root systems. Vadase Zone Journal, 7: 1089-1098.
Schneekloth, J. and Andales, A., 2009. Irrigation: seasonal water needs and opportunities for
limited irrigation for Colorado crops. Department of Agriculture, Colorado State
University, Colorado, USA.
Simunek, J., Jarvis, N. J., van Genuchten, M.Th. and Cardenas, A., 2003. Review and
comparison of modelsfor describing non-equilibrium and preferential flow and
transport in the vadose zone. J. Hydrol., 272: 14-35.
Simunek, J., van Genuchten, M. Th., and Miroslav, S., 2008. Development andapplications of
the HYDRUS and STANMOD software packages andrelated codes. Soil Sci. Soc. Am.
J., 7: 587-600.
Simunek, J, Sejna, M., and van Genuchten, M.Th., 2009. The HYDRUS-2D software
package for simulating two-dimensional movement of water, heat and multiple
solutes in variably saturated media, Version 2.0. Rep. IGCWMC-TPS-53, p 251, Int.
Ground Water Model.Cent., Colo. Sch. of Mines, Golden, CO.
Slawinski, C., Witkowska-Walczak, B., and Walczak, R.T., 2004. Determination of water
conductivity coefficient of soil porous media. Centre of Excellence for Applied
Physics in Sustainable Agriculture. Lublin, Poland.
Stroosnijder, L., 1987. Soil evaporation: test of a practical approach under semi arid
conditions. Neth. J. of Agric. Sci., 35:417-426.
The World Bank, 2005. Shaping the future of water for agriculture; A source book for
investment in agricultural water management. Rural and Agricultural Development,
Washington DC, USA.
39
Tabuada, M. A., Rego, Z. J. c., Vachuad, G., Pereira, L. S., 1995. Modeling of furrow
irrigation. Advance with two diomsnional infiltration. Agricultural water
management, 28:20 1-221.
van Rensburg, L. D. and Zerizghy, M. G., 2008. The BEWAB+ computer program for
irrigation scheduling. Department of Soil, Crop and Climate Science, University of
the Free State, Bloemfontein, South Africa.
Vereecken, H., Weynants, M., Javaux, M., Pachepsky, Y., Schaap, M. G. and van Genuchten,
M.Th., 2010. Using pedotrasnfer functions to estimate the van Genuchten-Mualern
soil hydraulic properties: A review. Vadase Zone J., 9:795-820.
Walker, W. R. and Skorgeboe, G. V., 1987. Surface irrigation. Theory and Practice. Prentice-
Hall, New Jersey. Engineering, Utah State University, Logan, Utah.
Walker S. and Tsubo, M., 2003. PUTURUN: A simulator of rainfall-runoff yield processes
with in-field water harvesting. WRC Report NO.KVI42/03. Pretoria, South Africa.
Walker W. R., 2003. SIRMOD Ill; Surface Irrigation Simulation, Evaluation and Designs.
Guide and technical document. Department of Biological and irrigation Engineering,
Utah State University, Logan, Utah.
Waskom, R. M., 1994. Best management practices for irrigation and management
Department of Agriculture. Colorado State University. Cooperative Extension Bulletin
# XCM-173. Colorado, USA.
Wildenschild, D., 1996. Characterization of unsaturated hydraulic parameters
forhomogeneous and heterogeneous soils.Department of Hydrodynamics and Water
Resources. Technical University of Denmark, Denmark.
Wilson, G. V., and Luxmoore, R. J., 1988. Infiltration, macroporosity, andmesoporosity
distribution on two forested watersheds. Soil Sci.Soc. Am. J., 52: 329-335.
Wilson, G. V., Periketi, R. K., Fox, G. A., Dabney, S. M., Shields, F.D., Cullum, R.F., 2007.
Soil properties controlling seepage erosion contributions to streambank failure. Earth
Surface Processes and Landforms, 32: 447-459.
Wind, G.P., 1955. Field experiment concerning capillary rise of moisture in heavyclay soil.
Neth. JAgric Sci., 3: 60-69.
Wosten, J. H. M. and van Genuchten, T. H. M., 1988. Using texture and other soil properties
to predict the unsaturated soil hydraulic functions. Division S-6, Soil water
management and conservation. Soil Sci. Soc. Am. J., 52: 1762-1770.
Yeh, M., Khaleel, R. and Yeh, T. C. 1. 2005. Steehastic analysis of moisture plume dynamics
of field injection experiment. Water Resource. Res., 41: W03013.
40
Zhang, H., Oweis, T. Y., Garabet, S. and Pala, M., 1998. Water use efficiency and
transpiration efficiency of wheat under rain-fed conditions and supplemental
irrigation in a Mediterranean-type environment. Plant and Soil, 201: 295-305
Znidarcic, D, I11angasekare, T. and Manna, M., 1991. Laboratory Testing and Parameter
Estimation for Two-Phase Flow Problems. GeotechnicalEngineering Congress 199 l ,
Vol. II, E G. McLean, D. A. CampbeJJ,and D. W. Harris, Eds., ASCE Geotechnical
Special Publication No.27, New York, USA.
41
CHAPTER3
IN SITU EVALUATION OF INTERNAL DRAINAGE IN LAYERED SOILS
(TUKULU, SEPANE AND SWARTLAND)
Abstract
The soil water characteristic curve (SWCC) and permeability properties of layered soils
following deep infiltration depends on the structural and layering composition of the profiles
diagnostic horizons. Three South African soils, the Tukulu and Sepane referred to Cutanic
Luvisols, and Swartland to Cutanic Cambisols in other countries were selected for internal
drainage evaluation. The soil water release curves as a function of suction (h) and unsaturated
hydraulic conductivity (K-coefficient) as a function of soil water content, SWC (8), were
characterised alongside the pedological properties of the profiles. The hanging water column
nd in situ instantaneous profile methods (IPM) were used for this were. Independently, the
saturated hydraulic conductivity (Ks) was measured using double ring infiltrometers. The
three soils had a generic orthic A- horizon but differed remarkable with depth. A clay rich
layer was found in the Tukulu and Sepane at depths of 600 to 850 mm and 300 to 900 mm,
respectively. The Swartland was weakly developed with a saprolite rock at a depth of 400-
700 mm. During the 50 days drainage period, soil water loss amounted to 21 mm, 20 mm and
51 mm from the respective Tukulu, Sepane and Swartland profiles. An abrupt drop in Ks in
conjunction with a steep K-coefficient gradient with depth was supported by the
hydromorphic colours on the clay-rich layers of the Tukulu and Sepane indicating that
drainage was restricted. These characteristics presented the Tukulu and Sepane as suitable for
optimum soil water storage under in-field rainwater harvesting production technique
compared to the Swartland with a much drier soil water regime.
Key words: internal drainage, layered soils, pedological properties, hydraulic conductivity
42
3.1 Introduction
Internal drainage has become a critical factor in soil water conservation especially on layered
soils. Since the large soil pores are the first to drain following deep infiltration, their soil
water release (SWR) and permeability is dependent on the structural and layering
composition of the profile's diagnostic horizons. The SWR and permeability functions could
be described by the SWR curve or soil water characteristic curve (SWCC) and unsaturated
hydraulic conductivity (K-coefficient), respectively. While the former is expressed as a
function of matrix suction (h) and SWC (8) (Dane and Hopmans, 2002) the latter is expressed
as a function of either h or 8 (Eching and Hopmans, 1993).
In practice internal drainage depicts the emptying of water filled pores under the influence of
gravity. In time and space, drainage rate diminishes with SWC and approaches negligible
rates at drainage upper limit (DUL) approximated by Ratliff et al. (1983) to be 0.1 to 0.2 %
per day. The DUL could further be reduced to approach the drainage lower limit (DLL),
equivalent to the permanent wilting point (Hillel, 2004), by evaporation or transpiration
(Bennie et al., 1994). In soils experiencing downward drainage DUL could be achieved
between 2 to 12 days for fine-textured soils and up to 20 days in soils with a restrictive layer
(Ratliff et al., 1983). The difference between DUL and DLL represents the plant available
water (PAW) that is important for estimating crop water requirements (Bennie et al., 1994;
Bittelli and Flury, 2006). Therefore accurate measurement of the soil hydraulic functions that
determines the extent at which soil water drains away from the profile is crucial for effective
soil water management.
In situ internal drainage methods of determining soil hydraulic functions are considered to be
more realistic than laboratory methods. Continuum of soil horizons represents the actual
boundary conditions under which water flows in a composite profile. Internal drainage under
in situ conditions was first studied by Richard et al. (1956). Difficulties in measurement of K-
coefficient under in situ conditions because of variations in air humidity and the need for
simultaneous measurement of many parameters, transient experiments were preferred from
the beginning. Early contributions included work by of Nielsen et al. (1964), Rose et al.
(1965) and Watson (1966) that culminated to the launching of the instantaneous profile
method (!PM) (Hopmans et al. 2002). Inversion of the Richard flow equation provided means
to estimate the K-coefficient (Vachaud and Dane, 2002) indirectly from the ratio of drainage
flux and effective hydraulic gradient. Other in-situ methods are the internal drainage and zero
43
flux plane methods described by Vachaud et al. (1978) and tension disk infiltrometers
(Hopmans et aI., 2002; Simunek et aI., 1999). The IPM was consolidated when the in situ
procedure to calculate the K-coefficient was published by Hillel et al. (1972). Among earlier
studies the unity gradient become popular but was later disputed by Richardts, (1993); Bacchi
and Richardts, (1993). To keep pace with the advances in the laboratory methods such as the
one step (Kool et al., 1985) and multi-step (van Dam et al. 1994; Eching et al., 1994) outflow
experiments, the !PM went through a series of modernization. This included the
computational contributions by Libardi et al. (1980) and Bacchi, (1988), and the
parameterization of the Richard inversion by Richardts et al. (2004). The first to use
parameters based models to inversely estimate soil hydraulic functions from in situ
experimental data was Dane and Hruska (1983), even though lack of uniqueness of their
solution was a concern. Hurtado et al. (2005) used the !PM to estimate hydraulic functions
below the DUL while Neto et al. (2007) introduced computer based software to calculate the
K-coefficient in internal drainage experiments.
Despite the remarkable works done to improve in situ internal drainage methods, laboratory
transient experiments are still by far dominant. Better control of conditions and parameters
influencing soil hydraulic functions as well as access to dynamic instrumentation and
powerful computers has made laboratory experiments attractive. Transferring some of these
technologies to developing farming communities has not been successful in many occasions.
Nevertheless, approaches such as the pedo-transfer (Leij et al., 1996; Acutis and Donatelli,
2003) and hanging water column technique (Vomicil, 1965; Dane and Hopmans, 2002) if
used in collaboration with the !PM could serve as relatively cheap and yet powerful
alternatives.
In the Free State province of South Africa, most of the rural farmers are concentrated on the
low agricultural potential areas predominated by layered soils. This group of soils belong to
the Duplex land type (Soil Classification Working Group, 1991) and occupy more than 10%
of the provincial landscape (Hensley et al., 2007). Erratic soil water regimes common among
this soil has led to the development of infield rainwater harvesting (IRWH) production
technique (Hensley, 2000; FAO, 2009). This technique seeks to turnaround the high surface
runoff and evaporation losses from the 550 mm annual rainfall associated with these soils
into deep infiltration and soil water storage. However, the occurrence of clay rich layer either
in the B or C-horizon among this group of soils suggests that their conservational prospects
for IRWH is dependent on the horizons physical characteristics and layering formation. On
44
this account, the South African Tukulu and Sepane soil types also referred as Cutanic
Luvisols and the Swarland also referred as Cutanic Cambisols of World Reference Base for
Soil Resources (1998) were studied. The following objectives were pursuit: Firstly, to
describe the pedological properties that relate to the presence of layering on the three soil
types. Secondly, to determine the soil water release, unsaturated hydraulic conductivity and
drainage-time functions that characterised the internal drainage outcomes of layered soils.
3.2 Material and methods
3.2.1 Site location and soil classification
The field experiments were carried out at Parady's Experimental Farm (29°13'24.69"S,
26°12'40.93"E, altitude 1422 m) of the University of the Free State. Three sites were selected
with different soil types to include a typical Swartland (Sw), Sepane (Se), and Tukulu (Tu)
soil form, according to the Soil Classification Working Group (1991). Three soil profile pits
were prepared in each site at distance intervals of 50 m. Each profile pit was excavated in a
step-wise downward fashion to allow double ring infiltrometers for the determination of
steady state or saturated hydraulic conductivity for individual horizons. Soil samples from
each horizon from these pits were taken for textural analysis and dry bulk density
determination. Samples were oven dried at 105°C for 24 hours and particle size distribution
was established using the pipette procedures proposed by the Non Affiliated Soil Analysis
Work Committee (1990).
3.2.2 In situ experimental set up and measurements
3.2.2.1 Saturated hydraulic conductivity
Saturated hydraulic conductivity (Ks) for the individual profile layers of the three soil types
was measured using double ring infiltrometers as described by Scotter et al. (1982). On each
excavated soil profile horizon both rings with diameters of 400 and 600 mm were fitted into
the surface to a depth of 20 mm. A floater with 10 mm calibrated depths was inserted in the
central position of the inner ring. The outside ring was first wetted and kept ponded through
the experiment to prevent lateral flows. Then the inner ring was ponded immediately over
some plastic material to prevent erosion of the surface until the depth of water settled to a
calibrated mark on the floater. The clock timer was set and further changes in depths along
the calibrated floater were recorded until a steady time was clocked by the falling water head.
45
The saturated hydraulic conductivity (Ks, mm hour') was obtained by the falling head
method described of Jury et al. (1991);
(3.1)
Where L is the depth of the soil layer in question (mm), bo is the initial depth of total head
above the soil column, bl is the lowest depth the falling is allowed to reach (mm) (mm), tI is
the time taken for bo to fall to bl (in hours). Once the steady state condition was established it
was assumed that the soil matrix suction was negligible and did not affect macro-pore domain
(Zavattaro and Grignani, 2001).
3.2.2.2 Instantaneous soil water measurement
A field representative site from each soil type was selected for the setting up of the in situ
instantaneous profile method (IPM). The IPM provides a more realistic approach in capturing
the flow dynamics in a heterogeneous soil profile subjected to variable hydraulic processes
such as internal drainage and evaporation (Hopmans et al., 2002). In the Sepane and Tukulu a
more central position was selected and monoliths of (4 m x 4 m) x 1 m depth were prepared
with three connected replications. Monoliths were centrally fitted with three neutron access
tubes in a V shape to a depth of 1.1 m. On the Swartland, the monoliths were detached and
setup at 30 m distance apart with dimensions of (1.2 m x 1.2 m) x 0.5 m depth. Shallow
depths on the Swartland allowed the use of DFM probe loggers at depths of 0.6 m. All
monoliths were laterally isolated with plastic sheet to prevent lateral water movement. An
earth bund was made to form a ridge around the monoliths to isolate them from surface
runoff and to support ponding.
Before soil water content (SWC) measurements commenced the monoliths were ponded with
water for three days until it was presumed to have reached saturation. At the end of the 3rd
day the monoliths were covered and sealed at the surface with a polythene plastic upon
depletion of ponded water. Measurements of SWC were taken using a neutron water probe
(NWP) at 200 mm, 500 mm, 850 mm and 1100 mm depth intervals. DFM probe loggers were
set to take readings at every hour during the entire drainage period at depths of 100 mm, 300
mm and 550 mm for the Swartland soil profile. Over the first seven days SWC measurements
were taken in the morning and afternoon and thereafter once a day for a 50 day period. At the
beginning of the field measurements both soil water instruments were calibrated for the
respective soil profiles. From each horizon a soil sample was taken using an auger at a 250
46
mm distance from the NWP and DFM probe. Samples were weighed shortly after packaging
in paper bags. They were then oven dried at 105°C for 24 hours to obtain gravimetric water
content from their in situ bulk densities.
The measured SWC that resulted in a drainage flux of 0.001 mm hour-I was assumed to have
reached the drainage upper limit (DUL). In this study the concept of DUL was adopted and
slightly extended by relating negligible drainage losses to annual average rainfall of semi arid
areas. Given the low and erratic rainfall distribution of semi-arid areas an accumulative deep
drainage losses amounting to 1 % of annual rainfall is generally acceptable. Practically a 1 %
of the 550 mm annual average rainfall received from the central Free State region would be
about 5 mm. This amount could be lost to deep drainage over the seven months (October to
April) rainfall period if drainage rate is 0.001 mm hour". Above this rate deep drainage
would be significant and could result to serious implications on the field water balance.
Interestingly, various field water studies have found that drainage becomes negligible at flux
rates approaching 0.001 mm hour" (Hensley et al., 2000; Chimungu, 2009; Nhlabatsi, 2011).
3.2.3 Laboratory experimental setup and measurements
Undisturbed soil core samples were taken from the centre of each horizon using a core
sampler mounted on a hydraulic jack. Samples were then trimmed, sealed and transported to
the laboratory for analysis of the soil water characteristics curve (SWCC). Sampling cores
had an inner diameter of 103 mm with a height of 77 mm.
Preparation of samples and desorption experiment was carried out in two phases. Core
samples and water used for saturation samples were initially de-aired using a vacuum
chamber pump set at -70 kPa for 48 hours under room temperature. Subsequentl y, de-aired
water was introduced on the second chamber with de-aired core samples allowing the inflow
to wet samples from below. Ponded water outside the wetted sample was cut-off before water
level overtopped the core samples. Core samples were left on the water bath for 24 hours to
allow total saturation. Three replications were used for each horizon. Before commencement
of the desorption experiment saturated samples were weighed and sealed to control
evaporation.
Although the desorption experiment involved three phases from 0-10 kPa, 10-100 kPa and
100-1500 kPa, in this work only the first was illustrated to capture the SWCC during internal
drainage. This procedure involved the hanging soil water column tension cup method
47
described by Dirksen (1999) and Dane and Hopmans (2002). Following saturation the
samples were weighted and water that dripped from the sample at weighing time was
attributed to the porosity of the samples. Placed on ceramic cups prepared for this purpose
hanging 1000 mm from reference point samples, elevation height was gradually reduced at
100 mm intervals for the entire 1000 mm. At every step, interval samples were weighed
before and after samples had equilibrated on the ceramic cups. During this experiment
samples were covered with a foliar sheath to protect them from radiation. Considering that
evaporation was assumed to be negligible under these conditions, change in samples mass
was considered to be a true reflection of the amount of water that drained from the samples.
The resulting soil water content-matric suction (O-h) relationship was then regressed to
estimate the corresponding 8-h relationship to internal drainage of soil profiles horizons from
saturation to drainage upper limit (DUL).
3.2.4. Data analysis
3.2.4.1 Estimation of unsaturated hydraulic conductivity
The unsaturated hydraulic conductivity (K-coefficient) was calculated using the IPM
procedure as described by Slawinski et al. (2004) and Neto et al. (2007). One dimensional
flux was computed using the mathematical expression:
(z t) = fZ ae(z,t) dzq, z=zo at (3.2)
Where q is the flux (mm hour"), 8 is volumetric soil water content (mm mm"), t is the time
in hours and z is the reference depth (mm).
The flux expression was then fitted to the extended Darcian law to express the K-coefficient
as a function of SWC (8);
(3.3)
Where K (8) is the K-coefficient (mm hour"), h is the matrix suction (mm) and t as defined in
equation 3.2. The classical exponential equation proposed by Hillel et al. (1972) was used to
obtain a linear regression function on the semi log scale for the plotted K-8 relationships.
48
3.2.4.2 Statistical analysis
The exponential function describing the K-8 relationships for the three soils were converted
into a semi-log scale after which a linear regression coefficient and intercept for each horizon
was determined. The coefficient of determination (R2) provided the accuracy of the fit on the
resulting semi-log linearized function. The linearized K-8 relationships were then evaluated
for homogeneity using the BartIett' s test or chi-square (l) distribution and pooled regression
coefficient F-test as described by Gomez and Gomez (1984). Also the Student's t-test was
conducted to compare the regression coefficients between horizons within the same soil
profile. Computed values were compared with tabular values, which serve as benchmarks at
95% confidence intervals. If the former was found to be larger than the latter the
homogeneity hypothesis was rejected.
3.3 Results
3.3.1 Pedological properties
Table 3.1 summarizes the physical and chemical properties of the Tukulu, Sepane and
Swartland soils according to the A-, B- and C-horizon layers. The results are also supported
by soil profile photos in Figures 3.1 to 3.3. Further reference is made to the profile
descriptions in Appendices A-C. All three soil types occurred within an area of less than 10
hectare, on a straight mid slope of less than 1%. However, both the Tukulu and Swartland
occupied the upslope position with the former occurring along the smooth drainage lines. The
Sepane soil form occupied most of the down slope areas of the experimental site. Although
all the soil types had a generic orthic A-horizon differentiation in the B- and C-horizons gave
each soil type unique layering properties.
Tukulu: The soil consisted of an orthic A-, neocutanic B- and prismatic C-horizon. The clay
plus silt fraction for the three horizons was 18 %, 31 % and 57 %, respectively with
corresponding bulk density of 1670 kg m', 1597 kg m-3 and 1602 kg rn'. Along with the
increase in clay plus silt content with depth, was the transformation in textural classes from
fine sandy loam in the A-horizon to a fine sandy clay loam in the B-horizon and to clay in the
C-horizon. Corresponding horizons structural representations were massive apedal, weakly
sub angular blocky and moderate to strong prismacutanic structure, respectively. A smooth
transition between the A- and B-horizons confirmed the common attributes of these horizons
including among others colour, texture and youthfulness of structure.
49
Table 3.1 Summary of the physical and chemical characteristics of the three soil types.
Soil Eh~sical EroEerties
Soil forms Tukulu SeEane Swartland
Master Horizons A BI C A BI C A BI C
Coarse sand (%) 5.3 9.2 2.1 5.2 3.5 2.3 4.7 3.2 54.3
Medium sand (%) 9.3 8.8 3.8 10 4.1 2.3 7.6 5.3 4.6
Fine sand (%) 41.2 31 28.3 41.9 41 31 42 37.6 17.2
Very fine sand (%) 25.3 21 8.4 21.5 10.5 18 31.7 26.6 2.5
Coarse silt (%) 2.1 2 3 1 3 1 2 3 3
Fine silt (%) 4.6 2.5 6.5 1 3 1 1 2 3
Clay (%) 11.3 26.4 47.9 19 35 45 11.3 21.9 15
Bulk density (kg m") 1670 1597 1602 1670 1790 1730 1670 1530 1450
Porosity (%) 34.0 33 32.4 34 33.5 33.8 35 39.9 41.6
Ks (mm hour .1) 36.1 40 9.6 35.2 18.1 1.9 23.5 42.8 76.5
Chemical EroEerties
pH (water) 5.8 6.1 6.5 5.9 6.2 6.3 5.5 5.8 6.8
Ca (c rnol.ikg") 35.3 40.8 62.7 30.53 46 50 23 25.78 45
Mg (c mol.ikg") 18.8 20.8 54.7 13.89 21.87 29.76 12.1 13.1 25.4
K (c rnol.ikg") 12.8 6.8 12.9 4.63 8.21 7.05 5.5 7.45 7.7
Na (c mol..kg") 2.4 2.5 3.4 1.96 3.8 3.91 0.7 1.52 2.3
Ksesaturated hydraulic Conductivity
50
(a)
(b)
Yellow, grey and oliv
motties
Figure 3.1 (a) Profile of the Tukulu and (b) motties present in the C-horizon.
(a)
Orthic
A-horizon:
(0-300 mm)
(b)
Figure 3.2 (a) Profile of the Sepane and (b) mottles present in the B-horizon.
In contrast, the clear abrupt tran ition between the B- and C-horizon explained the sudden
increa e in clay plus silt in the C-horizon, which gave the horizon grey colours compared to
the red colours of the upper horizons. The distinct layering on the lower strata also confirmed
51
the sudden drop in Ks from 40 mm hour" in the B-horizon to 9.6 mm hour" in the C-horizon.
Yellow, grey and olive mottled colours in the C-horizon shown in Figure 3.1 (b) could also
be regarded as indicators, not only of the intensity of layering, but also of the wet soil water
regime of this horizon
Sepane: The soil consisted of an orthic A-, pedocutanic B- and prismacutanic C-horizon as
shown in Figure 3.2. The clay plus silt fraction of the three horizons was 21%,41 % and 47%,
respectively. Corresponding bulk densities to the respective horizons were 1670 kg m", 1790
kg m-3 and 1730 kg m'. Accompanying the increase in clay plus silt content with depth, was
the transformation in horizon textural classes from fine sandy loam in the A-horizon to a fine
sandy clay loam in the B-horizon and to fine sandy clay in the C-horizon. Structural
properties unique to these horizon layers were the massive apedal, moderate to course sub
angular blocky, and medium to strong prismatic structure, respectively. An abrupt transition
between the A- and B-horizons and the smooth transition between the B- and C-horizons
explained the concentration of structure between lower horizons of this soil. Changes in red
colours from the surface to dull red and darker colours of the interior also pointed to layering
that occurred at 300 mm depth splitting this soil into a weakly developed surface and well
structured B- and C-horizons. Confirming the contrast in structure between the surface and B-
horizon were the well-developed and distinct peds with clay skins shown in Figure 3.2 (b).
Swartland: The soil consisted of an orthic A-, pedocutanic B- and saprolite C-horizon as
shown in Figure 3.3. The clay plus silt fraction for the three horizons was 14%, 27% and
21%, respectively. Corresponding bulk densities to the respective horizons were 1670 kg m',
1530 kg m-3 and 1450 kg m", respectively. A smooth transition occurred between the surface
A- and B-horizons indicating a gradual build-up in clay content, which resulted from the fine
sandy loam in the A-horizon transformed into a fine sandy clay loam in the B-horizon.
Between lower horizons there was an abrupt transition from a sub-angular blocky structure in
the B-horizon to the saprolite rock of the C-horizon. The presence of dolorite intrusions and
many black sesquioxides concretions shown in Figure 3.3 (b) at depths of 200-400 mm
explained the general course appearance and lack of abrupt transition in this soil. Although
levels of calcium and magnesium showed to increase with depth the predominance of a
weakly weathered saprolite at shallow depths of 400 mm also reflected the deficiencies in
structure and subsequent dry water regime of this soil.
52
(a)
Figure 3.3 (a) Profile of the Swartland and (b) dolorite intrusion present in the B-horizon.
3.3.2 O-hrelationship of soil horizons
Figure 3.4 a, band c shows the resulting 8-h relationships from the hanging column
laboratory desorption of the Tukulu, Sepane and Swartland soil, respectively. Linear
regression functions for each of the three soils horizons with the coefficient of determination
are shown in Table 3.2. Corresponding Student's Hest analysis for differences between the
horizons regressions coefficients are also presented in Table 3.3. Designated to capture the
soil water release behaviour under internal drainage conditions, the 8-h relationship was set
between the suction ranges of 0 to -1000 mm Cl0 -kPa). Spatial variations among the three
soils appeared to increase with suction, an essential indicator for evaluating the permissibility
of water release by the various soil horizons.
Tukulu: The 8-h relationships from the A-, B- and C-horizons appeared to cluster together at
near zero matric suctions but after matric suctions < -200 mm their spatiality increased. The
A-horizon contained 0.33 to 0.277 mm mm-I of water while the B-horizon contained 0.325
and 0.285 mm mm" water at matric suctions 0 and -1000 mm. On the other hand, the C-
horizon contained 0.324 and 0.321 mm mm-I water at matric suctions, resulting in a 8-h curve
with a very small slope (Table 3.2). Effective saturation satisfied porosity only in the C-
horizon while the A- and B-horizons were below with 3 % and 1 %, respectively (Table 3.1).
53
Table 3.2 8-h regression functions of soils horizons for the 0 -1000 mm suction range.
Soil type Horizons Regression R2
Tukulu A e = 0.332 - 0.00005 h 0.98
B e = 0.328 - 0.00004 h 0.96
C e = 0.324 - 0.000003 h 0.93
Sepane A e = 0.327 - 0.00007 h 0.99
B e = 0.333 - 0.00004 h 0.99
C e = 0.338 - 0.00004 h 0.99
Swartland A e = 0.339 - 0.00008 h 0.99
B e = 0.347 - 0.00006 h 0.99
C e = 0.345 - 0.00013 h 0.97
Table 3.3 Student's t-test for differences between horizons 8-h regression coefficients.
Soil type Horizons Pooled variance Tabulated variance Cl: 0.05)
Tukulu AB 3.002 2.145
BC 12.295 2.145
AC 18.688 2.145
Sepane AB 13.868 2.145
BC 2.776 2.145
AC 16.516 2.145
Swartland AB 5.819 2.145
BC 8.489 2.145
AC 5.541 2.145
Corresponding slopes to these 8-h relationships was 0.005%, 0.004% and 0.0003% for the
respective A-, B- and C-horizons with the C-horizon having the slowest water release
function (Figure 3.4). Although the A- and B-horizon lines could be observed to be close to
each other for the given matric suction range, the Student's t-test captured no similarities.
Linear regression functions fitted the measured data well with the R2 of not less than 93%.
Sepane: The 8-h relationships from the A-, B- and C-horizons represented soil columns that
exhibited different soil water retention properties for the matric suction range in question.
The A-horizon contained 0.328 and 0.257 mm mm-I water while the B-horizon contained
0.332 and 0.289 mm mm" water at matric suction 0 and -1000 mm. Similar to the B-horizon
was the C-horizon that contained 0.338 and 0.299 mm mm-I water. Effective porosity was
3.5% and 1% below porosity on the respective A- and B-horizons while in the C-horizon the
effective porosity was fairly satisfied (Table 3.1).
54
(a)
,-.._ 0.36
E
E 0.34
E 0.32
..E__,
.c..... 0.30
2
c 0.28
0
u
..... 0.26
.a...)..
<':l 0.24
~
0.22 -e-- A-horizon --+-- B-horizon _./Jr. C-horizon'0
Cl) 0.20
0 100 200 300 400 500 600 700 800 900 1000 I 100
(b) Suction (-mm)
0.36
,-.._
E 0.34 . '-'~
E ---- :_' - , - , - ,~
E 0.32 --------- :':-'-:~--:':''::_:'~'-:'':-:~ - , - , ~ , - , - bo
..E__, 0.30 ------+------..=- --.----,t&---' - . -6.,....
C
a) 0.28
C
0
u 0.26
.....
.ar..o.
).. 0.24
~ 0.22 -e-- A-horizon --+-- B-horizon - '/Jr' C-horizon
'0
Cl) 0.20
0 100 200 300 400 500 600 700 800 900 1000 I 100
(c) Suction (-mm)
0.36
E
E 0.34 ..__.-.-..........-------------+-
E
E 0.32 ~-~ ------------+------------ ...
'z 0.30 ... -
.--.-.-. .......
c ~', -----_t_
.a....)
-----+
.
c 0.28
0
u
..... 0.26 'A - , - , - , _ , -A
a) -. - °A.
êij 0.24 -. -8-
~ '''''h.
0.22
'0 -e-- A-horizon --+-- B-horizon - 'I!r' C-horizon
Cl) 0.20
0 100 200 300 400 500 600 700 800 900 1000 1100
Suction (-mm)
Figure 3.4 S-h relationships from the (a) Tukulu, (b) Sepane and (c) Swartland soil types.
55
The resulting slopes from these 8-h relationships (Figure 3.4) were 0.007%, 0.004% and
0.004% for the A-, B- and C-horizons, respectively. Interestingly, all three horizons 8-h
functions had R2 not less than 99%. Student's t-test also captured that the regression
coefficients from these horizons were different.
Swartland: Variations in the O-h relationships of all three soil horizons indicated low water
retention for the matric suction in question. For matric suctions 0 and -1000 mm, the A-, B-
and C-horizons contained 0.343 and 0.255 mm mm', 0.348 and 0.287 mm mm", and 0.355
and 0.227 mm mm", respectively. Effective saturation was 2%, 12.8% and 14.7% below
porosity from the respective A-, B- and C-horizons (Table 3.1). The A-horizon soil water
retention curve had a slope of 0.008% while the B-horizon resembled a slope of 0.005%. The
greatest slope was found in the saprolite C-horizon of about 0.012%. When compared,
regression coefficients from individual horizons O-hfunctions were significantly different.
3.3.3 K-9 relationships of soil horizons
Resulting K-8 relationships from the in situ desorption of A-, B- and C-horizons of the
Tukulu, Sepane and Swartland soils under internal conditions are shown in Figure 3.5 a, b
and c, respectively. Corresponding regression exponential K-8 functions are presented in
Table 3.4. Statistical homogeneity analysis is also illustrated in Tables 3.5 and 3.6. The
differences in K-8 relationships reflected the spatial responses of drainage to the
heterogeneity of soil layer permeability.
Tukulu: The resulting K-8 relationships from the A-, B- and C-horizons occupied the SWC
range between 0.297 to 0.328 mm mm" (Figure 3.5a). The K-8 lines of the A- and B-
horizons were close to each other close to the origin while that of the C-horizon was close to
saturation.The A-horizon had K-8 coefficients ranging from 36.1 to 0.0005 mm hour" for a
SWC ranging of 0.328 to 0.297 mm mm-I. The B-horizon K-8 coefficient ranged from 40 to
0.0009 mm hour" with a soil water range of 0.325 to 0.299 mm mm", For the C-horizon the
K-8 coefficient ranged from 9.6 to 0.0005 mm hour" for a drop in soil water content from
0.324 to 0.309 mm mm". The K-8 relationships regression functions had R2 of not less than
0.94 (Table 3.5) with the horizons significantly different for the Bartiett's test. However, the
Student's t-test it was found that K-8 relationships between A- and B-horizons were not
comparable (Table 3.6).
56
(a)
L~. 100
0 DA-horizon + B-horizon t.C-horizon
..c
E 10
E
'-"
.C;;
.;:
u~ 0.1
"0
t:
0
u 0.01
u
:;
('j 0.001
I-.
"»0
::r: 0.0001
0.24 0.25 0.26 0.27 0.28 0.29 0.3 0.31 0.32 0.33 0.34 0.35
(b) Soil water content (mm mrrr!)
,-...
L. 100 ~----------------------------------------------------~~ ,I
0 DA-horizon +B-horizon t.C-horizon
..c +',+E 10 ,
E / .
'-"
C ,I ,-t'.~.
.;>: -it .
u~ 0.1 It.,"0 ~.t:
0
u 0.01 IJ
.::!
~
('j
I-. 0.001 IJ
"»0::r:
0.0001
0.24 0.25 0.26 0.27 0.28 0.29 0.3 0.31 0.32 0.33 0.34 0.35
(c) C' :1 _ ""'+ + f.....,.., -1\
,-... 100 ~-------------------------------------------r~--~--~
L.
o~ DA-horizon +B-horizon t.C-horizon
..c 10
E
E .-.-.-.~ t.'-" 1 o
t::>-. _.-
0.1
0.01
.::!
~ 0.001
~
"0
::»r: 0.0001
0.24 0.25 0.26 0.27 0.28 0.29 0.30 0.31 0.32 0.33 0.34 0.35
Soil water content (mm mrrr')
Figure 3.5 K-8-relationship from the (a) Tukulu, (b) Sepane and (c) Swartland soil types.
57
Sepane: The resulting K-8 relationships from the A-, B- and C-horizons occupied the SWC
range between 0.292 to 0.338 mm mm-I (Figure 3.5b) The K-8 relationship of the A-horizon
inclined towards the interior while the B- and C-horizons were closer to each other and to the
exterior. The A-horizon had the K-8 coefficient ranging from 35.2 to 0.0007 mm hour -I for a
soil water content ranging of 0.326 to 0.292 mm mm-I. The B- and C-horizon K-8 coefficient
dropped respectively from 18.1 to 0.0015 mm hOU{1and 1.9 mm hOU{1to 0.0005 mm hOU(1
with a corresponding soil water content ranging from 0.332 to 0.316 mm mm-I and 0.338 to
0.321 mm mm-I. Horizons could were significantly different for the Barttle's test and could
not be described by one regression coefficient (Table 3.5). However, with the Student's test it
was found that K-8 relationships between B- and C- horizons were not compatible (Table
3.6).
Swartland: The resulting K-8 relationships from the A-, B- and C-horizons occupied the
SWC range between 0.222 to 0.344 mm mm-I (Figure 3.5c). The K-8 lines of the A- and B-
horizons were close to each other and found close to saturation while that of the C-horizon
was close to the origin. The A-horizon showed a K-8 coefficient ranging from 23.5 to 0.0002
mm hOU{1for a soil water content of 0.332 to 0.292 mm mm". The K-8 coefficient for the B-
horizon ranged from 42.8 to 0.0007 mm hOU{1with a SWC ranging from 0.344 to 0.299 mm
mm-I. For the C-horizon the K-8 coefficient ranged from 76.5 to 0.001 mm hour -I with a soil
water content ranging from 0.330 to 0.22 mm mm-I. The K-8 relationships from the three
horizons attained a coefficient of determination not less than 0.89 (Table 3.4) with horizons
significantly different for the Bartletts's test (Table 3.5). However with the Students t-tests it
was found that K-8 relationship between A- and B-horizons were not comparable (Table 3.6).
Table 3.4 K-8 relationships linear regression functions for the three soil types.
Soil type Horizon Regression R2
Tukulu A Log K = 186.1(8) - 58.42 0.94
B Log K = 167.4(8) - 52.24 0.99
C Log K = 274.1(8) - 87.82 0.97
Sepane A Log K = 143.9(8) - 45.21 0.99
B Log K = 281.3(8) - 91.58 0.98
C Log K = 260.4(8) - 86.51 0.87
Swartland A Log K = 124.3(8) -39.64 0.89
B LogK= 120.6(8) 39.15 0.97
C Log K = 43.29(8) - 12.62 0.99
58
Table 3.5 Homogeneity test for the three soils horizons.
BartIett' s test Regression coefficient
Computed Tabular X2 Computed F Tabular F
Soil types X2 (a; 0,05.2) (a; 0 05,2) (a; 005.2)
Tukulu 19.36 5.99 27.71 3.16
Sepane 26.75 5.99 51.99 3.16
Swartland 32.46 5.99 119.29 3.16
Table 3.6 Student's t-test for differences between horizons K-8 regression coefficients.
Soil type Horizons Pooled variance Tabulated variance ((1: 005)
Tukulu AB 1.71 ns 2.03
BC 9.91 2.03
AC 5.09 2.03
Sepane AB 14.56 2.03
BC 0.84 ns 2.03
AC 6.12 2.03
Swartland AB 0.34 ns 2.03
BC 18.58 2.03
AC 10.73 2.03
ns= not significantly different at 95 % confidence interval
3.3.4 0-T relationship of soil horizons
The 8-T relationship for the Tukulu, Sepane and Swartland soil types for the first 210 hours
of internal drainage are presented in Figure 3.6. The power regression functions for the A-, B-
and C-horizons for these soils are presented in Table 3.7 and the Student's t-test for the
comparison of regression coefficients between horizons 8-T functions presented in Table 3.8.
The presence of restrictive horizon layers influenced the manner in which these soils drained.
Tukulu: The 8-T relationship for this soil indicated an increase in SWC with depth although
each layer continued to drain instantaneously. The amount of water that drained away within
the first 3 hours was 5.1 mm from the orthic A- and 2.4 mm from the neocutanic B-horizon
from an initial field saturation of 0.328 and 0.325 mm mm' I, respectively. At field saturation
of 0.324 mm mm'l the C-horizon lost 1.5 mm for the same period. At the end of the 200 hour
drainage period the A-, B- and C-horizons drained away water amounting to 8.1 mm, 6.2 mm
and 3.1 mm, respectively. The drainage flux of 0.001 mm hOUr'I equivalent to DUL was
realised after 660 hours (28 days) after drainage for the A- and B-horizons while from the C-
horizon DUL was realised after 390 hours (16 days).
59
(a)
0.34
--e- A-horizon --+-- B-horizon -'I!r' (-horizon
E
E 0.33
E
E
'--" 0.32
.......
!:: ..:. .b f::r. . - . - 6,- . _ . _
(l.) .....~..-------+------- l!.- . - . - . - . - . - l!.- . - . - . - . - . - . - . - . - . - . - .t:r .C 0.310
u,_ ~----e-----~~&---_--_-_--_--~-=--=--~+~-:--=--:-:--=--:-=--=--=--=------------+---
..(.lr.
..).. 0.30 -=
;so:
'0 0.29
o:
0.28
0 20 40 60 80 100 120 140 160 180 200
(b) Time after saturation (hours)
0.34
,.-_ --e- A-horizon --+-- B-horizon -'I!r' (-horizon
E 0.33 \
E --.._.. l!.-l!r - . _ . -E -
A= - . - . -l!r - . - . - . - . - . -l!r - . - . - . - . - • - . - . - . _ . _ . _/!;._ .
E 0.32 ~-+-------+-------+----------------+-----------------
'--" ------------+--.......
!::
..(.l...).. 0.31
!::
0
U,_ 0.30
..(r.
l..)
;so
...
: 0.29
'0
r:/)
0.28
0 20 40 60 80 100 120 140 160 180 200
Time after saturation (hours)
(c)
0.34 @-------------------------------,
--e- A- horizon --+-- B-horizon -'I!r' (-horizon
0.32 -.."-++------+ -------t---------------,_--------------------- -t- _
0.30
\
\
0.28
0.26 ....b- .... ·"'·b· _. _.
'0 -'- '-'h,._._._._._._
o: O.24 .:j-,r-T"'''T''''''Ir-T'''''T''''''I-.-''T''''''I-.--r-I---r"-r-I-,-..,.......,.-,-..,.......,.-r--r-T-r--r-T-r--r-T-r--r-T. - • - • - •h. . - . -
o 20 40 60 80 100 120 140 160 180 200
Time after saturation (hours)
Figure 3.6 S-T relationships from the (a) Tukulu, (b) Sepane and (c) Swartland soil types.
60
Corresponding SWC was 0.30 mm mmo) for the A- and B-horizons and 0.31 mm mmo) for
the C-horizon. The Student's t-test also confirmed the similarities observed between the
upper horizons.
Sepane: From the 8-T relationship of the three horizon layers soil water content increased
with depth while drainage rates decreased. The amount of water that drained away from field
saturation of 0.326 mm mm", 0.332 mm mmo) and 0.338 mm mmo) within the first 3 hours
was 4.7 mm, 2.8 mm and 2 mm for the respective A-, B- and C-horizons. At the end of 200
hours the layers drained a respective water content amounting to 8.5 mm, 5.4 mm, and 3 mm.
The drainage flux of 0.001 mm hour" was attained after 660 (27 days), 480 (20 days) and
300 (13 days) hours from saturated SWC of 29.4 mm mm", 31.7 mm mmo) and 32.3 mm
mm", for the A-, B- and C-horizons, respectively. The Student's t-test also confirmed the
comparable differences found between these soil horizons.
Table 3.7 8-T relationships linear regression functions for the three soil types.
Soil form Horizons Regression R2
Tukulu A Y= 0.317x·0.009 0.96
B Y= 0.3l9x·0.oo9 0.99
C y = 0.320x·0005 0.99
Sepane A y = 0.3l6x·00" 0.99
B Y= 0.328x·0005 0.99
C y = 0.332x·0005 0.97
Swartland A y = 0.318x·00':l 0.92
B y = 0.328x·0013 0.97
C Y= 0.308x·0045 0.99
Table 3.8 Student's t-test for the differences between horizons 8-T regression coefficients.
Soil type Horizons Pooled variance Tabulated variance (u; 0.05)
Tukulu AB 1.153 ns 2.032
BC 12.977 2.032
AC 4.145 2.032
Sepane AB 5.178 2.032
BC 6.977 2.032
AC 7.854 2.032
Swartland AB 3.247 2.032
BC 44.447 2.032
AC 13.161 2.032
61
Swartland: Soil water content was concentrated on the upper soil horizons with the B-horizon
assuming a higher water content regime throughout the drainage period. The amount of water
that drained away within 3 hours was 4.6 mm from the A-horizon, 3.76 mm from the B-
horizon and 6 mm from the C-horizon. After 200 hours of drainage 8.6 mm, 7.5 mm and 26
mm drained from an initial field saturation of 0.332 mm mm", 0.344 mm mm" and 0.33 mm
rnm' for the A-, B- and C-horizons, respectively. A drainage flux of 0.001 mm hour" was
achieved after 672 (28 days) hours from the A- and B-horizons with a respective SWC of
0.29 mm mm' and 0.30 mm mm-I. For the C-horizon flux rates of 0.006 mm hOU(1 were
observed after 672 (28 days) hours of drainage and continued to decline to a rate of 0.003 mm
hOU(1by the end of the drainage period. Although the Student's t-test found all the horizons
to be comparable the pooled variation was smallest between A and B-horizons.
3.3.5 Total drainage and flux rates
Total drainage and flux rates from the internal drainage trial were summarized in Table 3.9
and Figure 3.7, respectively. The respective drainage amounts from the three soil types
showed a decrease with depth with the exception in the Swartland. In all the soils the amount
that drained away approached negligible levels (DUL) at drainage flux rate of 0.001 mm
hour". At this flux rate the total water loss was equivalent to 0.09% of the 550 mm annual
average rainfall of the area.
Table 3.9 Amount of soil water that drained away from horizons.
Accumulative water de12th(mm) at different drainage 12eriods(hrs)
Soil t~Qe Horizons 18 24 72 210 300 390 660 930 1200
Tukulu A 6.44 6.63 7.38 8.09 8.33 8.51 8.86 9.08 9.25
B 3.94 4.21 5.23 6.21 6.54 6.78 7.25 7.56 7.79
C 2.15 2.26 2.69 3.11 3.25 3.35 3.56 3.69 3.79
Sepane A 6.38 6.64 7.62 8.56 8.87 9.10 9.55 9.84 10.06
B 3.88 4.06 4.77 5.45 5.68 5.84 6.18 6.39 6.56
C 2.36 2.43 2.72 3.00 3.09 3.16 3.29 3.38 3.45
Swartland A 7.40 7.55 8.15 8.70 8.89 9.04 9.34 9.51 9.65
B 5.42 5.67 6.64 7.52 7.83 8.08 8.55 8.83 9.04
C 16.70 17.87 22.21 26.03 27.35 28.37 30.32 31.45 32.29
62
(a)
,-.. 100
l- -a- A-horizon ...+...B-horizon -. A· - C-horizon
::l
0 10
...c
E
E 1
x
::l
r:;:: 0.1
Cl)
bJ)
ê 0.01
'0...;; ......................
00.001 ,j-------=-_.:....:'8--·---.-_ - -._._._ . ...•...•..'.........•...........
0.0001
0 200 400 600 800 1000 1200
(b) Time (hours)
100
,-.. -a- A-horizon ........ B-horizon -'l!r' C-horizon
l- 10
::l
0
...c
E 1
E
'-"
x 0.1
::l
r:;::
Cl)
bJ) 0.01
~
gc: 0.001
0.0001
0 200 400 600 800 1000 1200
(C) Time (hours)
100
~ A-horizon --+-- B-horizon - 'l!r' C-horizon
,-.. 10
l-
::l
0
...c 1
E
E
'-" 0.1
x
::l .. Dr-'-
r:;:: 0.01 ............... 6 -.~_._.-.-.- ..
Celn) -------------- A _.-._._.~._._._._._~'0c:...;;0.001 -------~.-,_._.- _-------------
0
0.0001
0 200 400 600 800 1000 1200
Time (hours)
Figure 3.7 Drainage flux from the (a) Tukulu, (b) Sepane and (c) Swartland soil types.
63
3.4 Discussion
The internal drainage property was consistent with the presence and intensity of layering
from the Tukulu, Sepane and Swartland soils. Evidence of slight wind erosion in the absence
of flooding or water erosion could suggest that eutrophic conditions of the semi arid
environment played a critical role in the development of the layering pattern of these soils.
Moderate physical and chemical weathering of underlying material of solid rock of binary
origin affirm the in situ evolution of the three soil types. Despite the similarities in parent
rock material and orthic A-horizons, the differences in subsurface layers could be explained
by the geological position each soil occupied in the landscape. The Tukulu and Swartland had
a weakly developed A- and B-horizons with a smooth transition at the interface denoting a
weak layering between these horizons. Chimungu (2009) treated the Tukulu upper horizons
as homogenous in order to describe the internal drainage by the unity gradient approach
recommended by Yeh et al. (1985). The clay rich underlying horizons found in the Tukulu
and Sepane was indicative of the common drainage line shared by these soils. These clayey
horizons are of alluvial origin (Hensley et al., 2007).
Given that internal drainage is accountable to gravitational flow boundary conditions, the
hydraulic characteristics of the underlying soil profile layers are key to this process (Blume et
al., 2009; Chimungu, 2009; Nhlabatsi, 2010). Consistent with the restrictive clay layer of the
prismacutic C-horizons was 21 mm and 20 mm of water that drained away from the
respective Tukulu and Sepane soil forms. Corresponding to the weakly developed Swartland
was 51 mm drained away from which 32 mm was contributions by the saprolite C- horizon.
This was expected given the coarseness of the saprolite rock with Ks values of about 77 mm
hour". Hensley et al. (2000) and Botha et al. (2003) concluded that the saprolite layer was ill-
posed for deep soil water storage given its high water release characteristics. This was
consistent with the illustrations given by Hillel (2004) about soils with an underlying course
texture horizon having a drier soil water regime. However, the pedocutanic B-horizon when
fully developed could have clay content exceeding 30 % that was acceptable for soil water
conservation under IRWH technique (Hensley et al., 2000; Botha et aI., 2007).
Evidence of the slow drainage properties of the clay rich layer(s) in the Tukulu and Sepane
was the presence of hydromorphic mottles confirming that soil water was subjected to
restricted flow above these layer(s). In the Sepane hydromorphic mottles were also found at
depths of 300 to 700 mm reflecting the effect of the pedocunatic clay rich B-horizons on
64
internal drainage. Nevertheless, the prismacutanic layers were the most restrictive with 9.6
mm hour' and 1.9 mm hour-I from the Tukulu and Sepane, respectively. The latter realised a
flux rate of 0.001 mm hour" after 13 days of drainage while the former after 16 days. The
same flux rate from the Tukulu was recorded after 26 days of drainage by Chimungu (2009).
However, the more restriction on internal drainage of the Sepane could be explained by the
duality of heavy structured layers in this profile (Fraenkel, 2008). Prospects of having a
positive pressure head building up from the restrictive layers (Tartakovsky et aI., 2003) does
support the high drainage rate observed at the early hours of drainage from the Tukulu and
Sepane. Inconsistencies in drainage amounts within each soil profile could reflect poor
sealing of the monoliths side walls against lateral leakage, a scenario that was also a concern
to Hopmans et al. (2002) and experienced by Chimungu (2009) on the same monoliths. Over
and above, the C-horizons from the Tukulu and Swartland controlled the permissible
drainage from the upper horizons, a scenario that was confirmed by the fusion of the flux
curves from these two soils. Greater spatiality from the Swartland flux curves suggested that
the saprolite layer exhibited less control on how the upper horizons drained.
Considering the 8-h and K-8 relationships provided further evaluation of the presence of soil
profile layers on the drainage process. Driven by gravitational potential drainage the process
involves suction and K-coefficients of permeability near saturation (Schaap and van
Genuchten, 2006; Scheffler, 2007). Corresponding drainage pore space are those of diameters
ranging from 1000 urn to several mm (Luxmoore, 1981). The variation in these classes of
pores was noticeable within the suction range of 300 to 1000 mm from the soil layers of the
three soil profiles. However, the small volume of drainable pores (Dane and Hopmans, 2002;
Schaap and van Genuchten, 2005) found on clay rich horizons was confirmed by the almost
level or gentle slope of the 8-h relationship depicting a slow release. Deficiencies in clay and
structure from the Swartland profile layers, especially the saprolite C-horizon resulted to the
steep 8-h relationship that was consistent with the high water release during the internal
drainage. Intensity of layering among the three soils was also captured by the abrupt changes
in Ks downwards the profile within suction ranging from -10 to -100 mm (Zavattaro and
Grignani, 2001). Notable was the drop from 40 to 9.6 mm hour-I between the neocutanic B-
and prismatic C-horizons for the Tukulu. These discontinuities were consistent with the
behaviour of flow and unpredictable nature of hydraulic properties in layered soils (Mathews
et aI., 2004). Another drop of 18.1 to 2 mm hour -I between the Sepane pedocutanic B- and
prismatic C-horizons was recorded alongside the observation that increased clay content with
65
depth restricted internal drainage (Romano et aI., 1996; Bohne, 2005; Gopalakrishnan and
Manik, 2007; Jones et al., 2009).
Variation in K-8 relationship was also able to show the decline in available water for drainage
irrespective of SWC and depth. Distinguishable were the steep gradients of the K-8
relationships from the pedocutanic and prismatic horizons with corresponding hydraulic
gradient fairly above unity. From the orthic and neocutanic horizons this gradient was almost
equal to 1 suggesting that the adoption of the unity gradient by Chimungu (2009) in
approximating the K-8 relationship of these horizons was reasonable. However, application
of the unity gradient is questioned by most researchers especially on clay soils where the K-
coefficient could drop by several orders of magnitude within a narrow range of SWC
(Reichardt, 1993; Bacchi and Reichardt, 1993). Supporting this sentiment was the
observation that the K-coefficient from the Tukulu and Sepane was less than drainage flux at
DUL by order of one magnitude. The rapid fall of K-coefficient by several orders of
magnitudes while SWC remained within the near saturation range was also acknowledged by
Hensley et al. (2000) who related this function to the requirements of IRWH for deep soil
water storage. However, the affinity to water shown by soils predominated by micro-pores of
diameter less than 30 urn (Luxmoore, 1981), has raised questions on the availability of water
tightly held against gravity, to roots of crop plants (Whitmore and Whalley, 2009).
Nevertheless, it can be deduced from this work that the presence of soil layers and their
respective permeability from the three soils affected drainage differently, and hence soil
water storage.
3.5 Conclusions
The evaluation of soil layers on internal drainage from the Tukulu, Sepane and Swartland soil
types earmarked for IRWH production technique was the main objective of the study.
Integrating in situ and laboratory desorption procedures provided the opportunity to reflect on
the pedological presence of soil layers and also to relate their morphological features to the
hydraulic characteristics that governed the drainage process.
Critical conclusions were drawn from the total and rate of drainage expressed by the three
soil profiles. By virtue of their fine size (less than 0.002 mm) the clay fraction determined the
extent of layering, both in texture and structure. Consequently, the clay content influenced the
soil water release and permeability during drainage. Soil horizons with clay content ranging
from 26 to 48 % contributed less than 5 mm to total drainage irrespective of textural and
66
structural class. This suggests that from an annual rainfall of 550 mm drainage losses from
each profile layer with clay content falling in this category should be around 1 % of the total
rainfall. Considering drainage flux of 0.001 mm hour-I to correspond to DUL produced
realistic estimates that were consistent with the slope of 8-T relationships and could be used
to estimate DUL from other layered soils especially those earmarked for IRWH.
Proven by this work was the variation of drainage with the degree of soil layering. The
coefficient of both saturation and de-saturation allowed reflection on the extent at which
drainage was influenced by the presence of an abrupt or smooth transitional change in clay
content and structure. Increased matric suction against gravitational influence on clay rich
profile layers had a significant influence in lowering the K-8 coefficient to negligible levels at
relatively high soil water contents. This function was transferred to overlying layers
irrespective of their textural and structural orientation suggesting the importance of clay soils
on the overall permeability of the profile when occurring on the lower horizons in particular.
About 1 % of the total annual rainfall was lost from the soil profile layers of the Tukulu and
Sepane suggesting their greater potential for soil water conservation, especially for IRWH.
Aspects of deep infiltration and runoff could become management issues when clay rich soil
layers occur at shallow depths of 300 mm. On the contrary, the deficiency in layers rich in
clay and structure from the Swartland proven by the deep drainage and drier soil water
regime presented this soil as a lower potential for IRWH development.
Acknowledgements
We acknowledge the technical support rendered by the management and staff of Paradys
experiment station. Gratitude's to Prof. Malcom Hensley for his contributions in classifying
opened soil profiles and assistance in setting up the hanging soil column method. Special
thanks also to Ms. Liesl van der Westhuizen for her constructive suggestions on improving
the writing of this manuscript.
67
3.6 References
Acutis, M. and Donatelli, M., 2003. SOILPAR 2.00; Soft ware to estimate soil hydrological
parameters and functions. Eur. 1. Agron., 18: 373-377.
Bacchi, O.O.S., 1988. Comparative analysis of soil hydraulic conductivity methods for
unsaturated soils. Piracicaba, Escola Superior de Agricultua Luiz de Queiroz, Tese de
Doutorado.
Bacchi, O. O. S. and Reichrdt, K., 1993.0n simple methods for unsaturated soil hydraulic
conductivity determination. Sci. Agric. Piracicaba, 50:326-328.
Bennie, A.T. P., 1994. Managing the profile available water capacity for irrigation of annual
crops in arid regions. University of the Free State, Bloemfontein, South Africa.
Bittelli, M. and Flury, M., 2009. Errors in water retention curves determined with pressure
plates. SSAJ, 73: 1453-1460.
Bohne, K., 2005. An introduction into applied soil hydrology. Catena Verlag GMBH,
Reiskrchen, Germany.
Botha, J. J., van Rensburg, L. D., Anderson, J. J., Hensely, M., Macheli, M. S., van Staden, P.
P., Kundhlande, G., Groenewald, D. G. and Baiphethi, M. N., 2003.Water efficiency,
food security, and sustainable crop production. Water Research Commission, Report
No.,1176/l/03, Pretoria, South Africa.
Botha, J. J., Anderson, J. J., Groenewald, D. C., Mdibe, N., Baiphethi, M. N., Nhlabatsi, N.
N. and Zere, T. B., 2007. On-farm application of In-field rainwater harvesting
techniques on small plots in the central region of South Africa. Water Research
Commission, Report No., TT 313/07, Pretoria, South Africa.
Blume, T., Zehe, E. and Bronstert, A., 2009. Use of soil moisture dynamics and patterns at
different spatio-temporal scales for the investigation of subsurface flow processes.
Hydrol. Earth Syst. Sci., 13:1215-1234.
Chimungu, J. G., 2009.Comparison of field and laboratory measured hydraulic properties of
selected diagnostic soil horizons. M.Sc. (Agric) Dissertation, University of the Free
State, Bloemfontein, South Africa.
Dane, J. H. and Hopmans, J. W., 2002. Soil water retention and storage - Introduction. IN:
Methods of Soil Analysis. Part 4. Physical Methods. (J.H. Dane and G.c. Topp, Eds.).
Soil Science Society of America Book Series No. 5: 671-674.
Dane, J. H. and Hruska, S., 1983. In situ determination of soil hydraulic properties during
drainage. Soil Sci. Soc. Am. J., 47:619-624.
68
Dirksen, c., 1999. Soil physical measurements. Geo-ecology paperback, Catena verlag
GMBH, 35447. Reiskirchen, Germany.
Eching, S. O. and Hopmans, J. W., 1993. Optimization of hydraulic functions from transient
outflow and soil water pressure data. Soil Sc. Soc. of America, 57: 1167-1175, 1993.
Eching, S.O., Hopmans, J. W. and Wendroth, 0.,1994. Unsaturated hydraulic conductivity
from transient multistep outflow and soil water pressure data. Soil Sci. Soc. Am. J., 58:
687- 695.
FAO, 2009. Bio-Physical requirements and socio-economic acceptance of infield rainwater
harvesting and conservation In the semi-arid central region of South
Africa.http://www.fao.org.20/09/2009.10:30am.LT.
Fraenkel, C. H., 2008. Spatial variability of selected soil properties in and between map units.
M.sc. (Agric) Dissertation, University of the Free State, Bloemfontein, South Africa.
Gopalakrishnan, K. and Manik, A., 2007. A mathematical model for predicting isothermal
soil moisture profiles using finite different methods. Inter. Jour. of Computational
and Mathematical Sc. 1:128-134.
Gomez, K. A. and Gomez A.A., 1984. Statistical procedures for agricultural research. 2 ed.
John Wiley and Sons, New York, USA.
Hensley, M., Botha, J. J., Anderson, J. J., Van Staden, P. P. and Du Toit, A., 2000.
Optimising rainfall use efficiency for developing farmers with limited access to
irrigation water. Water Research Commission report 878/1/00, Pretoria, South Africa.
Hensley, M., Le Roux, P. A. L., Du Preez, C. c., Van Huyssteen, c., Kotze, E. and Van
Rensburg, L. D., 2007. Soils; the Free State agricultural base. South African
Geographical Journal, 88: 11-21.
Hillel, D., Krentos, V. D. and Styllianou, Y., 1972. Procedure and test of an internal drainage
method for measuring soil hydraulic characteristics in situ. Soil Sci., 114: 395-400.
Hillel, D. H., 2004. Environmental soil physics. Academic Press, New York, USA.
Hurtado, A. L. B., Cichota, R. and Jong van lier, Q., 2005. Parametrizacáo do método do
perfil instantáneo para a determinacáo da condutividade hidráulica do solo em
experimentos com evaporacáo. R. Bras. Ci. Solo, 29: 301-307.
Jones, E. W., Koh, Y. H., Tiver, B. J. and Wong, M. A. H., 2009. Modelling the behaviour of
unsaturated, saline clay for geotechnical design. School of Civil, Environmental and
Mining Engineering, University of Adelaide, Australia.
Jury, W., Gardener, W. R. and Gardner, W. H., 1991. Soil Physics (5th Ed.). John Wiley and
Sons, New York, USA.
69
Kool, J. B., Parker, J. C. and van Genuchten, M. T., 1985. Determining soil hydraulic
properties from one step outflow experiments by parameter estimation: Theory and
numerical studies. Soil Sci. Soc. Am. 1.,49: 1348-1354.
Leij, F. J., Alves, W. J. and van Genuchten, M. T., 1996. UNSODA: The unsaturated soil
hydraulic database. U.S. Salinity Lab., Riverside, CA.
Libardi, P. L., Reichardt, K, Nielsen, D. R. and Biggar, J. W., 1980. Simple field methods
for estimating soil hydraulic conductivity. Soil Sci. Soc. Am. 1.,44: 3-7.
Luxmoore, R., 1981. Micro-, Meso- and macroporosity of soil. Soil Sci. Sc. Am. 1.,45: 241-
285.
Mathews, C. J., Knight, F. J. and Braddock, R. D., 2004. Using analytic solutions for
homogeneous soils to assess numerical solutions for layered soils. ARC Discovery
Indigenous Research Development Program. C.mathews@griffith.edu.au, 31/041
2011, 17.00 pm, (LT).
Neto, D. D., Reichardt, K, Lucia da Silva, A., Bacchi, O. O. S., Timm, L.C., Oliveira, J. C.
M. and Nielsen, D. R., 2007. A software to calculate soil hydraulic conductivity in
internal drainage experiments (SHC, version 2). R. Bras. Ci. Solo, 31: 1219-1222.
Nielsen, D. R., Davison, J. M., Biggar, J. W. and Miller, R. J., 1964. Water movement
through Panoche clay loam soil. Hilgardia, 35: 491-506.
Non Affiliated Soil Analysis Work Committee, 1990. Handbook of standard soil testing
methods for advisory purposes. Soil Science Society of South Africa, Pretoria.
Nhlabatsi, N. N., 2011. Soil water evaporation studies on the GlenlBoheim ecotope. Ph D.
Thesis, University of the Free State, Bloemfontein, South Africa.
Ratliff, L. F., Richie, J. T. and Cassel, D. K, 1983. Field measured limits of soil water
availability as related to laboratory-measured properties. Soil Science Society.
American 1.,47:770-775.
Reichardt, K, 1993. Unit gradient in internal drainage experiments for the determination of
soil hydraulic conductivity. Sci. Agric. Piracicaba. 50: 151-153.
Reichardt, K., Timm, L. C., Bacchi, O. O. S., Oliveira, J. C. M. and Dourado-neto, D., 2004.
A parameterized equation to estimate soil hydraulic conductivity in the field. Aust. J.
Soil Res., 42:283-287.
Richards, L. A., Gardener, W. R. and Ogata, G., 1956. Physical processes determining water
loss from soil. Soil Sci. Soc. Am. Proc., 20: 310-314.
Romano, N., Brunone, B. and Santini, A., 1996. Numerical analysis of one-dimensional
unsaturated flow in layered soils. Advances in Water Res., 21: 315-324.
70
Rose, C. W., Stern, W. R. and Drummond, J. E., 1965. Determination of hydraulic
conductivity as a function of depth and water content for soil in situ. Aust. 1. Soil Res.,
3: 1-9.
Schaap, M. G. and van Genuchten, M. T., 2006. A modified Mualem-van Genuchten
formulation for improved description of the hydraulic conductivity near saturation.
Vadase Zone Journal, 5:27-34.
Scheffler, G., 2007. Determination of instantaneous moisture content and moisture potential
profiles. University of Technology, Dresden, Germany.
Scotter, D. R., Clothier, B. E. and Harper, E. R., 1982. Measuring saturated hydraulic
conductivity and sorptivity using twin rings. Australian Journal of Soil Research, 20:
295-304.
Simunek, J., Wendroth, O. and van Genuchten, M. T., 1999. Estimating unsaturated soil
hydraulic properties from laboratory tension disc infiltrometer experiments. Water
Resour. Res., 35: 2965-2979.
Slawinski, C., Witkowska-Walczak, B. and WaJczak, R. T., 2004. Determination of water
conductivity coefficient of soil porous media. Institute of Agrophysics PAS, Lublin,
Poland.
Soil Classification Working Group, 1991. Soil Classification - A taxonomic system for South
Africa. Memoirs on the Agricultural Natural Resources of South Africa No.15,
Department of Agricultural Development, Pretoria, South Africa.
Tartakovsky, D. M., Lu, Z., Guadagnini, A. and Tartakovsky, A. M., 2003. Unsaturated flow
in heterogeneous soils with spatially distributed uncertain hydraulic parameters.
Journal of Hydrology, 275: 182-193.
The non-affiliated soil analysis work committee, 1990. Handbook of standard soil testing
methods for advisory purposes. SSSSA, Pretoria, South Africa.
Vachaud, G. and Dane, J. H., 2002. Instantaneous profile. In: Dane, J.H. and Topp G. C.
(ed.) Methods of soil analysis. Part 4. Physical methods. SSSA Book Ser. 5. SSSA,
Madison, WI.
Vachaud, G., C. Donko, D. S. and Thony, J. L., 1978. Methodes decaracterisation
hydrodynamique in situ dun sol non sature. Aplication a deux types de sol du Senegal
en vue de la determination des termes du bilan hydrique. Ann. Agron., 29: 1-36.
van Dam, J. C., Stricker, J. N. M., and Droogers, P., 1994. Inverse method to determine soil
hydraulic functions from multistep outflow experiments. Soil Sci. Soc. Am. 1., 58:647
652.
71
Villagra, M. M., Michiels, P., Hartmann, R., Bacchi, O. O. S. and Reichardt, K., 1994. Field
determined variation of the unsaturated hydraulic conductivity functions using
simplified analysis of internal drainage experiments. Sci. agric., Piracicaba, 51: 113-
122.
Vomicil, J. A., 1965. Porosity. In: Black, C. A. (ed), Methods of soil analysis. Amer. Soc.
Agron., Madison Wisc, Part 1: 299-314.
Waston, K. K., 1966. An instantaneous profile method for determining the hydraulic
conductivity of unsaturated porous materials. Water Res. Res., 2: 709-715.
Whitmore, A. P. and Whalley, W. R., 2009. Physical effects of soil drying on roots and crop
growth. Journal of Experimental Botany, 60: 2845-2857.
Wildenschild, D., Hopmans, J. W. and Simunek, J., 2001. Flow rate dependence of soil
hydraulic characteristics. Soil Sci. Soc. Am. J., 65:35-48.
World Reference Base for Soil Resources (1998). World soil resources report, 84, ISSS /
ISRIC FAO, Rome, Italy.
Yeh, T. C. 1., Gelhar, L. W. and Gutjahr, A. L., 1985. Stochastic analysis of unsaturated flow
in heterogeneous soils, 2, statistically anisotropic media with variable a, Water
Resour. Res., 21: 457-464.
Zavattaro, L., and Grignani, c., 2001. Deriving hydrological parameters for modelling water
flow under field conditions. Soil Sci. Soc. Am. J., 65:655-667.
72
CHAPTER4
EVALUATING MODELS FOR PREDICTING HYDRAULIC CHARECTERISTICS
OF LAYERED SOILS
Abstract
Soil water characteristic curve (SWCC) and unsaturated hydraulic conductivity (K-
coefficient) are critical hydraulic properties governing soil water activity on layered soils.
Several parametric based models have been developed to describe these hydraulic functions
under laboratory and in situ conditions. Some of these popular models were validated for the
South African Tukulu and Sepane soil types similar to Cutanic Luvisols and Swartland soil
form also referred to Cutanic Cambisols in other countries. Laboratory measured SWCC was
first described using Brooks and Corey (1964), van Genuchten (1980) and Kosugi (1996)
parametric models. The conductivity functions of these models were then required to fit in
situ based K-coefficients derived from instantaneous profile method (!PM) transient
experiment. The same K-coefficient was also fitted by HYDRUS-ID using optimised SWCC
parameters. Although all parametric models fitted the measured SWCC fairly well their
corresponding conductivity functions could not do the same when fitting the in situ based K-
coefficients. Overestimates of more than two orders of magnitude especially at low soil water
content (SWC) were observed. This phenomenon was pronounced among the upper horizons
that overlaid a clayey horizon. However, optimized a and n parameters using HYDRUS-lD
showed remarkable agreement between fitted and in situ K-coefficient with root sum of
squares error (RMSE) recording values not exceeding unity. During this exercise the Brooks
and Corey was replaced by modified van Genuchten model (Vogel and Cislerova, 1988)
since it failed to produce unique inverse solutions. The models performance appeared to be
soil specific with van Genuchten (1980) performing fairly well on the orthic and neucutanic
horizons while its modified form fitted very well the prismatic and pedo-cutanic horizons.
The lognormal distribution model of Kosugi (1996) showed an extraordinary good fit among
the Swartland profile horizons especially the saprolite rock layer. It was therefore concluded
that in situ K-coefficient estimates from SWCC parameters could be acceptable if only rough
estimates were required. Optimization of parameters for in situ conditions especially for
HYDRUS-ID carried much prospects in characterising the hydraulic properties of most of
the layered soils earmarked for IRWH in the province.
Key words: hydraulic properties, layered soils, parametric models, HYDRUS-lD
73
4.1 Introduction
Characterising soil hydraulic properties of layered profiles has become more attractive
because of the increased availability of computer based models. These models rely on
parameterization of the soil water characteristic curve (SWCC) and unsaturated soil hydraulic
conductivity (K-coefficient) that serve as inputs in soil water flow and solute transport
simulations. The flexibility of this models to describe hydraulic functions for any given flow
domain through parameter optimization implies that most of the inefficiencies from
traditional methods could be avoided.
Successes of parameter based models have been based on integrating indirect and inverse
techniques. Hopmans et al. (2002) defined inverse modelling as a general mathematical
method to determine unknown causes on the basis of observation of their effects. Since
SWCC is the much easier function of soil water content (SWC) to measure, early models
used the Burdine theory (Millington-Quirk, 1961) based on SWCC to predict the K-
coefficient. The convenience of predicting the K-coefficient either as a function of SWC (8)
or suction (h) from SWCC was carried over to subsequent parameters based models with
greater accuracies (Brooks and Corey 1964; Mualem, 1976). Under controlled laboratory
conditions it later proved that SWCC and K-coefficient could be simultaneously determined
from transient outflow experiments (Gardener (1956) by inverse modelling (Valiantzas and
Kerkides (1990). Recent models including among others the van Genuchten-Mualem (van
Genuchten, 1980), modified van Genuchten (Vogel and Cislerova, 1988) and the Kosugi
(1996) have been reported to be more efficient under different conditions (Kosugi et al.,
2002). To improve parameters estimation of various soils the pedo-transfer functions were
developed that use soil physical properties such as pore-size distribution and bulk density as
well as the shape of the SWCC to predict the K-coefficient (Schaap et al., 2001; Hopmans et
al., 2002). Although the application of the pedo-transfer functions could be limited to specific
soil or horizons (Zavattaro and Grignani, 2001) their rough estimates could facilitate the
simulation of water flow and solute transport in the absence of reliable data (Simunek et al.,
2008).
Challenges of inverse modelling including parameter uniqueness, selecting of transient flow
variables and spatial variations of the porous media have been addressed (Zachman et al.,
1982, Kool and Parker, 1988, and Russo et al., 1991; Kosugi and Hopmans, 1998; William
and Ahuja, 2003; Simunek et al., 2008; Hunt and Ewing, 2009). Dane and Hruska (1983) was
74
also among the first to demonstrate that with the correct objective function and parametric
model it was possible to improve inverse solutions from in situ conditions from the
instantaneous profile method (IPM) transient experiment. Essential ingredient of transferring
the 8-h relationship from the SWCC to in situ conditions was that the selected model should
fit the measured data well for the SWC range in question (van Genuchten, 1980; Kosugi et
al., 2002). This was confirmed by Russo (1988) and Chen et al. (1999) who tested various
models, respectively for one-step outflow optimization and multistep outflow experimental
data and found that only a few fitted the measured data very well. Van Genuchten (1980)
model is among the highly rated models and has ability to improve stability of inverse
solution supported by other studies (Russo, 1991; Mallants et al., 1996; Chen et al., 1999;
Simunek et al., 2008). However, excellent results from other models have been reported
(pachepsky et al., 1996; TomaselIa and Hodnett, 1997; Kosugi et al., 2002; Kawamoto et al.,
2006; Lamara and Derriche, 2008), but because of the large number of unknown parameters
their applications is limited (van Genuchten, 1980).
Critical to in situ conditions is taking into account the spatiality of the flow boundary
conditions and horizons differentiation especially in layered profiles. To address the
sensitivity of the optimised parameters to lower boundary conditions and size of input data,
Kool et al. (1987) and Kool and Parker (1988) showed that coupling of the optimization
procedure with simultaneous measurements of pressure heads and SWC improves their
model's predictions. Increased number of measurements was also found to be able to account
better for spatial variations and to improve effect of scaling soil heterogeneity (Abbasi et al.,
2003; Sharma et al. 2011). These models have been shown to be more sensitive to the number
of optimised parameters than the optimization procedure (Zijlstra and Dane, 1996; Abbasi et
al., 2003; Brunone et al., 2003; Simunek et al., 2008).
To address the erratic soil water regime among the cultivated layered soils in the Free State
province of South Africa, the in-field rainwater harvesting (IRWH) technique (Hensley et al.,
2000) has been developed. This technique is geared to convert surface runoff losses into deep
infiltration and soil water storage. However, the success of this initiative and the
sustainability there off could not be possible without appropriate knowledge of the soil
hydraulic properties. Layered soils earmarked for IRWH developed constitutes about 10% of
the provincial landscape (Hensley et aI., 2006). Common in this group of soils includes the
Tukulu and Sepane that are similar to Cutanic Luvisols and the Swartland also similar to the
Cutanic Cambisols of the World Reference Base for Soil Resources (1998). Integrating
75
laboratory and field based analysis of soil hydraulic properties would be essential to this
work. It is therefore within the initiative of this study to attain the following objectives;
Firstly, to characterise the SWCC of the respective horizons of the three soil profiles and
secondly, to validate the conductivity functions based on the SWCC parameters for the
estimation of in situ K-coefficient; Thirdly, to estimate in situ K-coefficient from optimised
SWCC based parameters.
4.2 Material and methods
4.2.1 Experimental site and location
Three soil types cultivated at the Parady's Experimental Farm (29°13'25"S, 26° 12'08"E,
altitude 1417 m) of University of the Free State were selected. These included Tukulu (Tu),
Sepane (Se) and Swartland (Sw) (Soil Classification Working Group, 1991). The SWCC and
K-coefficient were determined under laboratory and in situ conditions, respectively.
4.2.2 Theory
Several parametric models are used to describe the measured SWCC using the shape (a) and
(n) pore size distribution parameters. Among the common parametric models include the
Brooks and Corey (1964), van Genuchten, 1980 and Kosugi (1996). These models constitute
of expression that aid in describing the 8-h relationship of the SWCc. Development of these
models was informed by the general retention equation (4.1) widely accepted to represent the
pore distribution of many of soils.
(4.1)
(4.2)
Where Se is effective saturation, 8s and 8r are the respective saturated and residual values of
the volumetric water content, 8 (mm mm"), h is the matric suction (mm), while a and nare
the shape and pore size distribution parameters, respectively.
Brooks and Corey (1964) reduced the expression (1) into the following general equation
(4.3)
Where a is is the inverse of air entry value and the rest as defined previously. This expression
allows a zero slope to be imposed on SWCC as h equals air entry value. Se equals to unity
76
when hz-L'e. This principle gives SWCC a zero slope at relatively higher suctions since
measuring SWCC above 85% of saturation was considered impractical and general
disconnection of the gas phase (Brooks and Corey 1999). Conductivity function
corresponding to this expression is written as:
= s:~+1+2K K5 e (4.4)
Where, 1 is the pore-connectivity parameter approaching 2. Similarly this expression allows
K-coefficient to approach Ks at or near air entry matric suctions.
The van Genuchten-Mualem retention model (van Genuchten, 1980) assumes the fol1owing
expression:
(4.5)
Where the condition m= l-lIn should be satisfied. According to the mathematical structure of
this model 8 (h) equals 8s when at zero or positive matric suction. The square root
relationship of 8 (h) to n and m parameters gives the SWCC a typical symmetrical shape with
zero slope approached both towards 8s and 8r. The conductivity function referred to as the van
Genuchten- Mualem (1980) equation is given as:
(h) = K,S; { 1 - {1 - S~rr (4.6)
where the parameters are as defined before with n maintained above unity value. This model
has undergone several revisions with the recent modification by Schaap and van Genuchten
(2005). However, one modification incorporated in many computer models is that by Vogel
and Cislerova, (1988). The modification assumes the same structure the expression except
that 8s and 8r are replaced by fictitious parameters given as 8m and 8a, respectively with 8m
slightly greater than 8s and 8a slightly smaller than 8r. This was done to force the SWCC to be
constant between 8s and some small negative matric suction value. Its corresponding
conductivity's expression is complex and was detailed by Schaap and van Genuchten (2005).
The introduction of some small capil1ary height just next to saturation allows the K-
coefficient to be more stable than the original version that exhibited a steep slope at suctions
very close to saturation with air entry value kept around -0.2 kPa. This modification has made
77
it to be highly recommended for clay textured soils (Schaap and van Genuchten, 2005;
Simunek et al., 2008).
Another parametric model of different form is that of Kosugi (1996) that assumes a
lognormal distribution for the retention and conductivity functions. The retention expressions
for Se(h):
(4.7)
K= (4.8)
where symbol a instead of h, and n instead of o as used in the original Kosugi (1996) paper
are adopted for uniformity reasons by some computer optimization programmes such as
RECT and HYDRUS-ID.
4.2.3 Experimental set up and measurements
4.2.3.1 Soil profile classification
Pedological classification of the three soils was carried out on excavated soil profile pits to a
depth of 1 m (Appendix A, B and C). Three soil pits were opened on each representative site.
Diagnostic horizons were described according to the physical, chemical and biological
features used by the SCWG (1991).
4.2.3.2 Soil particle distribution and bulk density
Undisturbed soil core samples from representative profile horizons were taken for the
determination of bulk density and soil textural analysis. Sampling cores had an inner diameter
of 103 mm and a height of 77 mm. Each core was mounted on a hydraulic jack that was
manually operated to extract samples from the various horizons. Particle size distribution was
established using the pipette procedures proposed by the Non Affiliated Soil Analysis Work
Committee (1990). For the determination of gravimetric soil water content soil samples were
oven dried at 105°C for a period of 24 hours.
78
4.2.3.3 Saturated hydraulic conductivity
Saturated hydraulic conductivity (Ks) for the individual profile layers of the three soils were
measured using double ring infiltrometers as described by Scotter et al. (1982). Soil profile
pits were excavated in a stepwise manner to allow the fitting of both rings with diameters of
400 and 600 mm to a depth of 20 mm. The falling head over a distance of 10 mm depth was
used to determine Ks with every fall recorded by means of a timer and a calibrated floater.
After a steady state was recorded for three consecutive times the Ks constant value (mm
hOU(I) was computed using the Jury et al. (1991) formula given as:
(4.9)
where L is the depth of the soil layer in question (mm), bo is the initial depth of total head
above the soil column, bl is the depth the falling head is not allowed to fall below (mm), ti is
the time taken for bo to fall to bl (in hours).
4.2.3.4 Internal drainage experiment
The instantaneous profile method (IPM) (Hillel et al., 1972, Marion et al., 1994) was used to
determine soil water flux and unsaturated hydraulic conductivity from the draining profiles.
Monoliths of 4 m x 4 m surface area with 1 m depth from the Tukulu and Sepane were
prepared in triplicates. From the Swartland monoliths were of 1.2 m x 1.2 m with a depth of
0.5 m. To minimise spatial variations monoliths were semi-detached but difficulty in
excavating the Swartland soil resulted in solitary monoliths at spacing of 30 m interval.
Neutron access tubes to a depth of 1.1 m were installed on the central area of each monolith
section from the Tukulu and Sepane while DFM probes were installed to a depth 0.6 m. Side
walls were isolated with polythene plastics with ridges around sides and a slurry prepared to
seal the sides from the surface. Monoliths were pre-ponded for three executive days and then
covered to keep weather elements from interfering with the trial. Measurements were taken at
the centre of each profile horizon daily for a period of 50 days. Changes in water storage
between time intervals were computed into drainage flux (qDr; mm hour") which was then
described using the Darcy's Law given as:
gDr = -K(e) {:: + 1} (4.10)
79
Where K (e) is unsaturated hydraulic conductivity (mm hour"), dh is the change in matric
suction (mm) between the neighbouring horizons, Z (mm) being the thickness of the horizon
layer in question.
4.2.3.5Measurements of the soil water characteristics curve
Determination of the SWCC for the three soil profile horizons was carried out by a laboratory
desorption experiment. Undisturbed soil samples in triplicates were first de-aired with a
vacuum chamber pump set at -70 kPa for 48 hours under room temperature. De-aired water
was then introduced to saturate samples by capillarity for a 24 hour period. Samples were
then desorbed through the following series of pressure heads including a 0 to -10 kPa, -10 to -
100 kPa and -100 to -1500 kPa. The first phase of desorption involved the hanging water
column method, described by Dirksen (1999). At every step, interval samples were weighted
before and after equilibration. At suctions of -100 to -1500 kPa samples were disturbed and
packed in 2000 mnr' PVC tubes at the measured bulk densities. Measured soil water content
(e) at every matric suction (h) level was plotted to produce the e-h relationship that depicted
the SWCC.
4.2.3.5.1 Mathematical description of soil water characteristic curve
Parametric models use hydraulic parameters to describe the SWCC. A parameter
optimization computer program RECT (van Genuchten et aI., 1991) included some of these
such as the Brooks and Corey (1964), van Genutchen (1980) and Kosugi (1996) models. A
summary of physical characteristics from the three soil profiles used as inputs into these
models are summarized in Table 4.1. Parameters related to pore-size distribution (n) and
shape (a) of the SWCC was initially fitted using the Rosetta pedo-transfer (Schaap et al.,
2001) and then optimised for the respective models.
4.2.3.5.2 Estimating unsaturated hydraulic properties for field based K-coefficient
Hydraulic parameters pertaining to the shape (a) and pore size distribution (n) of the SWCC
were inversely optimised for the description of in situ K-coefficient. The software HYDRUS
-ID (Simunek et al., 2008) was used for this optimization and also for the fitting of the in situ
K-coefficient at plot scale derived by the instantaneous profile method (IPM). This model
numerically solves the Richard flow equation in one dimension using a Galerkin-type linear
finite element scheme.
80
Table 4.1 Summary of the physical characteristics of the three soil types.
Soil Eh~sical EroEerties
Soil forms Tukulu SeEane Swartland
Master Horizons A BI C A BI C A BI C
Coarse sand (%) 5.3 9.2 2.1 5.2 3.5 2.3 4.7 3.2 54.3
Medium sand (%) 9.3 8.8 3.8 lO 4.1 2.3 7.6 5.3 4.6
Fine sand (%) 41.2 31 28.3 41.9 41 31 42 37.6 17.2
Very fine sand (%) 25.3 21 8.4 21.5 10.5 18 31.7 26.6 2.5
Coarse silt (%) 2.1 2 3 I 3 I 2 3 3
Fine silt (%) 4.6 2.5 6.5 I 3 I I 2 3
Clay (%) 11.3 26.4 47.9 19 35 45 11.3 21.9 15
Bulk density (kg m") 1670 1597 1602 1670 1790 1730 1670 1530 1450
Porosity (%) 34.0 33 32.4 34 33.5 33.8 35 39.9 41.6
Ks (mm hour") 36.1 40 9.6(1.9) 35.2 18.1(10.2) 1.9 (I) 23.5 42.8 76.5
Ks = Saturated hydraulic Conductivity
81
A wide variety of parameter optimization problems could be handled through the
minimization accomplished by the Levenberg-Marquardt nonlinear weighted least squares
approach (Hopmans and Simunek, 1999; Lazarovitch et al., 2009; Wollschlager, et al., 2009;
Kandelous and Simunek, 2010).
In this work the objective function contained pressure head from horizons centre block and
K-coefficient's determined by IPM as a function of soil water content (SWC). Preliminary
tests on the parametric models showed that the Brooks and Corey (1964) required
simultaneous optimisation of several parameters, an exercise that compromised the
uniqueness of the inverse solution (Simunek et al., 2008). Therefore, only the van Genuchten-
Mualem (van Genuchten, 1980), Modified van Genuchten (Vogel and Cislerova, 1988) and
Kosugi (1996) models were selected for the optimization of the conductivity function. To
enhance the convergence the Iexponent parameter from the Mualern's (1976) equation was
optimised as well. Initial conditions were set at saturated soil water content for the respective
soil profile horizons. Observation points were fixed at centre block corresponding to field
measurements. The constraint of respecting the model assumptions and the code requirements
was followed in all data analysis.
4.2.4 Statistical analysis
Measured and optimised drainage and unsaturated hydraulic conductivity as well as the
pertinent hydraulic parameters constituted the major findings. The coefficient of
determination (R2), root mean square error (RMSE) and the index of agreement or D-index as
proposed by Willmot et al., (1985), were the statistical tools used to quantify the quality of fit
and variability between measured and fitted data. Al: 1 line was used to show the degree of
scatter of the optimised model fit from the field based K-coefficient.
4.3 Results
4.3.1 Soilwater characteristics curve
Figure 4.1 shows the measured and fitted soil water characteristics curves (SWCC) for the A,
Band C horizons of the Tukulu, Sepane and Swartland soil forms. The fitted hydraulic
parameters from Brooks and Corey (1964), van Genuchten (1980) and Kosugi (1996) models
are summarized in Table 4.2 with their corresponding statistical measure of fit.
82
(a) (i) (ii) (iii)
0.4
,-.. 0.4 0.4
o ,-..
~ 0.3 ~ , ~ 0.3 ~ 0.3 - ----0Uli 0.21 Eli' E-- 0 MeasuredVan Genuchten ê 00.2 E 0.2 0 Measured~ e ~ --- van Genuchten
~ 0.1 - - - Brooks & Corey ~ 0.1 U
E
0.1 - -- Brooks & Corey
C/) ••••••••••• Kosugi ........... Kosugi
C/) ••••••••••• Kosugi ~
0 0 C/) 0
0.1 1 10 100 1000 0.1 I 10 100 1000 0.1 I 10 100 1000
(b) Suction (kPa) Suction (kPa) ~lIrt-tnn (,",Po;'!)
0.4 0.4 0.4
,-..
i;0.3 ~ 0.3E 0
E E 0
E 0.2 E 0_2 --- van Genuchten
S E --- van GenuchtenU - -- i:1:- -- Brooks & Corey0.1 Brooks & Corey~ 0.1 ~ 0_1~ ........... Kosugi........... Kosugi
C/) ••••••••••• Kosugi C/) C/)o ~ 0 0i • ii i iiii • i iiii
0.1 1 10 100 1000 0.1 I 10 100 1000 0.1 I 10 100 1000
(C) Suction (kPa) C"' ___ ._: _.. /,.1"'_ \ Suction -(kPa)
,-..0.4 0.4 -:;:-0.4 0 Measured
E Eo E van Gcnuchtcn
E 0.3 EO.3
~ E 0_3 _ _ _ Brooks & CoreyE 0 E E
E EO.2 o Measured~0.2 ~ ~ 0_2--- van Genuchten
- - - Brooks & Corey U U
~ 0.1 1 - - - Brooks & Corey........•.• Kosugi ~0.1C/) .••.••••••• Kosugi ~ 0.1 v -m.:1
o o ~ '''''hl'' ""hl '''''hl '"
0.0 0_1 10 100 1000
0.1 I 10 100 1000 0.1 1 10 100 1000
('1 ......... ; ........ (1,0 ...\. Suction -(kPa) Suction -(kPa)
Figure 4.1 Measured and fitted soil water content (SWC) and matric suction relationships for the Tukulu (a), Sepane (b) and Swartland (c)
diagnostic horizons A (i), B (ii) and C (iii).
83
Table 4.2 Fitting models to the hydraulic parameters of the SWCC for the Tukulu, Sepane and
Swartland form soils.
Tukulu soil
RetentionModels Horizons es er Ks a n m Dvindex
Brooks & Corey A 0.34 0.13 36.10 0.0019 0.62 1.000 0.977 0.0112 0.9945
van Genuiehen A 0.34 0.13 36.10 0.0012 l.77 0.436 0.995 0.0054 0.9987
Kosugi A 0.34 0.13 36.10 2084.0 1.41 0.359 0.995 0.0054 0.9987
Brooks & Corey B 0.33 0.116 40.00 0.0018 0.47 l.000 0.991 0.0066 0.9979
van Genutchen B 0.33 0.116 40.00 0.0009 1.62 0.381 0.990 0.0046 0.9990
Kosugi B 0.33 0.116 40.00 3378.9 1.59 0.359 0.987 0.0083 0.9967
Brooks & Corey C 0.32 0.26 l.90 0.0079 0.21 1.000 0.925 0.0043 0.9806
van Genutchen C 0.32 0.26 l.90 0.0059 1.22 0.182 0.940 0.0038 0.9841
Kosugi C 0.32 0.26 l.90 4770.7 3.38 0.359 0.964 0.0029 0.9905
Se ane soil
Brooks & Corey A 0.340 0.100 35.19 0.0044 0.31 1.000 0.985 0.0082 0.9963
van Genutchen A 0.340 0.100 35.19 0.0027 1.37 0.270 0.994 0.0054 0.9983
Kosugi A 0.340 0.100 35.19 2787.3 2.45 0.359 0.988 0.0063 0.9977
Brooks & Corey B 0.335 0.190 10.20 0.0025 0.47 1.000 0.988 0.0053 0.9971
van Genutchen B 0.335 0.190 10.20 0.0015 1.59 0.369 0.990 0.0018 0.9996
Kosugi B 0.335 0.190 10.20 2343.6 1.73 0.359 0.997 0.0028 0.9992
Brooks & Corey C 0.338 0.225 l.00 0.0025 0.54 1.000 0.987 0.0045 0.9968
van Genutchen C 0.338 0.225 l.00 0.0014 l.69 0.408 0.998 0.0016 0.9996
Kosugi C 0.338 0.225 l.00 1934.8 1.55 0.359 0.999 0.0013 0.9997
Swartland soil
Brooks & Corey A 0.350 0.100 23.48 0.0025 0.39 1.000 0.974 0.0053 0.9987
van Genutchen A 0.350 0.100 23.48 0.0014 l.50 0.333 0.992 0.0055 0.9999
Kosugi A 0.350 0.100 23.48 3010.0 1.84 0.359 0.990 0.0104 0.9996
Brooks & Corey B 0.399 0.105 42.80 0.0141 0.26 1.000 0.995 0.0095 0.9996
van Genutchen B 0.399 0.105 42.80 0.0088 1.30 0.231 0.999 0.0054 0.9999
Kosugi B 0.399 0.105 42.80 1333.9 3.06 0.359 0.989 0.0040 0.9999
Brooks & Corey C 0.416 0.061 76.50 0.0052 0.69 1.000 0.980 0.0094 0.9990
van Genutchen C 0.416 0.061 76.50 0.0041 1.76 0.433 0.992 0.0100 0.9996
Kosugi C 0.416 0.061 76.50 547.73 1.58 0.359 0.992 0.0162 0.9990
Tukulu: measured and fitted SWCC were fairly similar in all horizons. The measured SWCC
assumed the general segmoidale shape especially from the A and B horizons, while the C
horizon resembled a near linear and small slope. The A, Band C horizons had a 8-h
relationship with saturated (8s) soil water content (SWC) to residual water content (81')
ranging from 0.34 to 0.13 mm mm", 0.33 to 0.12 and 0.324 to 0.26 mm mm", respectively.
Corresponding Ks for these layers were 36.1, 40 and 1.9 mm hour". The van Genuchten-
Mualem and Kosugi models approached 8s at matric suctions of about 0.1 kPa with an air
84
entry of around -1 kPa from the A and B horizons and -1.5 kPa for the C-horizon. On the
other hand the Brooks and Corey's model approached 8s at about -5 kPa for the A and B
horizons and -2 kPa from the C horizon. Fitting a parameter from the Brooks and Corey for
the A, Band C horizons was 0.0019, 0.0018 and 0.0079 with corresponding n values of 0.62.
0.47 and 0.21. From the van Genuchten-Mualem model the A and B horizons shared the
same a value of 0.00 I while the C horizon had a value of 0.005. The n constant from this
model was 1.77, l.62 and l.22 for the respective A, Band C horizons. The Kosugi's model
had a lognormal a parameters of 2084, 3379 and 4770.7 for the respective A, Band C
horizons with corresponding n values of 1.41, 1.59 and 3.38. The R2 from the 1:1 linear
relationship between the measured and predicted values ranged from 0.99 to 0.93 with lowest
coefficient associated with the C horizon from Brooks and Corey model. Similarly, the D-
index ranged from 0.999 to 0.981 with the lowest value depicting the least good fit from the
C horizon. The RMSE from the A horizon ranged from 0.0115 to 0.005 with both van
Genuchten-Mualem and Kosugi models having the lowest values depicting a better fit. From
the Band C horizon the RMSE ranged from 0.004 to 0.005 and 0.003 to 0.008, respectively
with van Genuchten and Kosugi attaining a better fit with the former.
Sepane: a good fit between the measured and fitted SWCC was observed in all three
horizons. The A, Band C horizons had a 8-h relationship with SWC of 0.34 to 0.1, 0.335 to
0.19 and 0.338 to 0.225 mm mm·J with Ks of 35, 10 and 1 mm hour" from the respective A,
Band C horizons. At near saturation the van Genuchten-Mualem and Kosugi models
approached 8s at matric suctions between -0.1 to -0.01 kPa with an air entry value of around -
1 kPa from the three horizons. Brooks and Corey model approached 8s at around -Ll , -2.1
and -2.7 kPa for the A, Band C horizons, respectively. Fitting a parameters from the Brooks
and Corey were 0.004 for the A horizon and 0.003 for the Band C horizons with
corresponding n values of 0.31,0.47 and 0.54 for the A, Band C horizons, respectively. The
a parameter for the van Genuchten model were 0.003 from the A horizon, 0.002 for the B-
horizon and 0.0014 for the C-horizons with corresponding n values of l.37, l.59 and 1.69 for
the A, Band C horizons. Residual squares were in the range of 0.999 to 0.985 and Brooks
and Corey with the lowest coefficients. Similarly the D-index recorded lower agreement from
the Brooks and Corey especially from the A horizon. The RMSE ranged from 0.0082 to
0.0054 and 0.053 to 0.0018 from the A and B horizons respectively with the van Genuchten-
Mualem model recording the lowest value in both cases indicating a better fit. From the C
horizon the RMSE ranged from 0.0045 to 0.0013 with Kosugi attaining the better fit.
85
Swartland: Greater variability between measured and fitted SWCC were observed in this soil.
The A horizon recorded a SWC ranging from 0.35 to 0.1 mm mm-I and B horizon with a
range from 0.399 to 0.105 mm mm-I. The C-horizon had a SWC ranging from 0.412 to 0.06
mm mm-I. Saturated hydraulic conductivity was 23.5 mm hour-I for the A horizon with
42.8 and 76.5 mm hour" for the respective Band C horizons. Brooks and Corey's model had
0. parameter that increased with depth ranging from 0.0025 to 0.00141. In a similar manner
this parameter ranged from 0.0014 to 0.0088 for the van Genuchten model. From the Kosugi
model 0. decreased with depth at a range of 3010 to 547.73. Corresponding n values in the A,
Band C horizons ranged from 0.26 to 0.688 from the Brooks and Corey, 1.30 to 1.764 for the
van Genuchten-Mualern and 1.58 to 3.06 for the Kosugi models, respectively. Models
approached saturation at 0.1 to 0.01 kPa with the exception of Brooks and Corey that
recorded 2.6, 2.1 and 1.9 kPa in the A, Band C horizons. The R2 was in the range of 0.99 to
0.92 with Brooks and Corey recording the least from the A horizon. Similarly the D-index
recorded lowest agreement from the Brooks and Corey especially from the A horizon. The
RMSE ranged from 0.0104 to 0.0053 and 0.0095 to 0.004 from the respective A and B
horizons with better fit obtained from Brooks and Corey in the former and Kosugi in the
latter. From the C horizon the RMSE ranged from 0.0162 to 0.0094 with a better fit attributed
to Kosugi model.
4.3.2 Predicting K-coefficient from soil water characteristics curve
Figure 4.2 to 4.4 illustrates the variation between in situ K-coefficient derived from
instantaneous profile method (!PM) and estimated by conductivity based parametric models.
Fitting parameters from measured SWCC including Kswere used to predict K-coefficient as a
function of SWC. Statistical measure of fit from estimated K-coefficients are summarised in
Table 4.3 for the Tukulu, Sepane and Swartland soil horizons.
Tukulu: deviation between in situ and estimated K-coefficient could be observed especially at
lower SWC. The in situ K-coefficient in the A horizon ranged from 0.006 to 36.1 mm hOU(I,
for SWC range from 0.297 to 0.34 mm mm-I. For the same SWC range the estimated K-
coefficient ranged from 9.72 to 36.08, 3.36 to 36.1 and 2.18 to 36.1 mm hour-I from the
Brooks and Corey, van Genuchten-Mualem, and Kosugi models. In the same order these
models for the B horizon estimated K-coefficient within the range of 3.93 to 40, 14.13 to 40
and 2.73 to 40 mm hour", while the in situ ranged from 0.0006 to 0.0006 mm hour".
86
(a)
100 ..,----------------------------,
,_
o:l 10
...c
E
E o insitu
..... 0
c 0.1 0 -- van Genuchten-MualernQ)
u 00
<..:: 0.01 /00 - - - Brooks & CoreyQ)o
u ........... Kosugi
~ 0.001
0.0001
0.29 0.3 0.31 0.32 0.33 0.34 0.35
(b) Soil water content ( mm mm")
100
,-.._
\... 10 -----------------~:l
0 .......................,.'........ 0
...c o
E 1
E o
'.-...".
c 0.1
Q) o insitu
u
<..:: 0.01 -- van Genuchten- Mualem
Q)
0
u - - - Brooks and Corey
I
~ 0.001 ........... Kosugi
0.0001
0.29 0.3 0.:1 I 0.32 0.33 0.34 0.35
(C) .~()1 1 \AI::Jtprrnntpnt ( mm mm :1,
100
0 insitu
,-.._
10
\... -- van Genuchten & Mualem
:l
0 - -- Brooks & Corey
...c
E ........... Kosugi ..6..
E .... 0
'.-...". 0.1
c .. !
Q)
u
<..:: 0.01
Q)
0
u
I
~ 0.001 ..'
.'
0.0001 ..'
0.29 0.3 0.31 0.32 0.33 0.34 0.35
Soil water content (mm mrrr ')
Figure 4.2 Comparison of K-coefficient from in situ and fitted by retention models for the
Tukulu A (a), B (b) and C (c) diagnostic horizons.
87
(a)
100 ~--------------------------------------------------------~
,-... 10 _------------------cr--------~
\....
o:::l ----- 0 ...s:: 1 o . .................E
..E__, .................... ·················ëj··············· .
E 0.1
~
ë:i o 0/000 o insltu't 0.01 -- van Genuchten-Mualerno
CJ - - - Brooks & Corey
~.001 ........... Kosugi
0.0001 +--.---.---.---.-...,---,--,--.--.--.---r---r---.---.---,---.---.----..---..--,----,--,---,,--,.--,.--,....-.,.-l
0.29 0.3 0.31 0.32 0.33 0.34
(b) Soil water content (mm mrrr ')
100 ,-----------------------------,
,-...
\....
:o::l 10..s:: - -- ----- -----
E
E ................................ o..__,
ë o~ 0.1
CJ 00
;.;:
~ 0.01 /000 o insitu
CJ
~ -- van Genuchten & Mualem
0.001 - - - Brooks & Corey
........... Kosugi
0.0001
0.29 0.3 0.31 0.32 0.33 0.34
(c) ~oil w::.tpr rontpnt (mm rnrrr l)
100
,-... o insitu
\....
:::l 10 -- van Genuchten & Mualem
..0s:: - - - Brooks & Corey
E 1 ........... Kosugi
..E__, .---------...... ~ .., 0t: #
.~ 0.1
CJ
;.~;:
o
0 0.01
CJ /,000
I
~
0.001
0.0001
0.29 0.3 0.31 0.32 0.33 0.34
Soil water content (mm mm')
Figure 4.3 Comparison of K-coefficient from in situ and fitted by retention models for the
Sepane A (a), B (b) and C (c) diagnostic horizons.
88
o
o o o insituo -- van Genuchten & Mualem
00 - - - Brooks & Corey
........... Kosugi
0.27 0.29 0.31 0.33 0.35 0.37 0.39 0.41
Soil water content (mm mm')
100 ~---------------------------------------------------.
,...._,
10 ---\.... fPJ----
o;::l 1""\..c: o;y-
_,-
_ - "ï:>
E ..'
E
':=:' 0.1<I) ............................................
~ 0.01
<I)
o ,(o5)0 ......... 0 insitu
CJ ............... -- van Genuchten- Mualem
~0.001 .......... - - - Brooks & Corey
........... Kosugi
0.0001
0.23 0.25 0.27 0.29 0.31 0.33 0.35 0.37 0.39 0.41
(c) Soil water content (mm rnrrr')
\....
;::l 10
..oc:
E
E
'.-..".."
.c92 0.1 000
r.C;:J: 0
<I) o insitu
0 0.01
CJ
I -- van Genuchten & Mualem
~
0.001 - - - Brooks & Corey
........... Kosugi
0.0001
0.23 0.25 0.27 0.29 0.31 0.33 0.35 0.37 0.39 0.41
Soil water content (mm mm+)
Figure 4.4 Comparison of K-coefficient from in situ and fitted by retention models for the
Swartland A (a), B (b) and C (c) diagnostic horizons.
89
In the C horizon the in situ K-coefficient ranged from 0.0012 to 1.9 mm hour'l while
estimates from Brooks and Corey, van Genuchten-Mualern, and Kosugi models ranged from
0.0689 to 1.9, 0.003 to 1.9 and 0.00002 to 1.9 mm hour", respectively. The R2 from the A
and B ranged from 0.64 to 0.95 and 0.53 to 0.89 with the lowest values associated with
Brooks and Corey. In the C horizon R2 ranged from 0.78 to 0.99 with the lowest value.
Table 4.3 Statistical measure of fit for conductivity based parametric models on in situ K-
Coefficient for the Tukulu, Sepane and Swartland soil horizons.
Tukulu Seeane Swartland
Models Horizons R2 RMSE D-index R2 RMSE D-index R2 RMSE D-index
Brooks &Corey A 0.64 11.467 0.577 0.85 5.337 0.881 0.79 4.558 0.812
van Genutchen-Mualem A 0.90 4.046 0.928 0.96 2.35 0.974 0.99 0.381 0.998
Kosugi A 0.95 2.763 0.966 0.94 2.78 0.965 0.99 0.312 0.999
Brooks and Corey B 0.53 18.218 0.359 0.59 4.461 0.405 0.59 7.531 0.895
van Genutchen- Mualem B 0.83 6.678 0.851 0.91 1.160 0.932 0.93 4.089 0.95
Kosugi B 0.89 5.124 0.910 0.97 0.656 0.978 0.96 3.794 0.966
Brooks and Corey C 0.99 0.105 0.990 0.52 0.532 0.235 0.93 7.471 0.948
van Genutchen-Mualem C 0.82 0.357 0.852 0.87 0.148 0.884 0.99 1.229 0.999
Kosugi C 0.78 0.377 0.824 0.83 0.175 0.839 0.99 1.580 0.998
Sepane: considerable variability between in situ and estimated K-coefficient was observed
especially at lower SWc. The !PM captured SWC ranging from 0.292 to 0.34, 0.313 to 0.335
and 0.323 to 0.338 mm mm,l from the respective A, Band C horizons. Corresponding K-
coefficient ranged from 0.0013 to 35.19, 0.0014 to 10.2 and 0.0006 to 1 mm hou(l. For the
same SWC range the estimated K-coefficient from the A horizon were 3.95 to 35.19, 0.621 to
35.19, and 0.0953 to 35.19 mm hour" from Brooks and Corey, van Genuchten-Mualern, and
Kosugi models. Corresponding predictions from these models for the Band C horizons were
3.47 to 10.2 and 0.433 to 1,0.81 to 10.2 and 0.101 to 1, and 0.413 tolO.2 and 0.112 to 1 mm
hour", respectively. In situ based K-coefficients from the B horizon ranged from 0.0014 to
10.2 mm hour", while from the C horizon ranged from 0.0006 to 1 mm hour". The R2 from
the A, Band C horizons ranged from 0.85 to 0.96, 0.59 to 0.97 and 0.52 to 0.87, respectively,
with Brooks and Corey scoring the lowest value in all three horizons. Van Genuchten-
Mualem model had a better fit from the A and C horizons shown by RMSE of 2.35 and 0.148
and D-index values of 0.97 and 0.88, respectively. The Kosugi model recorded a better fit in
the B horizon with RMSE of 0.656 and D-index of 0.98.
90
Swartland: remarkable spatiality between in situ and estimated K-coefficient was also
observed. During the drainage period SWC ranged from 0.27 to 0.35, 0.28 to 0.35 and 0.23
to 0.42 mm mmo! from the A, Band C horizons, respectively. Corresponding to these
horizons was the in situ K-coefficient ranging from 0.0002 to 23.48, 0.0028 to 42.8 and 0.002
to 76.5 mm hour". Estimated K-coefficient by Brooks and Corey, van Genuchten-Mualem,
and Kosugi models from the A horizons were 1.4 to 23.48, 0.22 to 23.48, and 0.132 to 23.48,
respective. From the Band C horizons these models in their respective order estimated K-
coefficient within the range of 1.18 to 42 and 0.37 to 76.5, 4.63 to 42.8 and 1.42 to 76, and
0.52 to 42.8 and 0.13 to 76.5 mm hour". Estimates from Brooks and Corey had the lowest R2
of 0.79,0.59 and 0.93 from the respective A, Band C horizons. The Kosugi had a better fit in
the A reflected by RMSE of 0.312 and D-index of 0.999. In the B-horizon van Genuchten-
Mualem and Kosugi model shared similar RMSE and D-index values of 3.7 and 0.97,
respectively. Kosugi also had a better fit in the C horizon with RMSE of 0.853 and D-index
ofO.999.
4.3.3 Parameter optimisation for HYDRUS -ID application
Table 4.4 shows the optimised a and n parameters for the fitting of the in situ based K-
coefficient using HYDRUS-ID program. Statistical measure of coefficient of determination
from these optimised parameters corresponding to the fixed 8s, 8r and Ks values were also
shown. Figure 4.5 illustrates how the fitted K-coefficient compared with the field based K-
coefficient on a l:lline from the Tukulu (a) Sepane (b) and (c) Swartland soils for the A (i)
B (ii) and C (iii) horizons.
Tukulu: a general greater degree of accuracy could be observed from the fitting of the field
based K-coefficient by optirnised parameters using the inverse solution. The van Genuchten-
Mualem (van Genuchten, 1980) and modified van Genuchten (Vogel and Cislerova, 1988)
models shared the same optirnised o of 0.0015 and n of 2.511 to attain a better fit with a
RMSE of 0.324 and D-index of 0.9996 in the A horizon. The Kosugi (1996) models had used
the same optirnised a value of 1000 and n value of 1 for all three horizons. A better fit was
realised in the B horizon with a RMSE of 0.234 and a D-index of 0.9995. The modified van
Genuchten model showed a better fit with a value of 0.0055 and n value of 1.656 for the C
horizon. The RMSE from this soil ranged from 1.2 to 0.023 and D-index from 0.999 to 0.91.
The crowding of the fitted data along the 1:1 line confirms this statistics not only at higher
91
SWC but also for lower SWC with Band C horizons showing a good agreement for the entire
curve.
Table 4.4 Optimised parameters for the fitting of in situ K-coefficient from the Tukulu,
Sepane and Swartland soil horizons using HYDRUS -ID.
Tukulu soil
Conductivit~ models Horizons a n R2 RMSE O-index
van Genutchen & Mualem A 0.0015 2.511 0.5 0.9992 0.324 0.9996
Modified van Genuchten A 0.0015 2.511 0.5 0.9992 0.324 0.9996
Kosugi A 1000 1.000 0.5 0.9891 1.233 0.9937
van Genutchen & Mualem B 0.0019 2.000 0.5 0.9999 0.324 0.9997
Modified van Genuchten B 0.0039 2.000 0.5 0.9932 1.071 0.9888
Kosugi B 1000 1.000 0.5 0.9997 0.234 0.9996
van Genutchen & Mualem C 0.0026 1.500 0.5 0.9988 0.278 0.9046
Modified van Genuchten C 0.0055 1.646 0.5 0.9993 0.023 0.9995
Kosugi C 1000 1.000 0.5 0.9502 0.188 0.9638
Se ane soil
van Genutchen & Mualem A 0.0030 3.514 0.5 0.9997 0.200 0.9997
Modified van Genuchten A 0.0027 3.461 0.5 0.9997 0.202 0.9998
Kosugi A 405.62 0.590 8.9 0.9996 0.224 0.9998
van Genutchen & Mualem B 0.0007 2.368 0.5 0.9989 0.164 0.9978
Modified van Genuchten B 0.0023 1.915 0.5 0.9999 0.041 0.9998
Kosugi B 1113.70 1.228 8.7 0.9998 0.050 0.9997
van Genutchen & Mualem C 0.0004 1.500 0.5 0.9992 0.013 0.9992
Modified van Genuchten C 0.0005 1.751 0.5 0.9999 0.002 0.9998
Kosugi C 683.42 1.015 32.7 0.9999 0.002 0.9999
Swartland soil
van Genutchen & Mualem A 0.0017 1.85000 0.5 0.9998 0.025 0.9990
Modified van Genuchten A 0.0018 1.97700 0.5 0.9998 0.025 0.9990
Kosugi A 877.19 1.010 65.8 0.9999 5.092 0.7280
van Genutchen & Mualem B 0.0028 2.4300 0.5 0.9998 0.155 0.9980
Modified van Genuchten B 0.0028 2.4300 0.5 0.9998 0.155 0.9980
Kasugi B 539.08 0.910 16.0 0.9999 0.119 0.9990
van Genutchen & Mualem C 0.0019 1.760 0.5 0.8560 3.285 0.9890
Modified van Genuchten C 0.0015 1.760 0.5 0.8560 3.285 0.9890
Kosugi C 949.69 1.000 30.6 0.8900 2.877 0.9921
Bold letters = optimised to improve objective function solution
92
(i) (ii) (iii)
G&]100 100 v------::---;----------:.or
o van Genuchten-Mualem
~ 10 + 100 ~ ~10 ~ Modified van Genuchten o vrIn Genuchten-Mualem
-I... I:;. Kosugi 10
::I + Modified van Genuchten
..0c 1 1 --I:lline I:;. Kosugi
E 1 --I:llineg 0.1 0.1 0.1
::.: 0.01
" 0.01§ 0.01
ii: 0.001 0.001 0.001
0.0001 0.0001 iL", """'" ,,""'ij , , ..""I 0.0001 iL" ,L .."" '".".'I , " .. ".I
100u---------------~
~100 100) o van Genuchten-Mualem e van Genuchten- Mualem
~ 10 + Modified van Genuchten 10 10 + Modified van Genuchten
I...
::I I:;. Kosugi I:;. Kosugi
..0c 1 --I:lline 1 --l:lline
gE 0.1 0.1 0.1
::.: 0.01 0.01 0.01-e
OJ
.E 0.001 0.001 0.001
LI..
0.0001 0.0001 0.0001
00100 100 .---------------~0 van Genuchten- Mualern 100 0 van Genuchten- Mualem o van Genuchten- Mualem
~ 10 ~ 10 + Modified van Genuchten~ 10+ + Modified van GenuchtenI:;. Kosugi .db
::I 1 I:;. Kosugi
.0.c --I:lline ., +++E 0.1 0.1 .+ --I:lline0.1
~ 0.01 0.01 0.01
"g 0.001 0.001 0.001
li:
0.0001 0.0001 0.0001
0.0001 0.001 0.01 0.1 1 10 100 0.0001 0.001 0.01 0.1 10 100 0.0001 0.001 0.01 0.1 10 100
In situ K (mm hour') III situ K (mm hour') III situ K (mm hour')
Figure 4.5 Comparison of in situ and fitted K-coefficient for the Tukulu (a), Sepane (b) and Swartland (c) diagnostic horizons A(i), B(ii) and C
(iii) using HYDRUS-ID optimised parameters.
93
Sepane: the most accurate fit were recorded from this soil. The R2 and D-index from fitting
models exceeded 0.99 from all the soil horizons. Fitted data clustered closely to the 1:1 line
especially from the A and C horizons. The van Genuchten-Mualem model with a of 0.003
and n of 3.514 attained a better fit from the A-horizon with a RMSE of 0.2, followed by the
modified form with RMSE of 0.202. The latter also showed a better fit in the Band C horizon
alongside Kosugi model with a RMSE as low as 0.04 to 0.002, respectively.
Swartland: improved accuracy and better fit were limited to the upper horizons of this soil
profile. The modified van Genuchten model attained better fit from the A horizons with
RMSE of 0.02 while the Kosugi model attained better fit from the Band C horizons with
RMSE of 0.12 and 2.8. All model had the poorest fit coming from the C horizon given that
R2 ranged from 0.90 to 0.80.
4.4 Discussion
The goodness of fit from characterising SWCC to the estimating of in situ K-coefficient was
found to vary among models and soil types. Optimised parameters also exhibited the same
trend even though consistency with in situ K-coefficient showed remarkably improvement.
Could these findings therefore, indicate that these hydraulic models were well developed to
produce better fit than others or for specific soil types?
Fitting of SWCC was satisfactory from all the parametric models used. Of interest was the
manner at which the traditional'S' shape of the SWCC was modified among the three soils
from the models. Among the sandy textured orthic and neocutanic horizons this shape was
well defined while from strongly textured prismacutanic and pedocunatic horizons it was
diffused to almost a straight line. These findings were similar with those made by Wilson et
al. (1997), Wildenchild et al. (2001), Chimungu (2009), and Fraenkel (2008) on sandy and
clayey horizons. Narrow pore size distribution from the former could have triggered rapid
drainage rates at near saturation due to the small air entry values associated with these soils
(Kosugi et al., 2002). Consequently, the remaining SWC could have been subjected to very
steep hydraulic gradients as suctions are increased giving the SWCC a symmetrical shape
observed by van Genuchten (1980), Simunek et al. (2008) and Lamara and Derriche (2008).
Given the diverse pore pathways in clayey horizons a strong matrix activity could have been
triggered as suctions were increased resulting to an appreciable but slow release curve
consistent with structured soils (Wildnchild et al., 2001; Zavattaro and Grignani, 2001).
94
Nevertheless, the parametric models were consistent with the shape of the measured SWCC
with the exception of Brooks and Corey especially at near air entry value. This was not a
strange phenomenon since this model imposes a zero slope on the SWCC at near air entry
point given the assumption that measuring saturations above 85 % was impractical, because
of the general disconnection of the gas phase at this range of SWC (van Genuchten, 1980;
Brooks and Corey, 1999). The Brooks and Corey (1964) was also reported to produce poor fit
in soils with an S-shaped SWCC such as fine textured soils and undisturbed field soil (Kosugi
et al., 2002).
Application of the conductivity functions based on SWCC parameters to estimate in situ K-
coefficient derived by the IPM produced very high inconsistencies. Agreements were only
limited at and near the saturation domain. In situ K-coefficient was generally overestimated
by more than 2 to 3 orders of magnitude. This was well pronounced among the upper
horizons of the Tukulu and Sepane. Overestimates from Swartland were less than 2 orders of
magnitude especially at low SWC. Zavattaro and Grignani (2001) also observed that
overestimation of K-coefficient was common among horizons overlying clay rich soils given
the effect of abrupt transition on flow rates. Wildenchild et al. (2001) substantiated this by
demonstrating that K-coefficient was dependent on flow rates. Sensitivity of the inverse
problem to the lower boundary condition as could also be another factor (Dane and Hruska
1983; Hopmans et al. 2002). Given the absence of layering from soil columns used to
measure SWCC it was reasonable to consider that estimates of K-coefficient would be
generally greater than those derived from IPM under in situ environment. Similar results were
recorded by Abbasi et al. (2003) when infiltration data, SWC and solute concentrations from
layered soil were compared with predicted values.
Differences in suction range under which SWCC and in situ coefficient were determined
could be another source of inconsistency. This view was also supported by Zavattaro and
Grignani (2001). The SWCC was determined using matric suction range of -0.1 to -1500 kPa
while from the !PM ranged from -0.01 to about -1 kPa. The latter depicts structural pore
domain while the former constitutes all the entire pore size distribution (Nhlabatsi, 2010).
Therefore, desorption of pores in the !PM was only represented by a small section on the
SWCc. Although the parametric models were able to transfer the 8-h relationship from
SWCC to of the field, the accuracy of this function was said to depend on how good these
models fitted the SWCC (van Genuchten, 1980; Kosugi et al., 2002). This situation could be
95
illustrating why inconsistencies from Brooks and Corey model were pronounced irrespective
of soil type. Van Genuchten (1980) observed that if 81'was inaccurately measured among the
input parameters, given that it falls outside the !PM flow domain, inconsistencies when in situ
K-coefficient is estimates may arise. Under these circumstances the shape and pore-size
distribution parameters representing 8-h relationship affecting the !PM based K-coefficient
would need to be different from SWCC. This view was also shared by Zavattaro and
Grignani (2001) who suggested that among others the shape (a) and pore size distribution (n)
parameters of SWCC should be optimised for in situ applications. It was therefore not
surprising that the use of SWCC based conductivity functions generally produced poor
estimates of in situ K-coefficients especially at low SWC irrespective of soil type. Despite
these inconsistencies some estimates from Brooks and Corey in the Tukulu C-horizon and
Kosugi in the Swartland horizons were within acceptable range. The coarseness of these
horizons could be the reason.
The view of optimising SWCC parameters for in situ application was then validated using
HYDRUS-ID hydrological model. Inverse solution converged readily when a and n
parameters were optirnised allowing the rest of the parameters fixed under the constant SWC
lower boundary condition for the Tukulu and Sepane while Swartland was allocated free
drainage. The resulting matric suction distribution with depth from optimised parameters was
shown in Figure 6. Matric suctions ranged from 0 to -700, 0 to -550 and -150 to -650 mm in
the the Tukulu, Sepane and Swartland soil profiles consisted with matric suctions
corresponding to structural pore domain (Kosugi et al., 2002 ; Wang et al., 2003; Chimungu,
2009; Nhlabatsi, 2010). Evidence of differences in physical properties at interface could be
seen by the change of regime in matric suction with depth. This could also be used to
illustrate the extent of layering among the three soil profiles and the effect on hydraulic
properties. A build up in positive matric suction at depth corresponding to the prismatic C-
horizons of the Tukulu and Sepane after 24 hours of drainage indicates the restrictiveness of
this horizon an observation that was also made by Chimungu (2009) and Fraenkel (2008) on
the respective profiles. Interestingly, optimised a values appeared to increase with depth on
profile with restrictive horizons while from free draining Swartland profile the opposite was
observed. Could it be that this spatial; response of parameters is indicative of their sensitivity
to the soils profile layers boundary conditions. At present any meaningful insight about this
observation could not be drawn. Failure to attain saturation from the Swartland after deep
96
wetting was also indicative of the high permeability of this profile especially the underlying
saprolite rock with average Ks of 76.5 mm hour".
Improvements in R2 from optimised parameters were noticeable from the three soils
irrespective of model used. However, level of agreement was variable among soils and
horizon layers. The van Genuchten-Mualem model produced consistent K-coefficient
estimates among the weakly structured horizons of the Tukulu and Sepane soil profiles. The
modified van Genuchten model produced better fit on the clay rich horizons while Kosugi
model matched fairly well with K-coefficient from the Swartland soil. The use of lognormal
distribution to describe pore size distribution among the Swartland layers could have given
the Kosugi model a better urge over the van Genuchten expressions. Given the exponential
mathematical background of the van Genuchten-Mualem model it is reasonable to associate
its better fit in the orthic A and neocutanic B horizons to the well defined pore size
distribution of the sandy texture that this horizons have. Soils of the same texture were
associated with a typical S-shape SWCC and without a distinct air entry value (Kosugi et al.,
2002). However, due its exponential inheritance when n value approaches its lower limit of 1
then K-coefficient drops sharply at near saturation that could result to poor fit on clay soils
with an appreciable drainage (Schaap and van Genuchten, 2006). The introduction of an
extrapolated parameter that impose a non zero matric suction from saturation to some very
small matric suctions could be attributed to the better fit of the modified van Genuchten
model among the clay rich horizons (Schaap and van Genuchten, 2006; Simunek et al.,
2008).
4.5 Conclusions
The question on to what extent could in situ based K-coefficient be estimated from SWCC
based parameters was evaluated. Firstly the SWCC were parameterised using the Brooks and
Corey (1964), van Genuchten (1980) and Kosugi (1996). The three models fitted fairly well
the retention functions from the soil profile horizons. Brooks and Corey model showed some
deviation near or at the entry point given its assumption to impose a zero slope at this matric
suction from saturation. Secondly, the conductivity based expressions that rely on SWCC
parameters from these three models were used to fit the in situ K-coefficient derived from
IPM. Coefficient of determination from all models showed remarkable overestimated of the
in situ K-coefficient by more than two orders of magnitude, especially at lower SWc. This
finding gives the impression that SWCC parameters were not necessary appropriate to
97
describe in situ based K-coefficient but could be used when rough estimates are required
particularly from layered soil. Thirdly, the SWCC based parameters were inversely optimised
to serve as input to the HYDRUS-ID hydrological for the fitting of in situ K-coefficient from
a plot scale simulation. Agreement between fitted and in situ K-coefficient improved by more
than one order of magnitude from those obtained from SWCC parameters. The van
Genuchten-Mualem model by van Genuchten (1980) appeared to be well posed for sandy
textured soils while the modified van Genuchten model by Vogel and Cislerova (1988) was
well posed for clay rich horizons. On the other hand the lognormal distribution model of
Kosugi (1996) was able adapted for the Swartland soil profile layers. Over and above, the
optimization procedure as well as the application of HYDRUS-ID carries a lot of prospects
in improving the knowledge on the hydraulic properties that characterise layered soils
earmarked for IRWH.
Acknowledgements
Thanks to Prof. Malcom Hensley for his assistance during the laboratory desorption
measurements and classification of the three soil profiles. Special thanks also to Prof. A. H.
Cloot for the constructive comments and suggestions to improve the manuscript.
98
4.6 References
Abbasi F., Jacques, D., Simunek, J., van Genuchten, M.Th., 2003. Inverse estimation of soil
hydraulic and solute transport parameters from transient field experiments:
Heterogeneous soil. Am. Soc. Of Agric. Eng., 46: 1097-111l.
Brooks, R. H., and Corey, AT., 1964. Hydraulic properties of porous media. Hydrology
paper no.3.Civil Engineering Dep. Colarado State University, Fort Collins, USA.
Brooks, R. H., and Corey, AT., 1966. Properties of porous media affecting fluid flow. J.
Irrig. Drain Div., Am. Sc. Civil Eng., 92: 61-88.
Brunone, B., Ferrante, M., Romano, N., and Santini, A, 2003. Numerical simulations of one-
dimensional infiltration into layered soils with the Richard equation using different
estimates of the interlayer conductivity. Vadase Zone 1., 2: 193-200.
Campbell, G. S., 1974. A simple method for determining unsaturated conductivity from
moisture retention data. Soil Sci., 117:311-314.
Chen, J., Hopmans, J. W. and Grismer, M. E., 1999. Parameter estimation of two-fluid
capillary pressuresaturationand permeability functions. Adv. Water Resour., 22: 479-
493.
Chimungu, J. G., 2009. Comparison of field and laboratory measured hydraulic properties of
selected diagnostic soil horizons. M.sc. (Agric ) Dissertation, University of the Free
State Bloemfontein, South Africa.
Dane, J. H., and Hopmans, J. W., 2002 . Soil Water Retention and Storage - Introduction. IN:
Methods of Soil Analysis.Part 4.Physical Methods.(J.H. Dane and G.C. Topp,
Eds.).Soil Science Society of America Book Series No. 5: 671-674.
Dane, J. H., and Hruska, S., 1983. In-situ determination of soil hydraulic properties during
drainage. Soil Sci. Soc. Am. 1.,47:619-624.
Dirksen, C., 1999. Soil physics measurements. GeoEcology paperback, Catena Verlag
GMBH, 35447 Reiskirchen, Germany.
Fraenkel, C. H., 2008. Spatial variability of selected soil properties in and between map units.
M.sc. (Agric ) Dissertation, University of the Free State, Bloemfontein, South Africa.
Gardner, W. R., 1956. Calculation of capillary conductivity from pressure plate outflow data.
Soil Sci. Soc. Am. Proc., 20:317-320.
Gupta, S.D., Mohanty, B. P., and Kohne, J. M., 2006. Soil hydraulic conductivities and their
spatial and temporal variation in a vertisol. Soil Sci. Soc. Am. 1., 70: 1872-188l.
99
Hensley, M., Botha, J. J., Anderson, 1. J., Van Staden, P. P., and Du Toit, A., 2000.
Optimising rainfall use efficiency for developing farmers with limited access to
irrigation water. Water Research Commission report, 878, Pretoria, South Africa.
Hensley, M., le Roux, P.A. L., Du Preez, C. C., van Huyssteen, C.W., Kotze, E.,and van
Rensburg, L.O., 2007. Soils: The Free State's agricultural base. South African
Geographical Journal, 88: 11- 21.
Hillel, D., Krentos, V. D., and Stylianou. Y., 1972. Procedure and test of an internal drainage
method for measuring soil hydraulic characteristics in situ. Soil Sci., 114:395-400.
Hopmans, J. W., and Simunek, J., 1999. Review of inverse estimation of soil hydraulic
properties. 643-659.1n M.Th. van Genuchten et al. (ed.) Characterization and
measurement of hydraulic properties of unsaturated porous media. University of
California, Riverside, CA.
Hopmans, J.W.,Simunek, J., Romano, N., and Durner,W., 2002. Simultaneous determination
of water transmission and retention properties. Inverse Methods. IN: Methods of Soil
Analysis.Part 4.Physical Methods.(J.H. Dane and G.C. Topp, Eds.). Soil Science
Society of America Book Series No. 5: 963-1008.
Jury, W., Gardener, W.R., and Gardner, W.H., 1991. Soil Physics (5th Ed.). John Wiley and
Sons, New York, USA.
Kandelous, M. M., and Simunek, J., 2010. Comparison of numerical, analytical, and
empirical models to estimate wetting patterns for surface and subsurface drip
irrigation. Irrig. Sci., 28:435-444.
Kawamoto, K, Moldrup, P., Ferre, T. P. A., Tuller, M., Jacobsen, O. Komatsu, T., 2006.
Linking the Gardner and Campbell Models for Water Retention and Hydraulic
Conductivity in Near-Saturated SoiJ.Soil Science, 171: 573-584.
Kool, J. B., and Parker, J. C., 1988. Analysis of the inverse problem for transient unsaturated
flow. Water Resour. Res., 24: 817-830.
Kosugi, K, 1996. Lognormal distribution model for unsaturated soil hydraulic properties,
Water Resour. Res., 32: 2697-2703
Kosugi, K, Hopmans, J. W., and Dane, J. H., 2002. Water Retention and Storage -
Parametric Models. IN: Methods of Soil Analysis. Part 4.Physical Methods.(J.H.
Dane and G.C. Topp, (Eds.). Soil Science Society ofAmerica Book Series No. 5: 739-
758.
Lamara, M., and Derriche, Z., 2008. Predictions of unsaturated hydraulic properties of dune
sand on drying and wetting paths. Vol (13) Bund.B, EJGE.
100
Lazarovitch, N., Warrick, A.W., Furman, A., and Zerihum, D., 2009. Subsurface water
distribution from furrows described by moments analyses. J. of Jrrig. and Drain.
Eng., 135,7-12.
Mallants, D., Mohanty, B. P., Jacques, D., and Feyen, 1., 1996. Spatial Variability of
Hydraulic Properties in A Multi-Layered Soil Profile. Soil Science, T. A., 161: 167-
l81.
Marion, J.M., Rolston, D. E., Kavvas M. L., and Biggar, J.W., 1994. Evaluation of methods
for determining soil-water retentivity and unsaturated hydraulic conductivity. Soil
Sci., 58: 1-13.
Mathews, C. J., Cook, F. J., Knight, J. H., and Braddock, R. D., 2004. Handling the water
content discontinuity at the interface between layered soils within a numerical
scheme. Super Soil 2004: 3rd Australian New Zealand Soils Conference, 5 -9
December 2004, University of Sydney, Australia.
Mathews, C. J., Knight, F. J., and Braddock, R. D., 2004. Using analytic solutions for
homogeneous soils to assess numerical solutions for layered soils. ARC Discovery
Indigenous Research Development Program.C.mathews@griffith.edu.au , 31/4/2011,
17.00 pm, (LT).
Millington, R. J., and Quirk, J. P., 1961. Permeability of porous solids. Trans. Faraday Soc.,
57: 1200-1206.
Mualem, Y., 1976. A new model for predicting the hydraulic conductivity of unsaturated
porous media. Water Resour. Res., 12:513-522.
Nhlabatsi, N. N., 2010. Soil water evaporation studies on the GlenlBoheim ecotope. Ph D.
Thesis, University of the Free State, Bloemfontein, South Africa.
Pachepsky, Y. A., Timlin, D. J., and Ahuja, L. R., 1996. Estimating Saturated Soil Hydraulic
Conductivity Using Water Retention Data and Neural Networks. Soil Science, 164:
552-560.
Russo, D., 1988. Determining soil hydraulic properties by parameter estimation: On the
selection of a model for the hydraulic properties. Water Resour. Res., 24:453-459.
Russo, D., Bresler, E., Shani, U., and Parker, J.C., 1991. Analysis of infiltration events in
relation to determining soil hydraulic properties by inverse problem methodology.
Water Resour. Res., 27:1361-1373.
Schaap, M.G., and van Genuchten, M.T., 2005. A modified Mualem-van Genuchten
formulation for improved description of the hydraulic conductivity near saturation.
Resour. Res., 41:1-12.
101
Schaap, M.G., Leij, F.J., and van Genuchten, M.T., 2001. Rosetta: A computer program for
estimating soil hydraulic parameters with hierarchical pedotransfer functions. J.
Hydro!.,25:1163-176,
Scatter, D.R., Clothier, B.E., and Harper, E.R., 1982. Measuring saturated hydraulic
conductivity and sorptivity using twin rings. Australian Journal of Soil Research, 20:
295-304.
Simunek, J., van Genuchten, M.Th., and Sejna M., 2008. Development and applications of
the HYDRUS and STANMOD software packages, and related codes. Vadase Zone J.,
7: 587-600.
Simunek, J., and Hopmans, J.W., 2009. Modelling compensated root water and nutrient
uptake. Ecological modelling, 220: 505-521.
Sharma, P., Shukla, M. K., Mexal, J. G., 2011. Spatial variability of soil properties in
Agricultural Fields of Southern New Mexico. Soil Science, 176: 288-302.
Soil Classification Working Group, 1991. Soil Classification - A taxonomic system for South
Africa. Memoirs on the Agricultural Natural Resources of South Africa No. IS.
Department of Agricultural Development, Pretoria.
Tomasella, J., and Hodnett, M. G., 1997. Estimating Unsaturated Hydraulic Conductivity of
Brazilian Soils Using Soil-Water Retention Data. Soil Science, 162: 703-712.
Tuli, A, Denton, M.A, Hopmans, J. W., Harter, T., and Maclntyre, J. L., 2001. Multi-step
outflow experiments: from soil preparation to parameter estimation. LAWR Report
No.100037.
Valiantzas, J. D., and Kerkides, D. G., 1990. A simple iterative method for the simultaneous
determination of soil hydaulic properties from one-step outflow data. Water Resour.
Res., 26:143-152.
van Genuchten, M.Th., 1980. A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils. Soil Sci. Soc. Am. J., 44: 892-898.
van Genuchten, M. T., Leij, F. J., and Yates, S. R., 1991. The RETC code for quantifying the
hydraulic functions of unsaturated soils. Rep. EPA/600/2-911065. R.S. Kerr
Environmental Research Laboratory, USEPA, Ada, OK.
Vogel, T., and Cislerova, M., 1988. On the reliability of unsaturated hydraulic conductivity
calculated from the moisture retention curve, Transport in Porous Media, 3: 1-15.
Wang, Z., Wu, L., Harter, T., Lu, J. and Jury, W.A, 2003. A field study of unstable
preferential flow during soil water redistribution. Water Resour. Res., 39: 1075-1086.
102
Wilson, G.W., Fredlund, D. G., and Barbour, S. L., 1997. The effect of soil suction on
evaporative fluxes from soil surfaces. Can. Geotech. 1., 34: 145-155.
Wildnschild, D., Hopmans, J. W., and Simunek, J., 2001. Flow rate dependence of soil
hydraulic characteristics. Soil Sci. Soc. Am. 1., 65: 35-48.
Willmotti, CJ., Ackleson, S. G., Davis, R. E., Feddema, J. J., Klink, K. M., Legates, D. R.,
0' Dennel, J., and Rowe, CM., 1985. Statistics for the Evaluation and Comparison of
Models. Jour. of Geophysical Research, 90: 8995-9005.
Wollschlager, U., Pfaff, T., and Roth, K., 2009. Field-scale apparent hydraulic
parameterisation obtained from TDR time series and inverse modelling. Hydrol. Earth
Syst. Sci., 13: 1953-1966.
World Reference Base for Soil Resources, 1998. World soil resources report, 84, ISSS /
ISRIC FAO, Rome, Italy.
Yeh, T. J., Yeh, M., and Khaleel, R., 2005. Estimation of effective unsaturated hydraulic
conductivity tensor using spatial moments of observed moisture plume. Water Res.
Res.,41: 1-12.
Zachmann, D. W., Duchateau, P. C and Klute, A., 1982. Simultaneous approximation of
water capacityand soil hydraulic conductivity by parameter identification. Soil Sci.,
134: 157-163.
Zavattaro, L., and Grignani, C., 2001. Deriving hydrological parameters for modelling water
flow under field conditions. Soil Sci. Soc. Am. 1., 65:655-667.
103
CHAPTERS
IN SITU EVALUATION OF EVAPORATION IN LAYERED SOILS (TUKULU,
SEPANE AND SWARTLAND)
Abstract
Knowledge about the influence of soil layers on evaporation is essential for the optimization
of in-field rainwater harvesting (IRWH) in the semi-arid areas of the Free State province of
South Africa. Among the soils earmarked for IRWH development includes the Tukulu,
Sepane and Swartland with the first two referred as Cutanic Luvisols and the latter as Cutanic
Cambisols in other countries. In IRWH these soils are required to capture and store rainwater
within the soil profile layers away from the evaporation zone. To determine how the three
soils release and deliver soil water at the evaporating site, a 21 day evaporation experiment
was conducted on pre-drained monoliths. Instantaneous soil water content (SWC) from in
situ and soil water characteristic curve (SWCC) from laboratory was measured. Separate soil
samples of 15 mm thickness were also evaporated under the same conditions to establish the
extent of drying and hydraulic gradient at the soil surface. The Darcian evaporative flux and
unsaturated hydraulic conductivity (K-coefficient) were also determined. At the surface
suctions of magnitude greater than -1500 kPa were observed from all monoliths. Total
contribution to evaporation from the Tukulu, Sepane and Swartland was 43, 51 and 70 mm,
respectively. The low contributions were explained by the presence of the prismacutanic C-
horizon in the Tukulu and Sepane at respective depths of 600 and 700 mm. This layer was
associated with the steepest suction gradient that restrained further upward fluxes by
subsequent lowering for the K-coefficient with more than two orders of magnitudes within a
narrow range of SWc. However, the presence of the pedocutanic B-horizon at depths of 300
mm undermined this restrictive function through the appreciable capillary activity
demonstrated by clays at near evaporating surfaces. The shallowness and deficiency in
structure of the Swartland was consistent with the high contribution to evaporation that gave
this soil a dry soil water regime. It was therefore concluded that the Tukulu offered soil
profile layers that could reasonably satisfy the soil water conservation requirements for
IRWH.
Key words: soil water, soil layers, evaporation, soil hydraulic characteristics
104
5.1 Introduction
Field water storage from infield rainwater harvesting (IRWH) cannot be optimized without
evaluating the influence of soil layers on evaporation. In the absence of a crop soil, water
captured by IRWH over the growing season is likely to decrease by either deep drainage or
evaporation. The presence of clay rich layer at the bottom of most soils earmarked for IRWH
restricts deep drainage losses (Hensley et al., 2000, Botha, 2006; Chimungu, 2009). However,
information concerning the functional role of soil profile layers on evaporation is yet to be
demonstrated. Losses by evaporation could exceed 50 percent of the soil water resources
particularly in semi-arid areas. For example, layered soils including the Tukulu, Sepane and
Swartland of the central Free State province of South Africa are subject to an annual
evaporation potential of 2198 mm (Botha et al., 2007) and yet seasonal rainfall rarely exceeds
550 mm. Under these conditions the fate of the soil water balance is mainly determined by
the hydraulic responses of soil profiles against the high evaporative demand (at the soil
surface).
A number of variables determine the evaporation process from bare soil surfaces. Direct solar
radiation and ambient air temperature provides the energy responsible for transformation of
soil water to vapour state. About 590 calories of heat energy is required for every gram of
water evaporated at the soil surface in a vapour state at 15°C (Hillel, 2004). The vapour could
be carried away by windy conditions from the evaporating site and be replaced by drier air. In
this situation a steep vapour pressure gradient could be maintained. To sustain the
evaporative stream, water from neighbouring subsurface layers needs to be released and
conducted to the evaporating site. In the absence of a water table contribution this capillary
flux is dependent on soil water stored by the oil profile layers. Given that soil water is not
freely available its release and delivery to the surface is determined by the permeability of the
soil profile layers (Hillel, 2004; Wilson et al., 1997; Peters and Durner, 2008).
In drier climates evaporation is mainly characterised by the falling stage. Evaporation
declines once actual evaporation falls below evaporation potential since the soil becomes too
dry to deliver at the potential rate (Simunek and Hopmans, 2009). Upon de-saturation the soil
surface begins to rely on the delivery of water from underlying layers, enhancing the
progression of the drying front (Zeng et al., 2009).The lag time and the hysteric effect of soil
water release and permeability of subsurface layers towards the advancing drying front
triggers the subsequent drop in evaporation (Simunek and Hopmans, 2009). Wilson et al.
lOS
(1997) approximated that evaporation falls to constant levels once matrix suction at the soil
surface falls to around -1500 kPa. Soil surface crusting in semi-arid and desert soils could
limit evaporation to only the few millimetres of the soil profile (Hensley et al., 2000;
Sullivan, 2002; Hillel, 2004) Soil water content within the upper 100 mm of a dried sandy
loam was as low as 0.09 mm mm' with a matric suction of -1000 kPa (Leconte and Brissette,
2001). Total suctions including matric and osmotic suctions exceeded -3000 kPa at the
surface of various textured soils (Wilson et al. 1997). Despite drier conditions at the surface a
considerable amount of soil water remains stored within the profile (Sullivan, 2002).
Literature acknowledges the ability of clay soils to withstand the effect of temporal drying
periods. Increasing matric suction attraction upon desorption is one way that clay soils restrict
capillary water from being available for evaporation (Wilson et al., 1997; Wildenschild et al.,
2001; Jones et al., 2009). Restricting the propagation of the drying front is another. Some
clay shrinks upon drying, a phenomenon that increases the bulk density of the matrix and thus
retaining much if its water (Whitmore and Whalley, 2009). In either ways the unsaturated
hydraulic conductivity (K-coefficient) tends to drop by orders of several magnitudes at
relatively high soil water content (SWC) (Hillel, 2004; Bohne, 2006). In a similar fashion
clay soils restrict available water for extraction by crop plants roots (Whitmore and Whalley,
2009). However, opening of cracks on swelling clays may advance the drying to deeper soil
profile layers (Sullivan, 2002; Grismer, 1992). In many instances the amounts available for
evaporation has been modified by means of soil cover and mulching materials (Botha, 2006;
Sullivan, 2002; Arlauskien et al., 2010).
Analysis of evaporation from layered soils earmarked for the development of IRWH is
limited to only a few studies. Most these studies (Hensley et al., 2000; Botha, 2006;
Nhlabatsi, 2010) focused on the climatic and soil surface factors influencing soil water
evaporation. Under semi-arid conditions soil water evaporation occurs from unsaturated soils
and available water is only controlled by the hydraulic properties of the soil profile layers.
The aim of this paper was to test whether soil water evaporation differed for Tukulu, Sepane
and Swartland soil profile layers. The South African Tukulu, Sepane and Swartland soil types
were studied. The first two soil forms are similar to Cutanic Luvisols and the latter to Cutanic
Cambisols of the World Reference Base for Soil Resources (1998). In this regard the specific
objectives included; Firstly, to describe the amount of soil water available to evaporation by
the soil water release curve; Secondly, to describe the delivery rate of soil water to the
evaporating site by the unsaturated hydraulic conductivity function of soil profile layers; and
106
thirdly, to describe the evaporation flux and accumulative evaporation of the three soils with
respect to their hydraulic characteristics.
5.2. Material and methods
5.2.1 Site location and soil classification
The field experiments were carried out at the Paradys Experimental Farm 29° 13'24,73"S,
26° 12'40,93"E, altitude 1417 m) of the University of the Free State. Three sites were selected
with different soil types including a typical Swartland, Sepane, and Tukulu soil form
according to the Soil Classification Working Group (Soil Classification Working Group,
1991). The total surface land area for the three soil types was about 10 ha on a straight and
gentle slope of 1% with a Southerly aspect.
The three soils share a generic orthic A- horizon of a fine sandy loam with an average depth
of 300 mm for the Tukulu and Sepane and 200 mm for the Swartland (Appendix A,B and C).
The Tukulu has a neocutanic B-horizon with 26% clay content of sub angular structure that
forms a smooth transition with the apedal massive A-horizon with Il % clay content. The
prismatic C-horizon occurs at depths of 600 to 850 mm with a clay content of 48%. The
Sepane has a well developed pedocutanic B-horizon with a depth of 400 mm and a clay
content of 35%, and forms an abrupt transition with the surface horizon with clay content of
19%. Below the latter is the prismatic C horizon with a clay content of 45%. The Swartland
constitutes a course and sub-angular blocky pedocunatic B-horizon with 22% clay content
that forms a smooth transition with A- horizon with Il % clay content. The saprolite rock
appears at a depth of 400 mm and is weakly weathered and has a conspicuous course
appearance. The Tukulu and Swartland occupied the upland slope with the former occupying
the smooth drainage lines while the Sepane occupied the down slope areas of the
experimental site.
5.2.2 Soil water characteristic curve
The plot of SWC versus matric soil is referred as the soil water characteristic curve (SWCC)
(Fredlund, 2006) or release curve (Whitmore and Whalley, 2009). The curve provides a
measure of soil water available for desorption or the amount of water to be released at a given
matric suction i.e. the 9-h relationship. The SWCC was measured from a laboratory
desorption experiment involving three phases from 0 to -10 kPa, -10 to -100 kPa and -100 to
-1500 kPa. Corresponding to these phases are the regions of saturation that characterise the
107
SWCC known as the boundary effect, transitional and residual zones (Fredlund, 2006). The
9-h relationship for the first and second phases were determined from undisturbed core soil
samples using the hanging soilwater column (Dirksen, 1999) and pressure plate chambers,
respectively. Samples for lower matric potentials were disturbed and packed in triplicate at
the measured bulk densities using 2000 mm:' PVC tubes. Disturbance was carried out because
desorption at higher suction is not dependent on connective structural pores, but rather on
textural pores (Hillel, 2004). Soil samples were weighted after each and every step of
desorption until the retention characteristic was completed.
5.2.3 Determination of actual evaporation
The evaporation experiment for Sepane and Tukulu was carried out on 4 m x 4 m monoliths
with a depth of 1 m. Monoliths were prepared in triplicate attached to each other and three
neutron access tubes centrally installed to a depth 1.1 m. On the Swartland it was more
convenient to prepare 1.2 m x 1.2 m monoliths of 0.5 m depth, detached and 30 m apart given
the challenges encountered in digging the trenches. Special DFM probe loggers were inserted
to a depth of 0.6 m with soil water sensors at corresponding depths of 100, 300 and 550 mm.
Lateral sides of monolith trenches were isolated with plastic sheets to prevent lateral water
movement. An earth bund around the monoliths was prepared to prevent surface runoff and
support ponding during the soaking of the profile with water. After deep wetting the profile
surfaces were sealed from evaporation influences with a polythene sheet. The soil profiles
were allowed to drain and SWC measured with a Neutron water meter until the change in soil
water storage was negligible a stage referred as the drainage upper limit (DUL). Once DUL
was realised it was assumed that all macro-pores sensitive to gravitational head were
desorbed and the remaining soil water represented capillary and adsorbed water content
(Hensley et al., 2000; Nhlabatsi, 2010). Monoliths were then opened to the atmosphere to
encourage further drying of the soil profile through the evaporation process. Instantaneous
soil water measurement were made using a neutron water probe (NWP) at 200 mm, 500 mm,
850 mm and 1100 mm depth intervals for a period of 21 days. The first week measurements
were taken in mornings and evenings and then daily for the rest of the trial. The DFM probes
were set to take readings every hour. Drying properties of the soil surface from the three soils
was determined using an evaporation test on a thin soil surface section of about 15 mm
thickness set under the same in-situ environment. Triplicate samples were weighed daily to
determine gravimetric SWc. Measurements were stopped once the thin layer had dried to
almost a constant gravimetric soil water measurement.
108
5.2.4 Data processing and analysis
5.2.4.1 Soil water characteristic curve
The plotted SWCC was subdivided into the three degrees of saturation with more emphasis
on the mid-section that denotes the transitional zone. This section was able to describe the
soil water release during the evaporation process. A regression function was used to transfer
the ê-h relationships from the SWCC to in situ SWC measurements.
5.2.4.2 Estimation of evaporative flux
The upward flux per unit depth (Z) towards evaporating surface was depicted by the
mathematical expression:
(5.1)
Where qevis the evaporative flux (mm hour"), 8 is volumetric soil water content (mm mm"),
t is the daylight time (12 hours) and Z is the reference depth (mm) of the soil layer subject to
evaporation.
5.2.4.3 Estimation of unsaturated hydraulic conductivity
The unsaturated hydraulic conductivity (K-coefficient) was calculated using the
instantaneous profile procedure as described by (Hillel, 2004).
The flux expression (Eq. 5.2) was then fitted to the extended Darcian law to give K-
coefficient as function of SWC (8) for each respective soil profile horizon to be:
K(e)(Z, t) = (5.2)
Where K (8) is K-coefficient (mm hour"), dh is the change in matrix suction (mm) between
the evaporating surface and the soil layer in question of thickness Z (mm). Since flow was in
upward direction the negative sign was used, and t as defined in equation 2. Evaporative flux
(qev) was described by a positive hydraulic gradient and when negative or meaningless
gradients were observed on particular sections of the flow domain the unity gradient was
employed as described by Yeh et al. (1985). The classical exponential equation proposed by
Hillel et al. (1972) was used to obtain a linear regression function on a semi-log scale for the
plotted K-8 relationships.
109
5.2.4.4 Statistical analysis
Soil water characteristic curves and K-coefficients pertinent to the evaporation process after
DUL was realised were described using regression functions. The quality of fit of the
regression was estimated by the coefficient of determination (R\ Homogeneity between
horizon regression coefficients was evaluated using the Bartlett' s test or chi-square (X2)
distribution and pooled regression coefficient F tests as described by Gomez and Gomez
(1984). A Student's Hest was also used to identify horizons with comparable regression
coefficients. Regression coefficients were regarded to have failed the homogeneity test if
computed values were greater than tabular values at 95 % confidence interval.
5.3 Results
5.3.1 Soil water characteristic curves
Figure 5.1 shows the variations of the B-h relationship on the SWCC from the A, Band C-
horizons of the Tukulu, Sepane and Swartland soil types, respectively. The subdivisions of 0
to -3 kPa, -3 to -100 kPa and -100 to -1500 kPa were more applicable in describing the
SWCC of these soils. Soil water activities that are of importance to the soil water balance of
cultivated fields normally occur within the 0 to -100 kPa suction range. The -100 to -1500
kPa is referred as the residual zone because of the SWC in this range being unproductively
low for most economic crops. Consistent with the -3 to -100 kPa matric suction range results
coincided with the de-saturation during the evaporation experiment. The relationship
established the in-situ suction boundary conditions and Table 5.1 outlines the resulting O-h
regression functions and coefficient of determination for each of the three soil horizons.
Table 5.2 presents the corresponding Student t-test analysis for the differences between soil
horizons regression coefficients.
Water available for desorption between 0 and 100 kPa varied among soils and horizons. In
the Tukulu soil the A and B-horizons had a change in SWC ranging from 0.328 to 0.16 and
0.325 to 0.17 mm mm', respectively. The C-horizon desorbed from 0.324 to 0.29 mm mm-I.
Under in situ conditions the amounts available for desorption corresponded to 50, 47 and 8.5
mm for the A, Band C-horizons respectively. In the Sepane desorption effected a release
from 0.33 to 0.17, 0.335 to 0.22 and 0.338 to 0.24 mm mm-I for the respective A-, B- and C-
horizons.
110
(a)
,--.. 0.40
E
E 0.35
E 0.30
E
C 0.25
E
c 0.20
0 -e- A-horizon
,u_
Q.) 0.15 -+- B-horizon
~
~ 0.10 - .&. C-horizon
'0
o: 0.05
0.00
0.1 10 100 1000
(b) Matric suction (-kPa)
,--.. 0.40
E 0.35
E
E 0.30
E
C 0.25 -e- A-horizon
..Q....).
t: 0.20 - -t- B-horizon
0
,u_ 0.15 - .&. C-horizon
Q.)
~
~ 0.10
'0
\/) 0.05
0.00
0.1 10 100 1000
(c) Matric Suction (-kPa)
0.40 -:. ......
E --:::... .........0.35 "'lt......E .....
E 0.30
E I' :"-,,,
'.-...'.". 0.25 -e- A-horizon I '\'i-, ..
c
.Q.....).
t: 0.20 -+- B-horizon I 4
0 II
u,_ 0.15 -+-+
Q.) - 'l!r' C-horizon I '.. ""'6 .. -----+.r...o.. II
~ 0.10 I
.~ I
At:, -A IA- . _ .
'0 I -'-'-'-'-60.05 I
\/) I
I
I
0.00
0.1 10 100 1000
Matric suction (-kPa)
Figure 5.1 B-h relationship curves for the (a) Tukulu, (b) Sepane and (c) Swartland soil
forms.
111
The amount of water released under in situ conditions corresponded to 48, 46 and 20 mm for
the A, Band C-horizons, respectively. The Swartland released water from 0.35 to 0.17, 0.39
to 0.18 and 0.42 to 0.087 mm mmTrom the respective A, Band C-horizons. These amounts
corresponded to 36, 42, and 100 mm of soil water available for desorption under in situ
conditions.
The e-h relationships for the -3 to -100 kPa matric suction range was regressed using a power
function that yielded a coefficient of determination that was good for all the soil horizons (>
0.98). Student's t-test captured that the regression coefficients from the A and B-horizons
were significantly different from the Tukulu soil C-horizon while all horizons from the
Sepane and Swartland were not significantly different.
Table 5.1 e-h regression functions of soils horizons for the 10 -100 kPa matric suction range.
Soil type Horizons Regression R2
Tukulu A e = 0.466 h-o.23 0.99
B e = 0.463 h-o.22 0.99
C e = 0.317 h-o.o2 0.99
Sepane A e = 0.388 h-O.18 0.99
B e = 0.379 h-O.12 0.99
C e = 0.369 h-o.o9 0.99
Swartland A e = 0.456 h-O.21 0.99
B e = 0.364 h-o.IS 0.99
C e = 0.419 h-o.36 0.98
Table 5.2 Student's t-test for differences between soil horizons regression coefficients.
Soil type Horizons Pooled variance Tabulated Variance Ta;O.OS
Tukulu AB 0.13ns 2.26
BC 6.13 2.26
AC 6.13 2.26
Sepane AB 1.25ns 2.26
BC 0.96ns 2.26
AC 2.17ns 2.26
Swartland AB 1.40ns 2.26
BC 1.60ns 2.26
AC 0.47ns 2.26
ns= not significant at 0.05 Student t-test
112
(a) (b) (c)
Soil water content (mm mrrr') Soil water content (mm mrn') Soil water content (mm rnrn')
0.15 0.25 0.35 o 0.15 0.25 t' ,0.3,5 0.05 0.15 0.25 0.350.0 05 S I I , ,o ,5 , , " "'"~)(.G'!,'' " 9. O " "\:')(,' Ei' ''\, -.,.•.. OI"""'~G":t4.S. '" Q"I.. " ", ..... . ~':', '\"\ -,o \ \ °.... " '"0-. , '.,',.,'"-, 0,,<'\ \ '. "" .'. " ,
lOO ~"\
100 ~
,' \ .,.', \ .... 100 I ,\.~,\\'.~ . ' >t.~" i3 t I,'.,\\ ': \.;._ ."1 '\
~ m I.I \
'Vn '1 "". I ,,200 \ . I. I ,, 200 .·" ','\.1\ ~ 1 I ,
200 \11\ ~ ~Ohrs . I : 1II ''1 :
~Ohrs
12 hrs '~'h: \
\
: ....B....12 hrs 1 \ :
1'., 'I
....B.... ~ ~~,I :Ê ,, :. Ê I.I :300 --i::r-- 36 hrs § jj;: Ig 300 --i::r-- 36 hrs :: \ " 300j E ,• I
..c , : 72 hrs II .'.-c'-" - -€)- " . . I
0.. - -€)- 72 hrs v., :,t, : ..c 0..YII (l) I.·' , ,I(l)Cl - .::IE-. 120 hrs j 0.. 400 - .::IE-. 120 hrs(l) '1, : \ Cl 400: Cl .' I I ,
400 -+-240 hrs
,, :, : -+- 240 hrs ·X~I.' I·., ,, ,:'...:: :
-x· -264 hrs Jill. -x· -264 hrs ~500 .&.!.~
'~~,, :,:.: ~
500 ~: :, I '' , : ..:500 I ,':,,
~~',:\~~ 1:, :: ~Ohrs600 , : 600, :: ....B....12 hrs~:
600 :,.\,:~~ ~
, :: --i::r-- 36 hrs
~ ~: - -€)- 72 hrs700
~. .II: 700I~ I - .::IE-. 120 hrs\~ .1'", , -+-240 hrs700 ·1 "U:
~ h' -x· -264 hrs800 800
Figure 5.2 Change in soil water content during the evaporation period for the (a) Tukulu, (b) Sepane and (c) Swartland soils.
113
5.3.2 Change in profile water content during evaporation
Figure 5.2 shows the variation in soil water content from the near surface to deeper profile
layers during the 264 hour evaporation period. Allowing the three soils to drain to DUL
appeared to have given the profiles a more uniform soil water gradient (SWG) prior to the
evaporation experiment. However, this changed as the SWG showed a decline with depth as
available water for evaporation became less available.
Inter-block initial SWC from the Tukulu profile was about 0.30 mm mm-I from the surface to
a depth 450 mm while at 725 mm it was 0.31 mm mm-I. Within 36 hours of evaporation the
SWC at the surface was about 0.17 mm mm-I while at the bottom layer changes were
insignificant. At the end of the 264 hours evaporation period SWC had changed from 0.30 to
0.08 mm mm-I at the surface, 0.30 to 0.21 mm mm-lat 150 mm depth, 0.30 to 0.26 mm mm-I
at 450 mm and 0.31 to 0.29 mm mm-I at 725 mm.
Soil water content from the Sepane profile had stabilised at 0.29 mm mm' from the surface to
150 mm depth and about 0.32 mm mmTrom 500 and 800 mm depths prior to evaporation.
Within 36 hours of evaporation the SWC at the surface was 0.17 mm mm-I while consecutive
depths recorded 0.26, 0.29 and 0.32 mm mm-I. By the end of the 264 hours the SWC had
changed from 0.29 to 0.07 mm mm' at the soil surface while at depths of 150, 500 and 800
mm recorded changes from 0.29 to 0.19, 0.32 to 0.28 and 0.32 to 0.30 mm mm",
respectively.
The Swartland profile exhibited greater uniformity within its upper horizons with initial SWC
recording 0.29 from the surface to 150 mm depth. At 300 and 450 mm depth a respective
0.30 and 0.22 mm mm" was recorded. After 36 hours of evaporation the corresponding
amounts with respect to depth were 0.17, 0.22, 0.25 and 0.22 mm mm-I. At the end of the
evaporation period SWC had changed from 0.29 to 0.07 at the soil surface, 0.29 to 0.18 at
150 mm depth, 0.30 to 0.19 at 300 mm depth and 0.22 to 0.20 at 450 mm depth.
5.3.3 K-O relationships of soil horizons
Figure 5.3 shows the variation in K-8 relationship from the time the three soil profiles
reached DUL and were opened to evaporate until 264 hours. Table 5.3 presents the resulting
K-8 relationship linear regression functions from the A, Band C-horizons of the Tukulu,
Sepane and Swartland soils.
114
(a)
__..,
\..
::l
0 DA-horizon
..s:: 0.1
E +B-horizon
..E_.,
0.01
C 6e-horizon
I
.;; I + I).
·Z
u 0.001
::l
-0
c
0 0.0001 I~If
u
u
§.... 0.00001
-0
;>, I
:r::0.000001
0.13 0.18 0.23 0.28 0.33
(b) Soil water content (mm mm')
__..,
\..
::l
.s0: DA-horizon
0.1 +E
E +B-horizon.._., I
..;..>.. , 0.01 6e-horizon.;;
u +0.001 I
::l
-0
C
0
u 0.0001
.~
c::..e
l
.. lE-OS
-0
;>,
:r:: IE-06
0.13 0.18 0.23 0.28 0.33
(c) Soil water content (mm mm')
\..
o::l DA-horizon
..s::
E 0.1 +B-horizon
..E_.,
.q 0.01 6e-horizon
> ,
u 0.001
::l 6. ~"6"-0 6
Co
u 0.0001 ./~"
.~
e::l lE-OS 6 ,
-£0 I E-06 +-___r____,r___r_____r_-r----,---r-~__r_____.-,...__.._____.-_._____r____,r___r_____r_-.____l
0.13 0.18 0.23 0.28 0.33
Soil water content (mm mm')
Figure 5.3 K-8-relationship from the (a) Tukulu, (b) Sepane and (c) Swartland soil.
115
In all three soils the K-8 relationship from the A-horizon covered a wider range of the SWC
than the deeper Band C-horizons indicating steepening K-8 relationship with depth during
the evaporation process.The corresponding statistical test of homogeneity and Student t-test
for the comparison of regression coefficients between profile layers were shown in Tables 5.4
and 5.5, respectively. The accumulative and flux rate of evaporation shown in Table 5.6 and
Figure 5.4 respectively portrays the evaporation process from the Tukulu, Sepane and
Swartland soil layers.
The Tukulu had K-coefficients that ranged from 0.1 to 0.00004, 0.003 to 0.00002 and 0.004
to 0.00001 mm hour", from the respective A, Band C-horizons. The K-8 relationships were
regressed using a semi log linear exponential function that had a coefficient of determination
that was good for all horizons (>0.83). Failure of the Bartiett's homogeneous test was 1
consistent with pooled variance regression coefficient analysis.
Table 5.3 K-8 relationships linear regression functions for the three soil types.
Soil type Horizon Regression R2
Tukulu A Log K = 46.52(8) - 13.67 0.83
B Log K = 220.80(8) - 62.57 0.97
C Log K = 111.7(8) - 37.01 0.99
Sepane A Log K = 48.41(8) - 13.86 0.97
B Log K = 320.9(8) - 94.43 0.98
C Log K = 111.2(8) - 39.08 0.99
Swartland A Log K = 41.75(8) - 11.89 0.87
B Log K = 35.67(8) - 11.03 0.91
C Log K = 34.14(8) - 9.26 0.94
The Sepane had K-coefficients that ranged from 0.1 to 0.00001, 0.2 to 0.000004 and 0.01 to
O.OOOOOlmmhour", from the respective A, Band C-horizons. The quality of fit of the
corresponding regression function was good for all horizons (>0.97). Significant differences
were found by both Barttlets and pooled variance analysis among horizons. However, the
student's test confirmed similarity in regression coefficients between the Band C-horizons.
The Swartland evaporation process effected a K-coefficient ranging from 0.033 to 0.00002,
0.012 to 0.00001 and 0.008 to 0.00001 mm hour", from the respective A, Band C-horizons.
On the semi-log scale the exponential regression function had a good fit for all horizons (>
0.87). Bartiett's homogeneity test failed to capture significant differences among the horizon
116
regression coefficients while the pooled variance showed significant differences. A clearer
analysis was provided by the Student's t-test that indicated regression coefficients between B
and C horizons were not comparable.
Table 5.4 Homogeneity test for the K-8 relationships of the horizons of the three soils.
BartIett's test Regression coefficient
Soil tY.Qes Com.Qutedx2 Tabular X2; 0.05.2 Com.Quted F Tabular F;UO.05,2:57
Tukulu 6.92 5.99 328.48 3.15
Sepane 11.168 5.99 33.546 3.15
Swartland 0.348ns 5.99 5.367 3.15
Table 5.5 Student's t-test for the regression coefficient comparison between horizons.
Tabulated variance
Soil type Horizons Pooled variance Ta:0.05
Tukulu AB 25.63 2.02
BC 7.03 2.02
AC 15.54 2.02
Sepane AB 5.74 2.02
BC 0.22ns 2.02
AC 7.67 2.02
Swartland AB 2.07 2.02
BC 1.24ns 2.02
AC 3.38 2.02
5.4 Discussion
Given that the soil profiles were already desorbed to DUL at the onset of evaporation, the
drying process represented the emptying of capillary or meso-pores. Soil water release and
availability from this group of pores is controlled by matrix suction activity that is dependent
on soil texture and structure. Under these conditions available water is limited and its release
is dependent on the soil hydraulic characteristics rather than climatic parameters and hence,
the falling stage of evaporation.
117
(a)
10
,-..._ DA-horizon
\....
;:l
0 +B horizon..c 1
E /::,C. -horizon
..E_,
><
;:l
4: 0.1
tl.)
.>;:
~
~ 0.01
>
W
0.001
0 50 100 150 200 250 300
(b) Time (hours)
10
DA-horizon
\....
;:l
..0c 1
+B-horizon
E /::,.Chorizon
..E_,
><
;:l 0.1
4:
tl.)
.>~........
0 0.01
~0.,
>
W
0.001
0 50 100 150 200 250 300
(C) Time (hours)
10
,-..._
\....
;:l
0 DA-horizon..c
E q 0 +B-horizon
E +"'+ /::"C-horizon.._, ~ '--'+----=1'---- D D
><
;:l 0.1 . ~. ~ . . ----+--- -------------- ---- ---4: Is tr·tr· er- --- -- --- --
tl.) t:r .tr- 6.- t:r- 1:.- /::"-8 - I:.
.>
D
~........
&0.01
~
>
W
0.001
0 50 100 150 200 250 300
Time (hours)
Figure 5.4 Evaporative flux from the (a) Tukulu, (b) Sepane and (c) Swartland soil layers.
118
Table 5.6 Accumulative evaporation from the soils horizons and profile.
Accumulative deEth of water (mm) lost to evaEoration at Earticular time (hrs) intervals
Soil t~Ee Horizons 12 36 72 120 240 264 Total %
Tukulu A 9.05 19.61 22.34 24.05 26.15 26.45 61.96
B 3.91 7.81 9.04 9.79 10.78 10.90 25.53
C 0.038 0.19 1.64 3.16 5.09 5.34 12.50
Profile 12.99 27.60 33.01 37.00 42.01 42.68
Sepane A 2.24 8.56 20.76 25.83 30.72 31.31 61.18
B 2.64 10.61 12.82 13.86 14.92 15.06 29.42
C 0.10 1.13 2.86 3.71 4.69 4.81 9.40
Profile 4.98 20.30 36.44 43.40 50.33 51.18
Swartland A 4.35 13.44 16.65 18.74 21.33 21.66 31.08
B 3.16 10.56 17.57 19.46 21.92 22.43 32.18
C 0.11 5.47 9.17 16.95 25.01 25.61 36.75
Profile 7.63 29.47 43.39 55.15 68.26 69.71
*Total % = propotional contribution to total soil profile evaporation by horizons.
The amount of soil water made available during the 264 hour evaporation period varied
among soils and horizons. An amount of 43, 51 and 70 mm was accounted to accumulative
evaporation from the Tukulu, Sepane and Swartland soil profiles. This result reflects the
different influences the soil profile horizon layers had on evaporation. The surface orthic A-
horizon was generic to all three soils and it was reasonable to assume that it had a common
effect on evaporation (Hensley et al., 2000).
Presence of the prismacutanic C-horizon in the Tukulu and Sepane with a clay content of 48
and 45 percent, respectively could explain the low evaporation recorded from these soils. The
C-horizon was found at 600 to 850 mm depth in the Tukulu and at 700 to 900 mm depth in
the Sepane. The Sepane also had a pedocutanic B-horizon with 35 percent clay at depths of
300 to 700 mm in contrast to the Tukulu that had neocutanic B-horizon of fine sandy clay
loam at 300 to 700 mm depth. This contrast in layering within the midsection of the profile
explained the greater capillary activity that resulted in a higher evaporation loss from the
Sepane compared to the Tukulu. Botha et al. (2006) also observed that soils with a clay rich
layer near the soil surface were susceptible to high evaporation losses. However, soil water
retained by restrictive clay layer at depths below 300 mm was hardly affected by temporal
drying spells (Botha et al., 2007; Hillel, 2004; Whitmore and Whalley, 2009).
119
One aspect of the heavy structure of clay rich horizons is the ability to restrict the progression
of drying front. The steep SWG observed from the Tukulu and Sepane supports this notion.
The SWC from the prismacutanic C-horizon of these soils remained unchanged for the first
72 hours of evaporation, pointing to the time equivalent to the propagation of the drying front
at this depth. The drying front took approximately 36 hours to reach the saprolite C-horizon
at depth of 400 mm indicating the structural deficiency of Swartland. Consistent with the
high evaporation was the nearly uniform SWG that stretched to deeper layers of this soil
profile. The different drying regimes resulting from the lag time of the drying front from the
Tukulu and Sepane soil profiles was illustrated by the 60, 30 and 10 percent contributions to
total evaporation from the respective A, Band C-horizons. This skewed distribution of total
evaporation with depth was in agreement with the hypothetical shape of the drying front
illustrated by Hillel (2004).
Although the distribution of SWG with depth was consistent with the amounts contributed by
horizons to evaporation it could not explain the process. Within the first 12 hours of
evaporation SWC from the bare soil surfaces dropped from 0.3 to 0.2 mm mm' from the
Tukulu and 0.29 to 0.19 mm from the Sepane and Swartland. Corresponding matric suction
increased from -6.59 to -36.76, -4.38 to -43.17 and -7.75 to -52.32 kPa from the respective
Tukulu, Sepane and Swartland. By the end of the evaporation period, SWC at the surface was
around 0.07 mm mm" with corresponding matric suctions of -2231.15, -10965.39 and -
5708.84 kPa from the Tukulu, Sepane and Swartland, respectively. The greater suctions from
the Sepane were attributed to the higher clay content of 19 percent in the A-horizon
compared to 11.3 percent from its counterparts. Nevertheless these matric suctions fell in the
same category to those recorded by Wilson et al. (1997). The resulting hydraulic gradient was
in agreement with the upward evaporative flux. Steep gradients were associated with the
prismatic C-horizons and pedocutanic horizons with the steepest recorded for the Sepane.
This was not surprising because of the greater affinity of clay soils for water (Whitmore and
Whalley, 2009).
This phenomenon was also illustrated by the 8-h relationship from the horizons whereby a
within a narrow range of SWC the matric suction increased by more than two orders of
magnitude. Similar observations on clay rich horizons were made by Chimungu (2009) and
Nhlabatsi (2010). Therefore, the steep gradients of the K-8 relationships shown by subsurface
horizons especially from the Tukulu and Sepane were expected. Of interest was that the K-
coefficients at the onset of evaporation were greater than the final values attained during the
120
draining of these profiles (results not shown). This suggests that the orientation of pores
towards drainage and evaporation was accountable to different hydraulic gradients. The
complexity of this phenomenon could be explained by the non isotropic nature of the K-8
relationships especially in layered soils (Hillel, 2004). Beyond 0.001 mm hour", the SWC
irrespective of depth showed very little change from the three soils. However, this coefficient
was attained at different times and flux rates suggesting that the release and delivery rate of
water at the evaporating site varied among the profiles. Noticeable from the three soils was
also the failure of the K-8 relationship from upper horizons to follow a straight line on the
semi-log coordinate system. This was not strange given the interaction of high evaporative
demand of the atmosphere and the soils desorption restrictive responses at the surface. The
diurnal and hysteric nature of the movement of liquid and vapour near the evaporating
surface have been recognised to be sources of non-linearity in this regard (Hillel, 2004; Gupta
et al., 2006). Also the diffusive nature of evaporation near the surface especially on course
textured soils (Hillel, 2004) could be responsible for the lower but still acceptable goodness
of fit from all surface horizons. Given the coarseness of the Swartland horizons it could
therefore be understood why the K-8 relationship exhibited greater non-linearity and a fit that
was lower than its counterparts.
5.5 Conclusions
The effect of soil horizon layers on soil water evaporation from pre-drained Tukulu, Sepane
and Swartland soil profiles was evaluated. The amount and rate of evaporation was found to
differ among the three soil profiles under the same climatic conditions. The Tukulu profile
with a distinct prismatic C-horizon contributed 43 mm to evaporation during the 264 hour
period. The Sepane with dual clay-rich layers in the pedocutanic B and prismatic C-horizon
contributed 51 mm while the Swartland with youthful horizons contributed 70 mm for the
same period. Correspondingly the slopes of the 8-h and K-8 relationships from the clay-rich
horizon layers had a steeper gradient for a narrow change in SWc. This was consistent with
literature about the hydraulic properties of clay soils and based on this finding the Tukulu had
the least contribution to soil water evaporation. In this regard the layering of the Tukulu soil
profile horizons could be regarded to be appropriate for the development of IRWH and with
careful considerations in design, especially those that enhance early ground cover, soil water
evaporation could be further be reduced.
121
Acknowledgements
We acknowledge the technical support rendered by the management and staff of Paradys
experiment station. Special thanks also to Ms. Liesl van der Westhuizen for her constructive
suggestions on improving the writing of this manuscript.
122
5.6 References
Arlauskiene, A., Maiksteniene, S. and Slepetiene, A, 2010. Effect of cover crops and straw
on the humic substances in the clay loam Cambisol. Agronomy Research, 8:397-402.
Bohne, K., 2006. An introduction into applied soil hydrology. Catena Verlag GMBH,
Reiskrechen, Germany.
Botha, J. J., 2006. Evaluation of maize and sunflower production in a semi-arid area usingin-
field rainwater harvesting. Ph.D. (Agric ) Dissertation, University of the Free State,
Bloemfontein, South Africa.
Botha, J. J. Anderson, D. C., Groenewald, N., Mdibe N., Baiphethi, M.N., Nhlabatsi, N. N.
and Zere, T.B., 2007. On-farm application of In-field rainwater harvesting
techniqueson small plots in the central region of South Africa. Water Research
Commission, Report, TT 313/07, Pretoria, South Africa.
Chimungu, J. G., 2009. Comparison of field and laboratory measured hydraulic properties of
selected diagnostic soil horizons. Master's (Agric) Dissertation, University of the Free
State, Bloemfontein, South Africa.
Dirksen, C., 1999. Soil physical measurements. Gee-ecology paperback, Catena verlag
GMBH, 35447, Reiskirchen, Germany.
Fredlund, D.G., 2006. Unsaturated soil mechanics In engmeering practice. Journal
Geotechnical and Geo environmental Engineering, 132: 286-321.
Gomez, K.A and Gomez, AA, 1984, Statistical procedures for agricultural research, 2nd ed.,
A Wiley-Inter-science Publication, New York, USA.
Grismer, M. E., 1992. Cracks in irrigated clay soil may allow some drainage. California
Agriculture, 46: 9-11.
Gupta., S.D. and Mohanty, B. P. and Kohne, J. M., 2006. Soil hydraulic conductivities and
their spatial d temporal variations in a vertisol.Soil Science Society of America
Journal, 70: 1872-1881.
Hensley, M., Botha, J. J., Anderson, J. J., van Staden, P. P. and DuToit, A, 2000.
Optirnisingrainfall use efficiency for developing farmers with limited access to
irrigation water, Water Research Commission report 878/1/00, Pretoria, South Africa.
Hillel, D., 2004. Introduction to environmental soil physics, Academic Press, New York,
USA
123
Hillel, D., Krentos, V. D. and Styllianou, Y., 1972. Procedure and test of an internal drainage
method for measuring soil hydraulic characteristics in situ, Soil Science, 114: 395-
400.
Jones, E.W., Koh, Y. H., Tiver, B. J., Wong, M. A. H., 2009. Modelling the behaviour of
unsaturated, saline clay for geotechnical design, School of Civil Environment and
Mining Engineering, University of Adelaide, Australia.
Leconte, R. and Brissette, F. P., 200]. Soil moisture profile model for a two-layered soil
basedon sharp wetting front approach, ASCE. Journal of Hydrologic Engineering,
6:141-149.
Nhlabatsi, N. N., 2011. Soil water evaporation studies on the GlenlBoheim ecotope, Ph.D.
(Agric) Dissertation, University of the Free State, South Africa.
Peters, A. and Durner, W., 2009. Simplified evaporation method for determining soil
hydraulic properties. Journal of Hydrology, 356: 147-162.
Simunek, J. and Hopmans, J. W., 2009. Modelling compensated root water and nutrient
uptake. Ecological modelling, 220: 505-521.
Soil Classification Working Group, 1991. Soil Classification: A taxonomic system for South
Africa. Memoirs on the Agricultural Natural Resources of South Africa No. 15,
Department of Agricultural Development, Pretoria, South Africa.
Sullivan, P, 2002. Drought resistant soil, Agronomy Technical Note, National Centre for
Appropriate Technology.www.attra.ncat.org.I1/06/2011.
Whitmore, A. P. and Whalley,W. R., 2009.Physical effects of soil drying on roots anderop
Growth.Journal of Experimental Botany, 60: 2845-2857.
Wildenschild, D., Hopmans, J. W. and Simunek, J., 2001. Flow rate dependence of soil
hydraulic characteristics. Soil Science Society of America Journal, 65: 35-48.
Wilson, G. W., Fredlund, D.G. and Barbour, S.L., 1997. The effect of soil suction on
evaporative fluxes from soil surfaces. Canada Geotechnical Journal, 34: 145-155.
World Reference Base for Soil Resources, 1998. World soil resources report, 84, ISSS /
ISRIC FAO, Rome, Italy.
Yeh, T. C. J., Gelhar, L.W. and Gutjahr, A. L., 1985. Stochastic analysis of unsaturated flow
in heterogeneous soils, 2. Statistically anisotropic media with variable xa. Water
Resources Research, 21: 457-464.
Zeng, Y., Su, Z., Wan, L., Yang, Z., Zhang, T., Tian, H., Shi, X., Wang, X. and Cao, W.,
2009. Diurnal pattern of the drying front in desert and its application for determining
the effective infiltration. Hydrology and Earth System Science, 13: 703-714.
124
CHAPTER6
DETERMINING THE OPTIMUM INFLOW RATES FOR MICRO-FLOOD
IRRIGATION ON THE TUKULU SOIL
Abstract
The dependence of micro flood irrigation (MFI) on small inflow rates was evaluated on 90 m
closed ended furrows in the South African Tukulu soil also referred as Cutanic Luvisols in
other countries. A single irrigation was used to characterise the surface and subsurface soil
water distribution from the 20, 40, 80 and 160 L min" inflow rates treatments at every 10 m
distance intervals starting at 5 m from the inlet. Neutron access tubes were installed at these
stations to a depth of 1 m. The HYDRUS-2D software was also used to predict the initial and
final soil water content. Surface water distribution from the 20 L min-) sustained the advance
stream up to 60 m furrow length while from other treatments it reached the end. Agreements
between measured and predicted soil water content were fairly good although inconsistencies
due to spatial variations were to be expected. Effect of horizons layers was also observed
through discontinuities in subsurface water distribution. Distribution uniformity at longer
furrow distances was highest from the 80 and 160 L min-) with the 29 and 40 L min-)
recorded similar performances at shorter distances. Given the practicality and flexibility of
working with small inflow rates, it was suggested that the 40 L min-) inflow rate be adopted
for furrow distances of less than 60 m.
Key words: Furrow irrigation, Inflow rates, Infiltrated depth, HYDRUS-2D, Distribution
uniformity.
125
6.1 Introduction
Dependence of micro flood irrigation (MFI) on small inflow rates has made the achievement
of high application efficiencies comparable to pressured systems a reality in furrow irrigation.
Because furrow irrigation relies on the soil surface to convey and distribute water it has been
associated with poor distribution uniformity and inefficient water losses through deep
drainage and tail end runoff. Given the growing water scarcity and food insecurity in many
developing countries inefficient irrigation practices could no longer be acceptable. Advances
in technological breakthroughs have brought remarkable improvement in furrow irrigation
but very few of these have been sustainable among the rural farming communities where this
method is widely practiced.
In principle furrow irrigation is said to have attained high application efficiencies when the
infiltrated depth from the inlet to the furrow end is almost uniform. For this condition to be
realised the stream flow and opportunity time during irrigation should be about the same
along the furrow length. Practically this is not possible given that the inflow begins to
infiltrate at the inlet and if it not large enough it could go to zero before reaching the furrow
end. Even when the advance stream is sizeable to reach the end differences in the intake
opportunity time become the major source of non-uniformity (Walker, 2003).
Various approaches have been investigated for improving distribution uniformity on furrows.
One approach focuses on reducing the roughness and infiltration rate of the furrow surface.
This is usually achieved by compacting and smoothening of the furrows (Bargar et al., 1999;
Clemmens et al., 2001; Walker, 2003). Other modifications could include grading and
levelling the field to ensure that advance stream flows much faster and uniformly. Increasing
volumes of stream flows is another way of attaining more uniform irrigation especially on
long furrows (Walker, 2003). Difficulties of managing advance stream and runoff in long
furrows that could be as long as 1000 m (Wohling and Maihol, 2007) undermines this
approach. To improve control of large inflow rates and tail end runoff automatic cutback and
cyclical or surge flows are recommended (Walker, 2003 Rodriguez et al., 2004). Another
approach is to increase intake opportunity time at the lower quarter of the furrow by
encouraging ponding through blocking the furrow end (Walker, 2003).
Transfer of some of these technologies to rural farming communities that have access to
irrigation water has been a challenge. Although lack of infrastructural development and
access to capital are the common reasons (Awulachew et al., 2005), the unavailability of
126
alternative furrow irrigation methods is the major setback. Of critical importance is that most
of the rural communities' practices farming at a small scale and widely distributed on
marginal soils that are susceptible to erratic soil water regimes and fertility degradation.
Therefore, access to alternative furrow irrigation technologies that are user friendly and
preserve the resource base could not be overemphasised. Having considered these
contributing factors to food insecurity Austin (2003) proposed the (MFI) system. This system
combines the benefits of soils water intake and subsurface redistribution to optirnise irrigation
efficiencies using inflow rates as small as 20 L min-I. Distribution efficiencies as high as 95%
were recorded along unconventional furrows of 20 m long prepared on sandy loam soils
(Austin, 2003).
In the Free State province of South Africa, a similar situation of unavailability of alternative
irrigation methods is encountered by rural farming communities. These communities are
densely populated on the low productive areas with the majority of the soils belonging to
Duplex group (Hensley et aI., 2007). Common to this group is the occurrence of a clay rich
layer either in the B or C-horizon underlying a weakly developed course textured A- horizon.
Sensitivity of these soils to dry soil water regimes and runoff-erosion degradation is well
documented (Hensley et aI., 2000; Botha, 2006). Therefore, crop failure during prolonged dry
spells is common phenomenon among these communities.
Could the higher efficiencies achieved by the MFI be reproduced on any of the layered soils
cultivated in the province? In search for answers to this question the Tukulu soil also similar
to the Cutanic Luvisols that occupies 500 to 600 million hectares worldwide (World
Reference Base for Soil Resources, 1998), was selected in pursuit of the following objectives;
Firstly, to characterise the surface water distribution during the advance-supply phase.
Secondly, to describe the subsurface water distribution during irrigation with the aid of the
HYDRUS-2D software and thirdly: to evaluate the efficiencies from the irrigation outcomes.
6.2 Material and methods
6.2.1 Site location
The field experiments were carried out at Paradys Experimental Farm (29° 13'24, 73"S,
26°12'40, 93"E, altitude 1417 m) of the University of the Free State, South Africa. The
selected site was of a typical Tukulu (Tu) soil form according to the Soil Classification
Working Group (1991). This soil type represents a group of layered soils that covers 12%
127
land area of the province landscape. The A- and B-horizons are of orthic and neocutanic
structure, respectively with the underlying C-horizon with a moderate to strong prismacutanic
tructure. Physical properties of the Tukulu soil were summarized in Table 6.1 accompanied
with a soil profile photo Figure 6.1.
Table 6.1 Summary of the soil physical of the Tukulu soil form.
Physical properties Master horizon
A Bl C
Coarse sand (%) 5.3 9.2 2.1
Medium sand (%) 9.3 8.8 3.8
Fine sand (%) 41.2 31 28.3
Very fine sand (%) 25.3 21 8.4
Coarse silt (%) 2.] 2 3
Fine silt (%) 4.6 2.5 6.5
Clay (%) 11.3 26.4 47.9
Bulk density (kg m·3) 1670 1597 1602
Porosity (%) 34.0 33 32.4
(Ks, mm hour -I) 36.1 40 9.6
Ks= saturated hydraulic conductivity
(a)
(b)
motties
Figure 6.1 (a) Profile of the Tukulu and (b) moules present in the C-horizon.
128
6.2.2 Experimental design and measurements
Four plots with three furrow replications were prepared which were 90 m long and had 2 m
row spacing. Four inflow rate treatments including 20 L min-I, 40 L min-I, 80 L min-I and
160 L min-I were systematically allocated to the four plots with the largest assigned to the
plot nearest to the delivery line and the smallest inflow rate to the last plot. A 40 mm low
pressure delivery pipe was fitted to the distributor from the primary reservoir. Each plot had a
single delivery pipe with a control valve. A rigder was used to open furrows that assumed a
trapezoidal shape with a top width and bottom of 350 mm and 150 mm with a depth of 200
mm. All furrows followed a dead zero slope along the field contour lines and had a blocked
furrow end.
Soil water measurements were taken by a neutron probe soil water meter. A set of 10 neutron
probe access tubes of 1.2 m height were installed along the furrow shoulders of the central
furrow at 10 m distance intervals beginning from the 5 m from the inlet. Readings were taken
at depths of ISO mm, 450 mm and 725 mm corresponding to the mid points of the respective
horizons layers. A single irrigation event was used for this work and measurements were
taken before and after irrigation. Inflow controlling valves were calibrated just before
irrigation and one furrow was irrigated at a time. Above ground measurements focused on
the advance-supply and recession-depletion phases. During the irrigation period the times of
starting and cutting off supply was recorded with the latter initiated when the advancing front
was at 80 m distance. Times of the advancing front were also recorded at every neutron
probe. Depth of stream flow behind the advancing front was also measured at each station
and also during the ponding phase. The resulting arithmetic average was considered as the
stream flow wetting depth at that particular station.
6.2.3 Simulations and inverse analyses
Infiltration during furrow irrigation could be assumed to be two-dimensional in the plane of
the cross section of the furrow (Tabuada et al., 1995). The installation of one neutron probe at
each measurement stations restricted this work to observe vertical soil water distribution. In
addition, continuous measurement of soil water content (SWC) during the irrigation process
was impractical. Therefore, the HYDRUS-2D software code of Simunek et al., (2009) was
employed to simulate the two dimensional infiltration from the different inflow rates. This
model numerically solves the two dimensional form of the Richard flow equation and are
able to minimize the objective function for parameter optimization by using the Levenberg-
129
Marquardt non linear- squares approach (Marquardt, 1963). The HYDRUS-2D code has been
successfully employed in simulating studies of subsurface soil water distribution during
furrow irrigation and redistribution (Abbasi et al., 2003; Wohling and Maihol, 2007;
Lazarovitch et al., 2009).
In this work SWC before and after irrigation with corresponding opportunity times
constituted the objective function. Model predictions were compared with field-measured
SWC and infiltrated depths. The 5 m, 55 m and 85 m furrow distances were selected for
model predictions. Selected distances from the furrow that received the 20 L min-I were 5 m,
35 m and 55 m because the advancing stream flow only reached the 60 m mark. Laboratory
based soil water characteristic curve (SWCC) for all soil horizons was parameterised using
the closed-form equation of van Genuchten (1980):
(6.1)
Where es and er are the respective saturated and residual values of the volumetric water
content, e (mm mm'), h is the matrix suction (mm), while a and n are the shape and pore
size distribution parameters, respectively. The condition m=l-lIn should be satisfied and e
(h) equals es when at zero or positive matric suctions. Parameters related to pore-size
distribution (n) and shape (a) of the SWCC were estimated from the Rosetta pedo-transfer
(Schaap et al., 2001) program included in the HYDRUS-2D hydraulic parameters option. A
summary of the fitted hydraulic parameters were summarized in Table 6.2. To be in
agreement with the rest of the parameters the field based saturated hydraulic conductivity
(Ks, mm hour") derived from the double ring infiltrometer (Scotter et al., 1982) was
optimised during estimation of n and a parameters.
Table 6.2 Soil hydraulic characteristics of the Tukulu soil.
Soil horizons
Parameters A B c
er(mm mm") 0.13 0.12 0.26
es(mm mm") 0.34 0.33 0.32
a 0.00116 0.00094 0.0052
n l.773 1.617 1.226
Ks(mm hour") 36.1 40 9.6 (1.9)
(*) = value from parameter optimisation procedure
130
The van Genuchten (1980) model with air entry value of -0.2 kPa was used to describe soil
hydraulic properties during the infiltration simulation. A soil profile flow domain of 4 000
mm by 850 mm dimensions and a finite element mesh of 50 mm was defined with the furrow
at the ymmetric position (Figure 6.2). The profile was also discretisized into three layer
materials corresponding to the in situ soil horizons. The pressure head was selected for initial
pressure conditions with the assumption that pressure head was uniform in each horizon.
Constant head was appropriate for the furrow boundary condition while along the vertical
sides a no flux condition was assigned. At the bottom of the profile a free drainage boundary
was selected.
3S0mm
1S0mm
8S0mm
2000 mm
Figure 6.2 Fine element mesh and boundary conditions assigned inside the furrow (constant
head), zero flux at the surface and vertical sides and free drainage at the bottom of the profile.
6.2.4 Experimental data analysis
Processing of measured and predicted SWC into infiltrated depth and function was the
measure focus of this experiment. The geometric mean was selected for averaging infiltrated
depth at different points along the furrow length in consideration of soil spatial variability
that could be a source of non linearity (Roman a et al., 1996; Abbasi et al., 2004). Infiltration
functions were assumed to be directly proportional to the actual opportunity time determined
by the advance-recession phase. From the selected three furrow measurement points (5, 55
and 85 m) the infiltrated depth with corresponding opportunity times were averaged to
determine the average furrow infiltration function (AFIF). This function was validated by
averaging the one point time rate change in SWC simulated by HYDRUS-2D. Since one
131
point values were used to estimate the AFIF this approach was considered to be
representative of the one point method described by Clemmens et al. 2001). The AFIF was
also simplified to a simple power function and expressed as (Clemmens et al., 2001);
(6.2)
Where D is infiltrated depth (mm), r is the infiltration opportunity time and k and a are
parameters. Best fit curves were used to estimate the parameters with the objective to
improve the conservation of mass with the final average infiltrated depth. Where measured
and predicted final infiltrated depth varied the parameters were adjusted to agree with final
infiltrated depth at the specified opportunity time. The AFIF was then used to approximate
infiltrated depths 30 and 60 m furrow distances with respect to the corresponding average
opportunity times.
Estimating the irrigation performance of the 20, 40 80 and 80 L min-I inflow rates at different
furrow lengths the distribution uniformity (DU), application efficiency (AE) and deep
percolation ratio (DPR) (Clemmens et al., 2001; Oyonarte et al., 2002). In the absent of tail
end runoff on blocked ended furrows these criterion using average accumulative infiltration
could take the following expressions;
DU = infiltrated depth at the lower quarter of the furrow(mm) (6.3)
furrow infiltrated depth (mm)
AE = infiltrated depth within the root zone (mm) (6.4)
infiltrated depth of the profile
DPR= infiltrated depth below the root zone (mm) (6.5)
Total infiltrated depth (mm)
6.2.3 Statistical analysis
Agreement between the measured and predicted SWC content measurement was evaluated
using the root mean square error (RMSE) and the index of agreement or D-index as proposed
by Willmot et al. (1985). AI: 1 line was further used to assess the quality of fit alongside the
coefficient of determination (R\ The model predictions are considered satisfactory if the
values of RMSE approach zero and both the D-index and R2 approach unity. Subsequent
analysis of secondary data was subjected to the BartIetts or chi-square (X2) homogeneity test
and Student's t-test analysis. The homogeneity hypothesis was accepted if computed values
were equal or less than the reference tabulated value.
132
6.3 Results
6.3.1 Surface flow characteristics
Figure 6.3 shows the time-distance relationships of the advance and recession phase from the
20, 40, 80 and 160 L min-I inflow rates tested on the Tukulu soil 90 m furrow runs. The
advance time recorded from the 20 L min-I was 160 minutes at a 60 m distance before the
advancing stream approached zero. The 40, 80 and 160 L min-I inflow rates reached the
furrow end at different times of 105, 38, and 23 minutes, respectively. Corresponding cut off
times were 160 minutes (2.7 hours), 105 minutes (l.08 hours), 35 minutes (0.6 hours) and 23
minutes (0.4 hours) for the 20, 40, 80 and 160 L min-I inflow rates. Recession phase from the
20 and 40 L min-I inflow rates was observed from the 162 to 166 and 106 to 112 minutes,
respectively. The 80 and 160 L min-I inflow rates had recession period from the 38 to 100 and
28 to 99 minutes from irrigation.
The average depth of the advancing stream flow and opportunity times along the 10 m
interval measuring points were summarized in Table 6.3. The average depth of stream flow
ranged from 15 to 60 mm from the 20 L min-I inflow with an opportunity time range of 2 to
162 minutes. In the 40 and 80 L min-I had a respective stream flow depth ranging from 33 to
80 mm mm and 63 to 80 mm with a corresponding opportunity time range of 7 to 105 and 37
to 65 minutes. The 160 L min-I recorded an average stream flow depth ranging from 70 to
125 minutes with opportunity times from 27 to 76 minutes.
6.3.2 Subsurface soil water content
Figure 6.4 presented the measured and predicted SWC before and after irrigation on a 1:1
line. Optimised parameters for the improvement of model's predictions were summarized in
Table 6.4 Statistical measure of fit between the measured and predicted SWC are also
presented in Table 6.5.
133
180 A
,,-.. 160 - - ~- - - *- - - ->E-- - - ~ - - -*, ~
t:
ï~ 140120
v
E 100
~ 80
60
40
20
180 B
,,-.. 160
t:
ï.._~_, 140
v 120
E
~ 100
___ ~ >+ _ - - ~_ - - - )E - - - -x- - - - l>( - - - -)Eo - - <'~
80
60
40
20
0
180
160 c
,,-..
e 140
ï.._~_, 120
v 100 X __-x
E 80 ..... -- ... -'--
~ 60 --- _- - --x- - -- - - - x I
40 _~_-----~-------x--- x I>
20
0 I
120
D
100
80 ~-------~-,,-..
t: ?:<- - - - --
'E x
..__, 60
v ---
E 40
~ _----x
--- x
20
0
5 15 25 35 45 55 65 75 85
Furrow length (m)
o Advance x Recession
Figure 6.3 Measured advance and recession times for the 20 (A), 40 (B), 80 (C) and 160 (D)
Lrnin".
134
Table 6.3 Stream depth and opportunity time of stream-flow along furrows under different
inflow rates.
Inflow rates (L min-I)
Furrow length 20 40 80 160
SD OT SD OT SD OT SD OT
(mm) (min) (mm) (min) (mm (min) (mm) (min)
5 60 162 80 105 80 37 90 27
15 60 155 77 98 85 34 125 28
25 60 136 67 89 92 31 IlO 34
35 50 112 65 80 83 33 lOS 38
45 50 85 60 71 80 35 100 58
55 20 32 55 55 75 39 90 63
65 15 2 50 31 70 46 85 71
75 38 I 1 65 64 80 71
85 33 7 63 65 70 76
OT;opportunity time in minutes
SD; stream depth
Table 6.4 Optimised parameters to improve model's predictions.
Inflow rates (L min-I)
Distance 20 40 80 160
Sm Horizons er a er a er a er a
A 0.0006 0.13 0.0022 - 0.0028
B 0.0005 0.19 0.0005 - 0.0007
C 0.0457 0.25 0.0159 - 0.0433
55 (35) A 0.13 - 0.0042 0.0070 - 0.0013
B 0.05 - 0.0005 0.0006 - 0.0009
C 0.26 - 0.0175 0.0056 - 0.0015
85(55) A 0.0023 - 0.0143 0.13 - 0.0017
B 0.0005 - 0.0003 0.16 - 0.0008
C 0.0467 - 0.0639 0.26 - 0.0015
() = Furrow distance from the 20 L min-
There was generally good agreement between measured and predicted SWC at various
lengths of the furrow from all inflow rates (Figure 6.4). At depth of 150 mm all furrows had
an initial SWC measurement ranging from 0.13 to 0.145 mm mm" while the predicted ranged
from 0.13 to 0.139 mm mm". Following irrigation the measured SWC ranged from 0.24 to
0.32 mm mm-I while the predicted ranged from 0_27 to 0.34 mm mm-I.
135
20 Lmin" 40 Lrnin' 80 Lmin" 160 Lmin"
~ 0.4 ] 04] 04] 0.4
E
E 0.3 Sm y I 0.3 Sm ~ I 0.3 Sm 3 I J 00.3 Sm 0
E
E
U 0.2 i .> I 0.2 ~ /' I 0.2 i ./ I 0.2
o::::
"0 0.1 i 0.1 i
u0)
-: I 0.1 ~ /' I ./ I 0.1
:0
0..).. 0.0 0 0
0.. 00.0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
0.4 ] 0.4 J 0.4 J 0.4
E 0.3 35 m _!V I 0.3 55 m _lJ/" I 0.3 55 mE - ~ I 0.3
E
-5 0.2 ~ ./ I 0.2 ~ -: I 0.2 i .,,/ I 0.2
U:::
r/l 0.1 ~ ~ I 0.1 i /- I 0.1 i -: I 0.1
"0
u0)
:0 0.0 0 0 0
2 0.0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.40.. 0 0.2 0.4
0.40 ] 0.4 0.4E 0.4 ]
E 85 m 85m 0
0.3 _D~ I 0.30 n/'" I 0.3 i .....,,/.E I 0.3
g
u::: 02 / 0.20 1 .: 0.2 ~ .A') I 0.2r/l 0"0u0) 0.11 0.10 o.pJ _/ I 0.1
:0
0..).. 0
0.. 0 0.00 V I I I I DV I I I I
0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Observed SWC (mm mm -I) Observed SWC (mm mm -I) Observed SWC (mm mm -I) Observed SWC (mm mm -I)
Figure 6.4 Measured and predicted soil water content (SWC) for the profile depth layers (0-300; 300-600, and 600-850 mm) before and after irrigation
HYDRUS-2D software from the 20, 40,80 and 160 L min' inflow rates at three furrow distances.
136
At depth of 450 mm the measured initial and final SWC ranged from 0.20 to 0.27 mm mm-I
and 0.22 to 0.327 mm mm', respectively. Predicted initial SWC ranged from 0.26 to 0.29
mm mm' and final SWC within the range of 0.265 to 0.30 mm mm-I. At depth of 725 mm
the measured initial SWC was from 0.265 to 0.29 and the predicted from 0.23 to 0.30 mm
mm-I. After irrigation the measured and predicted values ranged between 0.27 to 0.32 and
0.29 to 0.33 mm mrn', respectively. The values of R2 was good (> 0.98) for all inflow rates
throughout the furrow lengths. The D-index also indicated a good agreement (> 0.97)
between the measured and predicted values especially at the inlet. Further away from the inlet
the D-index showed a decrease but still within the acceptable range (> 0.90). Consistent with
this result was the RMSE of 0.02 near the inlet and further away the furrow length recording
a value of 0.04 in all the furrows.
Table 6.5 Statistical measure of fit from measured and predicted soil water content at three
furrow distance measuring points.
Inflow rate Distance
(L min") (m) RMSE D-index R2
20 5 0.025 0.966 0.997
35 0.026 0.957 0.998
55 0.035 0.902 0.995
40 5 0.022 0.975 0.998
55 0.039 0.910 0.990
85 0.025 0.944 0.999
80 5 0.020 0.974 0.999
55 0.016 0.984 0.999
85 0.018 0.981 0.998
160 5 0.019 0.973 0.998
55 0.031 0.935 0.994
85 0.039 0.906 0.989
6.3.3 Subsurface soil water distribution
Variation in measured and predicted infiltrated depth along the wetted furrow was shown in
Figure 6.5 with reference to the three selected distance points. At the 5 m distance from the
furrow inlet measured infiltrated depth was 90, 101, and 57 and 59 mm from the 20, 40, 80
and 160 L min-I inflow rates, respectively. In the same order corresponding predicted values
were 92, 85, 68 and 72 mm. The 20 L min-I wetted the furrow up to the 60 m with infiltrated
depth decreasing monotonously with distance recording an infiltrated depth of 46 mm at 55 m
distance with predicted value of 67 mm. Having reached the 90 m distance the 40,80 and 160
137
L min-I inflow rates recorded measured infiltrated depths of 35, 80 and 88 mm while the
corresponding predicted depths were 41, 80 and 89 mm. Two-dimensional predicted soil
water distributions from selected furrow measurement point for the different inflow rates are
presented Figures 6.6, 6.7, 6.8 and 6.9. The saturated infiltrated depth (SID) was at different
levels of developments with the 20 and 40 L min-I at the inlet 5 m point showing a
continuous blocky infiltrated zone reaching the C-horizon. The SID for the wetted furrow end
was limited to only the upper layer within the wetted perimeter of the furrow. In the 80 and
160 lImin inflow rates the Sill was restricted at the inlet to the upper layer while further away
it had progressed towards the C-horizon. On all the snapshots constrictions on the Sill,
transitional zone and wetting front could be noted at depth of 300 and 600 mm. The R2 values
given by the HYDRUS-2D were good (>0.95) at the 5 m distance from all inflow rates. At
the last wetted point of the furrow the R2 value had decreased sharply, but was still
acceptable (> 0.69).
6.3.4 irrigation infiltration functions
Average furrow infiltration functions derived from one point method were shown in Figure
6.10. Parameters for the simple power based infiltration function and the outcomes of
homogeneity tests are summarized in Table 6.6.
Average infiltration function projected infiltrated depths that varied from those from
predicted measurements, especially at the early stages of infiltration. In the 20 L min-I the
predicted measurements reflected a rapid start, while that of the infiltration function was slow
up to the 0.78 hour mark where both curves matched fairly well until the end of the plot. In a
similar manner the curves from the 80 and 160 L min-I inflow rate matched at about 0.67 and
0.81 hours. A different trend was observed from the 40 L min-I with predicted depths
overestimated by the infiltration function up to the 0.18 hour mark after which the two curves
run close to each other. The final average opportunity time was 0.38 hours compared to the
1.1,0.83 and 1.01 hours for the 20, 80 and 160 L min-I inflow rates.
Table 6.6 Bartiett's homogeneity test between measured and predicted infiltration functions
for the different inflow rates
Inflow rates (L min-I) Computed X2 Tabular X2 (0.05.2)
20 0.4221 3.84
40 0.0000012 3.84
80 0.0038 3.84
160 0.0048 3.84
138
Sm
A-horizon
B-horizon
C-horizon
0.12-9 0.14-8 0.18-7 0.1e-8 0.20-8 0.22-5 0.2" 0.283 0.28-2 0.30-2 0.32-1 O.MO
35 m
A-horizon
B-horizon
C-horizon
0.131•0.150•0.16•St 0.18•8 0.20•7 0.22•8 0.245 0.2&4 0.283•0.302•0.321•0.340
Water Cont.nt - th(-), Mln-O.131, Ma.-O.:wG
55 m
A-horizon
B-horizon
C-horizon
0.13'-0.15-4 0.17-3 0.18-2 0.21-0 0.2-28 0.2"70.21' 0.214-0.30-3 0.32-10.340
Figure 6.5 Subsurface soil water distribution following irrigation at 5 rn, 35 rn, and 55
furrowlength for the 20 L mm-I inflow rate.
139
Sm
A-horizon
B-horizon
C-horizon
0.131•0.1550•.1730.•1820.2•100.22•80.247• 0.218 0.284•0.303•0.32•1 0.$40
55 m
A-horizon
B-horizon
C-horizon
0.13•1 O.15•00.HI•IO.111•0.20•7 0.2•21 0.245 0.214 0.28•3 0.30•20.321•0.340W.ter Content - th(-J, Mln-G.131, M... -O.340
85 m
A-horizon
B-horizon
C-horizon
0.12•10.148•0.18•7 0.1"•0.20•5 0.22•5 0.244 0.283 0.28•2 0.302 0.321 0.340W.ter Content - th(-). Mln.(t.UI, M... O.S40 ••
Figure 6.6 Subsurface soil water distribution following irrigation at 5 m, 55 m, and 85 m
furrow length for the 40 L mm-I inflow rate.
140
Sm
A-horizon
Bvbcrizen
C-horizon
Q.13-e 0.15-7 0.17-6 0.1-94 0.2-12 0.2-30 0.249 0.287-0.285 0.303 0.322 0.340Water Cont.nt - th[-). Mln-O.139, Ma.-O.3-40 --
55 m
A-horizon
B-horizon
C-horizon
0.130-0.140-0.1tU-. 0.18-8 0.2-07 0.2-26 0.245 O.26ot 0.283-0.30-2 0.32-10.340Wet.r Cont.nt - thH, Mln-O.130, Max-O.340
85 m
A-horizon
B-horizon
C-horizon
0.131-0.15-0 0.18-8 0.1-'. 0.20-7 0.2-2. 0.245 0.2'"-0.28-3 0.302 0.321 0.340Wat.r Cont.nt - th(-]. Mln-O.i31, M•.x-O.340 --
Figure 6.7 Subsurface soil water di tribution following irrigation at 5 m, 55 rn, and 85 furrow
length for the 80 L mm" inflow rate.
141
Sm
A-horizon
B-horizon
C-horizon
0.12-20.142-0.1120-.1810.-2010.2-210.24-10.2810.2810-.300 0.320 0.3<40Wat.r Cont.nt - the-J, Mln-a.i22, M.x-O.M-O --
55 m
A-horizon
B-horizon
C-horizon
0.12-10.141-0.1110-.1110-.2010.-2210.2-410.280 0.28-0 0.30-0 0.32-0 0.340W.t.r Cont.nt - th(-), Mln-O.121, M•• -O.J.40
85 m
A-horizon
B-horizon
C-horizon
0.12-10.141-0.1110-.1810.-2010.2-210.24-10.211 0.2100.300 0.3200.3<40W.t.rCont.nt -the-J, Mln-O.121, Max-O.-340 --
Figure 6.8 Sub urface oil water di tribution following irrigation at 5 rn, 55 rn, and 85m
furrow length for the 160 L nun- J inflow rate.
142
~
E 120
E A
'---' lOO
"'0
il)
~,_ 80
- --t\. ---------6
.......
c..::: 60
t:
. 40....t..:..
0.
il) 20
0
0
~ 120
E B
E
'---' lOO
"'0
..i.l..).. I!r----- _
C';l 80 - - - - - - - -t;,. _
.t:
c..:::
t: 60
..t:
0. 40
il)
0 20
0
~ 120
E c
E lOO
'---'
"'0
il)
~ 80,_ ----
.......
1.;: 60
t:
..t: 40
0.
il)
0 20
0
~ 120 D
E
E lOO -
'---'
"'0 80 1!r 4F----------- -eil)
~,_
....... 60 13-"
c..:::
t: 40
.....t..:..
0.
il) 20
0
0
0 lO 20 30 40 50 60 70 80 90
Furrow distance (m)
-e- Measured - -6- Predicted
Figure 6.9 Distribution of infiltrated depth along the furrow length for the 20 (A), 40 (B), 80
(C) and 160 L min-I (D) inflow rate.
143
,-.., 90
E
._E
80 A
_.
-0 70
Q)
~ 60 ---
.:.:.:. 50 0 ---- ----
r.;:
c 40 --- ---
---
--a=0.296
..c 30 --- --- ---
---
0- a= 0.453
Q)
Cl 20la 0 Predicted
0
,-.., 90
E B
._E_. 80 --a=0.228
-0 70
..Q...).
('j 60 a= 0.227
.t:
r.;: 50 I::. Predicted
c 40
.....c...
0.. 30
Q)
Cl 20
la I::.
0 I::.
90 C
,.-...
E 80
._E_. 70
-0
.Q......
). 60 --a=0.40
('j
....... 50 a= 0.66
r.;:
c 40 I::. Predicted
..c 30
0-
Q) 20
Cl la
0
90
,.-... D
E 80
._E_. 70 ............
-0
.Q....
)
...
60
('j
....... 50 --a=0.39
r.;:
e 40
.....c... 30 .. ..
a= 0.74
0.. I::. Predicted
Q) 20
Cl la ..
0
0 0.2 0.4 0.6 0.8 1.2
Opportunity time (hours)
Figure 6.10 Average furrow irrigation functions for predicted (solid line) and for measured (broken
line) corresponding to the different constants (a) adjusted to conserve mass for the 20 (A), 40 (B), 80
(C) and 160 (D) L min-I inflow rates.
144
6.3.5 Average infiltration depth and measures of performance
Average infiltrated depths from measured and predicted SWC measurements from actually
wetted furrow distanced are presented in Table 6.7. Average infiltrated depths approximated
from the average furrow infiltration function (AFIF) for the 60 and 30 m long furrows are
also shown. In addition appropriate measures of irrigation performance including DU, AE
and DPR were given.
Average infiltrated depth for the 90 m furrow distance ranged from 66 to 87 mm and 74 to 71
mm from the respective measured and predicted SWC measurements. The highest depth from
the measured came from the 40 L min·1 inflow rate while from the predicted was from the
160 L min-I. At this furrow distance average infiltrated depths between measured and
predicted SWC were found to be homogeneous. At the 60 m distance only the 20 L min-I was
able to actually wet this furrow distance with average depths of 67 and 82 mm from the
measured and predicted SWc. These two averages were comparable according to the Student
t-test analysis. For same 60 m furrow lengths, the average infiltrated depth approximated
using the AFIF ranged from 64 to 95 mm and 91 to 64 mm for the measured and predicted
SWc. Highest depths were from the 160 L min-I and the least from the 40 L min-I .The
Student's t-test analysis found the approximated infiltrated depths from the measured and
predicted SWC of the 40, 80 and 160 L min-I not comparable for the 60 m furrow length.
Corresponding to the 30 m furrow length, average infiltrated depths ranged from 34 to 58 mm
and 58 to 69 mm for the respective measured and predicted SWC. The highest average
infiltrated depth was from the 40 L min-I under measured SWC while from the predicted
SWC was given by the 160 L min-I inflow rate. Comparable differences between the
measured and predicted SWC average infiltrated depths were found from the 80 and 160 L
min-I inflow rates.
Measures of performance included the DU, EA and DPR and were found applicable to the
actual furrow distances wetted by the selected inflow rates. Where the AFIF infiltrated depth-
oppotunity time relationship was employed to intepolate irrigation parameters for the 30m
and 60m furrow lengths, only DU was used. In the 90 m long furrows the DU ranged from
0.69 to 0.85 and 0.67 to 0.92 from the respective measured and predicted SWC.
Corresponding EA ranged from 0.93 to 0.99 and 0.88 to 0.78 while the DPR ranged from
0.01 to 0.07 and 0.12 to 0.22, respectively.
145
------------------------------------------------~~-------------------------------------------------- ..------------------------------------------------------------------------------------ ....~
Table 6.7 Average infiltrated depth and performance indicators for different furrow lengths and inflow rates.
ID (m) DU AE DPR
Furrow Inflow
length rates
(m) (I/min) Measured Predicted Measured Predicted Measured Predicted Measured Predicted
30 20 34.30 57.71 0.89 0.93
40 58.22 58.16 0.96 0.96
80 50.51* 60.65* 0.96 0.97
160 55.29* 68.79* 0.89 0.94
60 20 67.15* 82.40* 0.70* 0.87 0.93 0.88 0.06 0.16
40 63.68 63.64 0.83 0.83
80 81.19 81.05 0.998 0.99
160 94.38 91.14 0.87 0.93
90 20
40 86.50 72.90 0.69 0.67 0.93 0.88 0.07 0.12
80 65.99 71.15 0.82 0.95 0.94 0.86 0.06 0.14
160 74.23 73.74 0.85 0.92 0.99 0.78 0.01 0.22
* significant different at 0.05 T-test
ID = Infiltrated depth
AE = Application efficiency
DPR= Deep percolation ratio
DU = Distribution uniformity
146
The 20 L min-I was able to actually wet the 60 m furrow distance with a DU and EA values
of 0.7 and 0.87 from the measured SWC and 0.93 and 0.88 from the predicted SWC with a
corresponding DPR value of 0.06 and 0.16. From the 20, 40,80 and 160 L min-I inflow rates
the DU ranged from 0.9 to 0.7 and 0.99 to 0.83 for the measured and predicted SWC. In the
30 m furrow distance the corresponding DU values were ranging from 0.86 to 0.89 and 0.93
to 0.93.
6.4 Discussion
Selecting an inflow rate that satisfies the claims of soil water infiltration and runoff of the
advance stream to the furrow end is one of the many challenges in designing furrow
irrigation. Surface and subsurface water distribution from single run irrigation with a 20, 40,
80 and 160 20 L min-I inflow rates on a 90 m long furrows provided insight on the
appropriate inflow rate-furrow length combination for the Tukulu soil.
Among the inflow rates it was only the 20 L min-I that could not sustain an advance stream to
the furrow end, but approached zero at the 60 m point. Despite the inflow rate being small the
high surface roughness of the newly prepared furrow is one reason the advance stream went
to zero. According to the Manning Roughness Index a newly prepared furrow has a
roughness coefficient of about 0.06 and 0.02 for very smooth furrows (ASAE, 2003).
Restrictive effect of furrows with high surface roughness on the advance stream is well
acknowledged in literature (Clemmens et al., 2001; Walker, 2003; Hulugalle et al., 2007;
Clemmens and Bautista, 2009). Raising the inflow rate to 40 L min-I was able to reach the 90
m, but at the advance time of 105 minutes compared to the 35 and 23 minutes clocked by the
80 and 160 20 L min-I inflow rates. The small advance times were expected given the high
runoff potential of large inflow rates (Walker, 2003). The high roughness could have had
little effect on the advance times of the larger inflow rates. However the immediate rise in
averaged depth at the 25 m point to 92 and 105 mm does suggest that advance at this distance
were slowing down and high roughness of the furrow surface being the most appropriate
reason. Following cut off the recession periods between the furrow inlet and end from the 20
and 40 L min-I is about 4 and 7 minutes, respectively. Because of the small recession periods
irrigation could be implied to have stopped simultaneously at cut off an assumption that is
upheld by the kinematic wave model (Walker and Skorgeboe, 1987). In the 80 and 160 L
min-I the time difference of recession period between the furrow inlet and end was 62 and 71
minutes, respectively. Occurrence of ponding at the furrow end was the main reason for the
147
large differences in recession for the 80 and 160 L min-J• The generally high average stream
flow depth and the signs of furrow shape deformations as well as potholing near the inlet
despite provisions made to prevent erosion could be regarded as indicators of the risk of
erosion and overtopping from the 80 and 160 L min-J inflow rates. Walker (2003) shared the
same sentiment and suggested emergency facilities to drain excess water especially when
large inflow rates are applied. Management of advance on long furrows was also seen to be
more challenging compared to short furrows by Wohling and Maihol (2007).
Application of the HYDRUS-2D software in predicting the infiltrated depth from initial and
final SWC measurements was satisfactory. Agreement between the measured and predicted
initial and final SWC were observed from all inflow rates. Optimised parameters for all
horizons was the er from the 20 and 40 20 L min-J and the a from the 80 and 160 L min-J
Overestimates of infiltrated depth was common among the predicted SWC measurements
especially at the inlet from the larger inflow rates while from the smaller inflow rates it was
towards the furrow end. The model assumed es to be equal to the porosity value and yet under
field conditions the presence of dead end pores, entrapped air and bulk density variations
makes achievement of maximum saturation impractical (Hillel, 2004). Underestimation of
infiltrated depth was only pronounced in the 40 and 20 L min-J with a fairly good agreement
at the furrow end. These variations were expected given that SWC measurements were one
dimensional while the predicted values were computed using a two dimensional model. Much
of the erratic variations could be attributed to spatial variability of the soil properties along
the 90 m furrow length. Similar situations were also encountered by Raghuwannshi and
Wallender (1996), Abbasi et al. (2003), and Wohling and Maihol (2007). However, good
agreement at the lower quarter of the furrow especially from the 80 and 160 L min-J could be
pointing out to the homogenizing effect of ponding on this soil.
Infiltrated depth from both measured and predicted SWC showed greater consistency to the
opportunity time and average depth of stream flow. This was not surprising because surface
and subsurface water distribution should conserve mass. Total water volume applied during
irrigation according the inflow rate and irrigation-cut off time was equivalent to 3.2, 4.2, 2.8,
and 3.7 m3 from the 20, 40, 80 and 160 L min" inflow rates. The higher volume of water
applied for small inflow rates corresponds to the large irrigation times, 162 and 105 minutes
for the respective 20 and 40 L min", attributed to the slow advance stream observed from
these inflow rates. Of interest was the distribution of the water in the profile since no water
escaped at the blocked ended furrow. As expected infiltrated depth was greater at the inlet
148
from the 20 and 40 L min-I while from the 80 and 160 L min-I was skewed towards the lower
quarter of the furrow. This variation could be explained by the fact that most of the delivered
water from the smaller inflow rates was infiltrated before making significant advances
implying that the inflow rate were very close to the soil water basic intake. On the contrary,
the 80 and 160 L min-I inflow rate recorded small advance times indicating that a significant
proportion of its streamflow was able to reach the furrow end before making time for
infiltration. This observation was supported by the high opportunity times at the furrow end
due to ponded water implying the 80 and 160 L min-I were way above the soil water basic
intake. The graphical snapshots from HYDRUS-2D did capture clearly the two situations
represented by the two groups of inflow rates. At the inlet of the 20 and 40 L min-I the
saturated infiltrated depth (SDI) extends uniformily to the C-horizon where it developed
sideways. This phenomenon suggests the presence of a restrictive layer at the 600 mm
suggesting that deep drainage losses was not a major problem in this soil. Confirming this
observation was the abrupt transition of 40 to 1.9 mm hour" in Ks values from the B to the C-
horizon. Hensley et al. (2000) shared the same sentiment about the formation of heavy
structured clay rich underlying C-horizon among group of layered soils in the Free State
province as key to profile soil water storage. Between the A and B horizon a discontinuity in
SID and wetting front could be observed suggesting that the layering between these horizons
was not severe and that it had little effect on deep infiltration.
The use of one point method to determine the AFIF was thought to be able to give reasonable
estimates of infiltrated depth for any furrow depth since it was based on improving the
conservation of mass. The backwater effect from the 80 and 160 L min-I inflow rates as a
result of ponding at lower end of the blocked furrow should have made the two point method
(Clemmens et al., 2001; Walker, 2003) ill posed for this work. The AFIF was based on
infiltrated depth-opportunity time relationship because ponded infiltration is a function of
time rather than surface infiltration characteristics (Clemmens et aI., 2001). Application of
the geometric mean, considered by some to be appropriate for non linear conditions, could
not eliminate the inconsistencies caused by spatial variations and the use of a simplified
power law infiltration model. Therefore, guarantees for accuracy from the AFIF could not be
made. Differences in infiltration parameters between measured and predicted AFIF was the
major source of variations. This was expected given the differences in opportunity times
between the furrow inlet and end from all inflow rates. A similarity between measured and
predicted average infiltration parameters by AFIF for the 40 L min-I was an exception and
149
can be explained by the spatial distribution of infiltrated depth and opportunity times
relationship depicted in Figure 6.9 (B). This observation could be indicating the challenges
this inflow rate encountered in sustaining adequate stream flow to wet the furrow end. For the
20 L min-I it was able to attain a higher opportunity time at the 55 m point because the
advance stream was given some time to demonstrate if it could progress beyond the 60 m
mark or not. Austin (2003) under similar circumstance noted that once the advance stream
along the furrow length has been reduced to soil basic water intake, it instantly goes to zero.
Further supply of water increases infiltrated depth rather than advance distance. This analogy
explains why infiltrated depth was high near the inlet of the 20 and 40 L min-I. Better water
distribution along the furrow contributed to the higher average infiltrated depth from the 160
L min-I. The 40 and 20 L min-I had higher distribution at shorter distances. Underestimations
from the infiltration function were not uncommon especially at early stages of infiltration;
overshooting tendencies of infiltration on initially dry soil is the contributing factor (Hillel,
2004).
A strong agreement between the AFIF with the final infiltrated depth from the three selected
measurement point made it feasible to generate infiltrated depth for the entire wetted furrow
based on the corresponding opportunity times. The resulting distribution of infiltrated depth
was typical of the subsurface distribution from various irrigation regimes. For the 90 m
distance the 40 L min-I had the highest average infiltrated depth of 86.5 mm and this was
consistent with the 4.2 m3 volume of water applied in this furrow. Although 80 and 160 L
min-I supplied smaller volumes of 2.8 and 3.7 nr', respectively they had a better
corresponding DU and AE. This was consistent with observations of Austin (2003) that
higher inflow rate produce better irrigation efficiencies despite difficulties in managing large
advance streams. Reduction of furrow length to 60 m produced average infiltrated depths, as
determined by AFIF that improved distribution uniformity from all inflow rates. Even from
the 20 L min-I was from actual observations produced reasonably high measures of
performance. This was in agreement with the view that shortening the furrow length would
reduce differences in opportunity time between the furrow inlet and the end (Walker, 2003;
Austin, 2003). For the 30 m furrow length the average infiltrated depths were decreasing
proportional to opportunity time. Inconsistencies from the 80 and 160 L min-I in the 60 m
length could be attributable to overestimation of the opportunity time that included the
ponding phase for the lower quarter of the furrow. Improvement of DU was remarkable from
the 20 and 40 inflow rates while from the large rates it was stable. Such a phenomenon could
150
be explained by view upheld by Austin (2003) that small inflow rates on short furrows could
achieve efficiencies as good as those from pressured systems. Interestingly, the average
infiltrated depths between measured and predicted SWC from all inflow rates that wetted the
90 m point passed the homogeneity test, but the same inflow rates with the exception of the
40 L min'I failed the same test for the 30 m furrow length. Could this mean that the
applications of the larger inflow rates were ill posed for the 30 m furrow length?
6.5 Conclusion
The exercise of determining the appropriate inflow rate for micro-flood irrigation on the
Tukulu soil was undertaken by a single irrigation with four different inflow rates on a 90 m
long furrow. Surface water distribution during irrigation measured during the advance-supply
and recession phases varied among the inflow rates. During the irrigation period the advance-
supply phase demonstrated its dependence on the size of inflow rate. The 20 L min'I required
2.7 hours to support an advance stream that went to zero at the 60 m point after 2.7 hours of
irrigation time. The 40 L min'I reached the 90 m after 1.8 hours while the 80 and 160 L min'I
becuase of their large streamflow size required 0.4 and 0.6 hours of irrigation time,
respectively. The recession time was directly related to the size of inflow rate with the
smaller inflow rates lasting less than 7 minutes while the larger rates lasted more than an
hour.
Involvement of the HYDRUS-2D to predict initial and final SWC for the soil profile horizons
provided further insight on the subsurface water distribution under the wetted furrows. It also
aided to the application of the one point method for determining the AFIF for the respecti ve
inflow rates. Model inverse predictions were improved with minimum number of parameters
optimised. Variations between the measured and predicted SWC as well as the AFIF were to
be expected given soil spatial variations common in layered soils. However, conservation of
mass was able to be improved when average infiltrated depth and adjustment of infiltration
parameters was carried out.
Consistency of the infiltrated depth to opportunity time made it practical for the AFIF to
approximated corresponding depth for the 30 and 60 m distance furrows. Assessment of
subsurface water distribution indicated that the prismatic C-horizon of the Tukulu soil
restricted deep drainage losses from all inflows rates irrespective of furrow length. Measures
of performance reflected higher AE and DU from the 80 and 160 L min" inflow inflows at 90
m furrow distances. Reducing the furrow length to 60 and 30 m showed a remarkable
151
improvement in DU among the 20 and 40 L min-I while for the larger inflow rates it remains
more stable. Given the practicality of working with small inflow rates and the flexibility to
adjust the time and frequency of irrigation it would be recommended that the 40 L min-I
inflow rate be adopted for furrow distances of less than 60 m.
Acknowledgements
Special thanks to the staff and management of Paradys experiment station of the technical
and hospitality they provided during this experiment. Gratitude's to Ms Liesl van der
Westhuizen for constructive suggestions on improving the writing of this manuscript.
152
6.6 References
Abbasi, F. Fayen, J. and van Genuchten, M. Th., 2004. Two-dimensional simulation of water
flow and solute transport below furrows: model calibration and validation. Journal of
Hydrology, 290: 63-79.
Abbasi, F. Jacques, D. Simunek, 1. van Genuchten, M. Th., 2003. Inverse estimation of soil
hydraulic and solute transport parameters from transient field experiments:
Heterogeneous soil. Am. Soc. OfAgric. Eng., 46: 1097-1111.
ASAE, 2003. Evaluation of Irrigated Furrows, ASAE EP419.1 FEB03, American Society of
Agricultural Engineers.
Austin, c., 2003. Micro flood, a new way of applying water. http://waterright.com.
22/0912011, 10.00 am (LT).
Awulachew, S B, Merrey, DJ, Kamara, A B, van Koppen B, Penning de Vries F, Boelee E.
2005. Experiences and opportunities for promoting small-scale/micro irrigation and
rainwater harvesting for food security in Ethiopia. International Water International
Water Management Institute, Working paper 98, Addis Ababa, Ethiopia.
Bargar, B, Swan, J. B. and Jaynes, D., 1999. Soil water recharge under un-cropped ridges and
furrows. Soil Sci. Soc. Am. 1., 63: 1290-1299.
Botha, 1. J., 2006. Evaluation of maize and sunflower production in a semi-arid area using In-
field rainwater harvesting. Ph.D. (Agric ) Dissertation, University of the Free State,
Bloemfontein, South Africa
Clemmens, A. J., Eisenhauer, D. E. and Maheshwari, B. L., 2001. Infiltration and roughness
equations for surface irrigation: How form influences estimation. The Society for
Engineering in Agricultural, Food and Biological Systems, Paper number 01-2255.
Clemmens, A. J. and Bautista, E., 2009. Toward physically based estimation of surface
irrigation. J. of Irrig.and Drain. Eng., 135: 588-596.
Hensley, M. Botha, J. J., Anderson, J. J., Van Staden, P. P., Du Toit, A. 2000. Optimising
rainfall use efficiency for developing farmers with limited access to irrigation water.
Pretoria, Water Research Commission report 878/1/00, South Africa.
Hensley, M., Le Roux, P., Gutter, J. and Zerizghy, M.G., 2007. A procedure for an improved
soil survey technique for delineating land suitable for rainwater harvesting. WRC
Report No. TT 311/07.Water Research Commission, Pretoria, South Africa.
Hillel, D. H., 2004. Environmental soil physics. Academic Press, New York, USA.
153
Hulugalle, N. R., McCorkell, B.E., Weaver, T.B., Finlay, L.A. and Gleeson J., 2007. Soil
properties in furrows of an irrigated Vertisol sown with continuous cotton
(Gossypium hirsutum L.). Soil and Tillage Research, 97: 162-171.
Lazarovitch, N., Warriek, A.W., Furman, A. and Zerihum D. 2009. Subsurface water
distribution from furrows described by moments analyses. J. of Irrig.and Drain. Eng.,
135: 7-12.
Marquardt, D.W.,1963. An algorithm for least squares estimation of non-linear parameters. J.
Ind. Appl. Math., 11: 431-441.
Oyonarte, N. A., Mateos. L. and Palomo, MJ., 2002. Infiltration variability 10 furrow
irrigation. J. of Irrig.and Drain. Eng., 128:26-33.
Raghuwanshi, W.W. and Wallender, W.W. 1996. Modelling seasonal furrow irrigation. J. of
lrrig. and Drain. Eng., 122: 235-242.
Rodriguez, J. A., Diaz, A., Reyes, J. A., and Pujols R. 2004. Comparison between surge
irrigation and conventional furrow irrigation for covered black tobacco cultivation in
a Ferralsol soil. Spanish Journal of Agricultural Research, 2: 445-458.
Romano, N., Brunone, B. and Santini, A.,1996. Numerical analysis of one-dimensional
unsaturated flow in layered soils. Advances in Water Res., 21: 315-324.
Schaap, M.G., Leij, F. J. and van Genuchten, M. Th., 2001. Rosetta: A computer program for
estimating soil hydraulic parameters with hierarchical pedotransfer functions. J.
Hydrai., 25: 1163-176.
Scotter, D. R., Clothier, B. E. and Harper, E. R., 1982. Measuring saturated hydraulic
conductivity and sorptivity using twin rings. Australian Journal of Soil Research, 20:
295-304.
Simunek, J. Sejna, M., and van Genuchten, M.Th., 2009. The HYDRUS-2D software
package for simulating two-dimensional movement of water, heat and multiple
solutes in variably saturated media, Version 2.0. Rep. IGCWMC-TPS-53, p. 251, Int.
Ground Water Model.Cent., Colo. Sch. of Mines, Golden, CO.
Soil Classification Working Group, 1991. Soil Classification - A taxonomic system for South
Africa. Memoirs on the Agricultural Natural Resources of South Africa No. IS.
Department of Agricultural Development. Pretoria, South Africa.
Tabuada, M. A., Rego, Z. J. C., Vachaud, G. and Pereira, L. S., 1995. Two-dimensional
infiltration under furrow irrigation, its validation and application. Agricultural water
management, 27: 105-123.
154
van Genuchten, M.Th., 1980. A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils. Soil Sci. Soc. Am. J., 44: 892-898.
Walker, W.R., 2003. Surface irrigation simulation evaluation and design - User Guide and
Technical Documentation.Utah State University, Logan, Utah.
Wohling, T. H., and Maihol, J. C., 2007. A physically based coupled model for simulating I-
D and surface 2-D subsurface flow and plant water uptake in irrigation furrows. II:
model test and evaluation. J. of lrrig.and Drain. Eng., 133: 543-553.
World Reference Base for Soil Resources (1998). World soil resources report, 84, ISSS /
ISRIC FAO, Rome, Italy.
155
CHAPTER 7
CHARECTERISING VERTICAL REDISTRIBUTION ON IRRIGATED SHORT
FURROWS IN THE TUKULU SOIL
Abstract
Subsurface soil water redistribution in the Tukulu soil of South Africa also referred as the
Cutanic Luvisols in other countries was evaluated on a closed ended, 90 m furrow. Following
single run irrigation with 20, 40, 80 and 160 L min-I inflow rate treatments, changes in soil
water content (SWC) at the three profiles diagnostic horizons were monitored using a neutron
water meter. Measurements were made at every 10 m distance starting from 5 m from the
furrow inlet for a period of 455 hours from irrigation. The HYDRUS-2D software was used
to estimate the soil hydraulic parameters through the inverse optimization algorithms for the
redistribution process at the inlet, midpoint and furrow end of the different furrows. Models
optimised parameters were comparable with initial estimates and the agreement between
measured and predicted SWC was quite satisfactory, despite noticeable spatial variability.
Calculated effective K-coefficient (Keff) for the 0-600 and 0-850 mm profile flow domains
demonstrated linearity with average SWC although inconsistencies under field conditions
were inevitable. The underlying layer restricted gravity and augmented redistribution with
Keff assuming a steeper gradient than normal. Converting the linearity between Kerf and
average SWC into a ratio assisted in quantifying redistribution of the infiltrated depth for the
consecutive profile flow domains.
Key words: Soil water redistribution, inflow rates, effective unsaturated hydraulic
conductivity, infiltrated depth, HYDRUS-2D.
156
7.1 Introduction
Subsurface soil water redistribution is one of the important determinants of furrow irrigation
efficiencies. In recent times efforts towards improving furrow irrigation techniques has been
complimented by advances in mathematical tools to describe the infiltration dynamics and the
water flow and solute transport long after irrigation has ceased. The latter is referred as
redistribution; a hydrological process depicting the transient flow of infiltrated water from the
nearly saturated upper layers to the less infiltrated lower layers of the profile.
Soil water redistribution along furrows is dependent to a number of factors. Design factors
include furrow shape, stream flow depth and wetted perimeter (Clemmens et al., 2001) while
management factors embraces initial soil water content, crop and climate factors, and
infiltrated depth (Lazarovitch et al., 2009). Another important determinant of redistribution is
the soil hydraulic properties and their pertinent parameters. The soil water characteristics
curve (SWCC) represents the unique relationship between the soil water content (SWC) and
matric suction of a particular soil and the unsaturated hydraulic conductivity (K-coefficient)
are the primary hydraulic functions widely used as input data into the governing flow
equations. Despite the latter being the most difficult hydraulic property to measure, physical
based approaches of describing the transient nature of redistribution requires accurate
estimates of the K-coefficient (Hillel, 2004). The K-coefficient usually varies with depth as
the upper and lower profile layers desorbs and wets monotonically. The effect of textural and
structural spatiality on soil water redistribution across the flow domain results to a unique
relationship between the infiltrated depth to the matrix suction and K-coefficient (Rubin,
1967; Leconte and Brissette, 2001; Hillel, 2004).
Various numerical methods with varying levels of complexity have become available for the
estimation of K-coefficient along partially wetted furrows. One of these techniques is spatial
moments (Aris, 1956), that has a strong pedigree with the description of movement and
spread of a solute plume (Sudicky, 1986; Barryand Sporto, 1990; Srivastava et al., 2002).
Recently, it has become popular in quantifying redistribution of soil water plume especially
from drip and furrow irrigation systems (Gee and ward, 2001; Yeh et al., 2005; Lazarovitch,
et al., 2007, 2009). This technique relies on SWC measurements to calculate the zeroth
moment, first moment and second moments in one or more flow dimensions to describe the
change in soil water storage, spatial variance of the centre of the plume and the spread of the
plume, respectively. In a layered soil, the spatial heterogeneity of the flow domain is usually
157
treated as an equivalent homogeneous medium by means of ensemble mean or spatial
average of the hydraulic properties. This concept was successfully demonstrated by Yeh et al.
(2005) who estimated average or effective K-coefficient from the spatial moments. Previous
work on effective hydraulic properties have found this approach acceptable and offering
reasonable estimates of the average flow behaviour under spatially variable conditions
(Warrick, 1991; Baker et al., 1999; Wildenschild and lensen, 1999; Gastro et al., 2002).
An alternative method of estimating the hydraulic properties including the K-coefficient
during redistribution is the inverse technique. Numerous studies have used this approach with
great success (Wohling et al., 2004 and Wohling and Maihol, 2007) also in layered soils
(Abbasi et al., 2003). Lazarovitch et al. (2009) used this technique to estimate the soil
hydraulic parameters which were processed further to generate spatial moments for soil water
plume below furrows. In essence the inverse method estimates the K-coefficient either as a
function of SWC or matrix suction (h) from physical properties such as SWC and h that are
easier to measure (Kool and Paker, 1988; Hopmans et al., 2002). Having evolved under
laboratory conditions several hydraulic models that predict the K-coefficient from the SWCC
curve (Brooks and Corey, 1964; van Genuchten, 1980; Kosugi, 1996) have become more
available. Development of the physically based models such as the HYDRUS-2D code
(Simunek et al., 2009) that includes inverse optimization algorithm has made the inverse
method more attractive.
Growing need to improve food production has increased the cultivation of layered soils in the
semi-arid areas of the Free State Province of South Africa. Such an initiate has been
confronted with the dry soil water regime that is exacerbated by high runoff and evaporation
losses from these soils. In response the in-field rainwater harvesting (IRWH) production
technique (Hensley et al., 2000) was developed to transform surface runoff between crop
rows into deep infiltration. Because of the restrictive layer underlying the profile of this
group of soils, deep infiltration contributes to soil water storage. However, the need for
supplementary irrigation during the prolonged dry spells has resulted into the adoption of
micro-flood irrigation (MFI) from Austin (2003) that is currently being developed to operate
alongside the IRWH technique. Given that subsurface water storage is a function of
redistribution, it is therefore critical that the redistribution hydraulic properties be
characterised before the two techniques are merged.
158
It was therefore against this background that the Tukulu (Tu), as a representative of the
layered group of soils occupying more than 12 % of the provincial landscape, was used to
describe how its profile horizon affected redistribution along short furrows under different
inflow rates. The South African Tukulu is similar to the Cutanic Luvisols of the Reference
Soil Group and covers an estimated 500 to 600 million hectares worldwide (World Reference
Base for Soil Resources, 1998). The specific objectives of this study were firstly, to estimate
the soil hydraulic parameters of the Tukulu soil that describes soil water movement during
redistribution using the Levenberg-Marquardt parameter optimization algorithm. Secondly, to
compare the HYDRUS-2D models predictions of field measured soil water content and
calculated effective K-coefficient and thirdly to estimate the relationship between the rate of
redistribution and infiltrated depth from furrows under different soil water regimes.
7.2 Material and methods
7.2.1 Experimental site
The field experiment was carried out at Parady's Experiment Farm (29°13'24, 69"S, 26°12'40,
93" E, altitude 1422 m) of the University of the Free State, South Africa. The selected site
was predominately of the Tukulu soil form according to the Soil Classification Working
Group (1991). The soil profile description is summarized in Table 7.1 and illustrated in a
photo in Figure 7.1.
Table 7.1 Summary of the soil physical properties of the Tukulu soil form
Physical properties Master horizons
A Bl C
Coarse sand (%) 5.3 9.2 2.1
Medium sand (%) 9.3 8.8 3.8
Fine sand (%) 41.2 31 28.3
Very fine sand (%) 25.3 21 8.4
Coarse silt (%) 2.1 2 3
Fine silt (%) 4.6 2.5 6.5
Clay (%) 11.3 26.4 47.9
Bulk density (kg m-3) 1670 1597 1602
Porosity (%) 34.0 33 32.4
(Ks, mm hour -I) 36.1 40 9.6
Ks= saturated hydraulic conductivity
159
Ca)
(b)
(600-850
motties
Figure 7.1 (a) Profile of the Tukulu and (b) mottles present in the C-horizon.
7.2.2 Experimental design and measurements
Redi tribution of soil water in the Tukulu soil wa characterized on 90 m distance furrows
that has been irrigated with a single run. Four plots with three furrow replications were
systematically a igned to four inflow rates including the 20, 40, 80 and 160 L min-I. Only
the central furrow in a plot was installed with neutron access tubes of 1. 2 m long at 10 m
distance intervals starting from the 5 m from the furrow inlet. An area of 2 m by 2 m around
each access tube was covered with a plastic to minimise the influence of evaporation on
sub urface water redistribution as hown in Figure 7.2. A site calibrated neutron soil water
meter was used to monitor the change in SWC during the redistribution period that
commenced immediately after the depletion-rece sion phase of irrigation. Measurements
were taken two times a day for the first week and then daily for a fortnight taken at depths of
150 mm, 450 mm, and 725 mm.
160
Figure 7.2 Shading of the furrow section measurement stations at 10 m intervals.
7.2.3 Description of the soil hydraulic properties
Soil water characteristic curve (SWCC) corresponding to the diagnostic A-, B- and C-
horizons of the Tukulu soil was determined by using a combination of the hanging water
column method and the pres ure chamber plate desorption technique. The hanging water
column measured the matric suction (h) and soil water (e) relationship for a range of 0 to -
lOkPa, while suctions from -10 to -1500 kPa mm were measured using chambers of various
pressure sizes. The resulting e-h relationship was then parameterised using the closed-form
equation of van Genuchten (1980) expressed as:
(7.1)
Where es and er are the respective saturated and residual values of the volumetric soil water
content, e (mm mm"), h is the matric suction (kPa), while a and n are the shape and pore
ize distribution parameters, respectively. The condition m=l-lIn should be satisfied and e
(h) equals es when at zero or positive suctions. Parameters related to pore-size distribution (n)
and shape (a) of the SWCC was estimated from the Rosetta pedo-transfer (Schaap et al.,
2001) in agreement with the field measured saturated hydraulic conductivity (Ks). Table 7.2
summarizes the initial estimates of soil hydraulic parameters for the Tukulu soil horizons
before optimization.
161
Table 7.2 Initial estimates of soil hydraulic parameters.
Parameters A B C
8s (mm mm") 0.34 0.33 0.32
8,.(mm mm") 0.13 0.12 0.26
a 0.00116 0.00094 0.005
n 1.773 1.617 1.226
Ks (mm hour") 36.1 40 9.6
7.2.4 Prediction of two-dimensional redistribution
A two-dimensional soil water movement during redistribution on irrigated furrows was
predicted using the HYDRUS-2D software (Simunek et al. 2009). This model numerically
solves the two dimensional form of the Richard flow equation and able to improve parameter
optimization by using the Levenberg-Marquardt non linear- squares approach (Marquardt,
1963). The HYDRUS-2D code has been successfully employed in simulating studies of
subsurface soil water distribution during furrow irrigation and redistribution (Abbasi et al.,
2004; Wohling and Maihol, 2007; Lazarovitch et al., 2009).
In this paper soil water movement during redistribution was inversely predicted using SWC
measurement versus time as the objective function. To improve the models predictions the
laboratory based SWCC hydraulic parameters had to be optimised (Abbasi et al., 2004). Any
of the parameters that were found to improve predictions was selected, but a combination of
more than three parameters was discouraged given the instability of the inverse optimization
method (Abbasi et al., 2003; Simunek et al., 2009). A soil profile flow domain of 2 000 mm
by 850 mm with the trapezoidal furrow shape at the symmetric position were used in this
exercise. The flow domain was described by a finite element mesh of 50 mm and
discretisized into three layer materials. Final pressure head from the irrigation phase was
exported as the initial pressure condition for the redistribution phase. The zero flux boundary
condition was assigned at the soil surface, inside the furrow, and along the vertical sides of
the flow domain. A free drainage boundary condition was selected for the bottom of the soil
profile. Three furrow distant sections were used for model predictions including the 5 m, 35
m and 55 m from the 20 L min-I inflow rate furrow, because the advance stream could not
reach the furrow end. In the 40, 80 and 160 L min-I inflow rate furrows the three distant
sections used were the 5 m, 55 m, and the 85 m.
162
7.2.5 Experimental data analysis
The measured and predicted SWC were used to characterise the redistribution process with
respect to depth and time. Firstly, the measure of dispersion between the measured (M) and
predicted (P) soil water content (8) from the various soil profile horizons of the three
respective furrow distant sections was determined. To this condition the following simple
mathematical expression was found to be convenient for this purpose:
(7.2)
Wherever, the predicted and measured variations of SWC are similar, this ratio being of the
order of one. Soil water content measurements at 150 mm, 450 mm, and 725 mm depths
allowed the profile to be discretised into two flow domains, 0-600 mm and 0-850 mm,
applicable to describe redistribution under different soil water regimes. The redistribution
flux across these flow domains was quantified using the unsaturated hydraulic conductivity
(K-coefficient) and the hydraulic gradient that operated between the upper nearly saturated
layer and the lower layer that has been less infiltrated. The inter block or effective K-
coefficient (Kefr) was thought to be appropriate for the Tukulu soil profile. Given the
assumption that furrow irrigation usually leaves the upper layer nearly saturated compared to
the lower unsaturated layers, the Darcian flux expression for redistribution from Gasto et al.,
(2002) was simplified to become:
hL -hU ]
qeff = Keff [-- + 1 (7.3)ZL-ZU
Where, qeffis the average flux between the upper (U) and lower (L) horizons, hl and hu and
Zl and Zu representing the corresponding matrix suctions and depths of the horizons in
questions. Unity value was assigned to the gravitational gradient (6.z/6.z) with the positive
sign for downward water flow. The horizons thickness was also considered to be large
enough for the redistribution process to be consistent with the principles of K eff. Before
irrigation and at the onset of redistribution it was assumed that profile layers not infiltrated to
near saturated are near residual SWC and that redistribution flux was a function of Kerf and
infiltrated depth. The geometric mean as described by Warriek (1991) was used to estimate
the effective SWC and K-coefficient for the 0-600 mm and 0-850 mm selected soil profile
flow domains.
163
The rate of distribution was therefore taken to be represented by the ratio of the derivatives of
the estimated Kerr and soil water content of the flow domain (0- 600 mm or 0-850 mm) of
interest. This analogy is fairl y similar to that of Yeh et al. (2005) used to describe the rate of
change of the first spatial moments in the z direction (Vz) written as:
ddKez = Vz(e) (7.4)
The resulting derivatives from the three furrow distant sections were then plotted against the
corresponding measured infiltrated depth to obtain a linear relationship between
redistribution and infiltrated depth. This function could be been extended, if adequate data
was available, to determine the moisture diffusivity length (A) described by Yeh et al., (2005)
as follows:
D(e)
A(e) = dK(8) (7.5)
CTë"""""
7.2.6 Statistical analysis
The quality fit between the measured and estimated SWC during the redistribution period
was evaluated using the 1:1 line. The coefficient of determination (R2), root mean square
error (RMSE) and the index of agreement or D-index as proposed by Willmot et al. (1985)
were also used. The Bartiett's test (Gomez and Gomez, 1984) was preferred for evaluating
the homogeneity of the error variances between K-coefficients calculated from averages of
measured and predicted SWC
7.3 Results
7.3.1 Parameter optimization and HYDRUS-2D inverse solution
Table 7.3 shows a summary of the various parameters optimised at various points of the
furrow that received different inflow rates. Statistical measure of quality of fit was also
shown. A comparison between the measured and predicted SWC was drawn on al: 1 line
presented in Figure 7.3 along with the coefficient of determination (R2). The parameter most
sensitive to the optimisation procedure was the es followed by the er and n, with the a and Ks
being the least.
164
Table 7.3 Optimised parameters to improve HYDRUS-2D model prediction.
Hydraulic 12arameters
nflow rate Distance
Lrnrn') (m) Horizons es er a n Ks RMSE D-index
20 5 A 0.331 - - 1.522 - 0.008 0.905
B 0.319 - - - - - -
C 0.313 - - 1.297 - - -
35 A 0.331 - - 1.522 - 0.014 0.853
B 0.319 - - - - - -
C 0.313 - - 1.297 - - -
55 A 0.160 - - - 0.013 0.931
B 0.206 - - - - -
C 0.199 - - - - -
40 5 A 0.327 0.271 - - 139 0.012 0.57
B 0.329 0.303 - - 266 - -
C 0.330 0.267 - - 4 - -
55 A 0.346 - 0.001 - - 0.027 0.565
B 0.349 - 1.222 - - - -
C 0.294 - 0.005 - - - -
85 A 0.275 - 0.001 - - 0.022 0.844
B 0.381 - 0.0004 - - - -
C 0.276 - 0.012 - - - -
80 5 A 0.337 - - 1.570 - 0.025 0.295
B 0.333 - - 1.290 - - -
C 0.306 - - 1.342 - - -
55 A 0.340 - - 1.525 - 0.022 0.800
B 0.343 - - 1.223 - - -
C 0.274 - - 1.182 - - -
85 A 0.324 0.241 - - - 0.008 0.957
B 0.286 0.259 - - - - -
C 0.323 0.318 - - - - -
160 5 A 0.328 0.260 - - - 0.007 0.925
B 0.278 0.261 - - - - -
C 0.307 0.261 - - - - -
55 A 0.326 0.225 - - - 0.012 0.922
B 0.280 0.240 - - - - -
C 0.332 0.318 - - - - -
85 A 0.340 0.181 - - - 0.019 0.788
B 0.330 0.167 - - - - -
C 0.320 0.308 - - - - -
8s = Saturated volumetric water content; 8r= Residual volumetric water content;a = pore
shape or geometry parameter; n = pore size distribution parameter; Ks = Saturated hydraulic
conductivity; RMSE = root mean square error; and D-index= index of agreement
165
20 L min-I 40L min-I 80 L min-I 160L min-I
E 0.4 ...------------"71 0.4 .,...---------71 0.4 ..,..------------",
E
Sm 0.4 r_--------~E Sm Sm
S 0.3 0.3 0.3
Sm
0.3
u:s 0.2 0.2 0.2
C/l 0.2
"0
0,) 0.1 R2 = 0.77 0.1 R2 = 0.80 0.1 R2 = 0.66.~ 0.1 R 2= 0.58
"0
0,)
'- o ~-_r-~--r_-~
0.. o o :-' --.--...------r-_J 0..jL----r"-----r--~-__I0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4 o 0.1 0.2 0.3 0.4 o 0.1 0.2 0.3 0.4
E 0.4 .,...---------/1 0.4 .,...---------___.. 0.4 ..,----------------,.. 0.4 ..,------------""?I
E 35 m 55 m 55 m 55 m
E 0.3 0.3 0.3
0.3
S
u:s 0.2 0.2 0.2 0.2
C/l
"0 R2 = 0.68 0.1 0.1
0,) R2= 0.57 0.1 R2 = 0.63 R2= 0.56u 0.1
"0
0,) o ~' O-;<'----,----,------r-----I o ~--,---_r----r_--~ o ~--_.----._--_.--____I0': --r---.--...---_j
o o 0.1 0.2 0.3 0.4 o 0.1 0.2 0.3 0.4 o 0.1 0.2 0.3 0.40.1 0.2 0.3 0.4
0.4 .,...---------~
'E 0.4 .,...---------'" 0.4 .,...-----------"71 0.4 .,..--------------:>1
E 0.3 55 m 85 m 85 m
E 0.3 0.3 0.3 85 m
E
u'--" 0.2:s 0.2 0.2 0.2
C/l 0.1
"0 R2 = 0.77 0.1
0,) R2 = 0.83 0.1 R2 = 0.88 0.1 R2 = 0.52
.~
"0 o :-'----.-----.-----.--_j o ~--r-~--r_-~ o ~,~-_.--._--_r__-__J o ~---,----_.----r_--~
0,)
0'.-. o 0.1 0.2 0.3 0.4 o 0.1 0.2 0.3 0.4 o 0.1 0.2 0.3 0.4 o 0.1 0.2 0.3 0.4
Measured SWC (mm mm')
Figure 7.3 Measured and predicted soil water content (SWC) of the soil profile during redistribution from the 20, 40,80 and 160 L rnin-I inflow
rates at various furrow distances covered by the advance stream.
166
Optimised 8s were highly variable among the inlet, midpoint and furrow end with the
exception in the 20 L min"1 inflow rate furrow where values were similar for both 5 and 35 m
distances. Optimised 8s values that were similar to the initial estimates were only observed in
the 160 L min"1 inflow rate furrow at the 85 m distance point. Comparable to the initial
estimates were noticed among the A- and B-horizons, especially in the 80 L min "I inflow rate
furrow. Only in the 160 L min"1 inflow rate furrow was the 8r optirnised at all three distance
points and all cases showing greater variability with the exception in the 5 m point where
almost the same value was assigned to the soil horizons. Similarly, the n assumed different
values along the furrow length and in depth that were not comparable with the initial
estimates with the exception in the C-horizon of the 55 m point from the 80 L min"1 inflow
rate furrow. The a and Ks were optimised once and it was in the 40 L min"1 inflow rate
furrow at the 85 m and 5 m points, respectively. The a value showed similarities with the
initial estimates especially in the A and C-horizons while the Ks values were not comparable
at all depths.
The quality of fit from the optimised parameters when the HYDRUS-2D was initialized
exhibited variability along the furrow distance points from the four inflow rate treatments.
The R2 ranged from 0.57 to 0.77, 0.68 to 83, 0.63 to 0.88 and 0.52 to 0.58 for the 20, 40, 80
and 160 L min"1 inflow rate furrows, respectively. Corresponding RMSE and D-index ranged
from 0.013 to 0.008 and 0.85 to 0.93, 0.027 to 0.012 and 0.57 to 0.84, 0.25 to 0.008,0.295 to
0.96,0.019 to 0.007 and 0.788 to 0.925, respectively.
7.3.2 Changes in soil water content during redistribution
The changes in SWC with depth from the inlet, mid and furrow end from the different inflow
rates at different times of redistribution were summarized in Figure 7.4. Results are given by
means ofaXY one dimensional curves to provide a simply comparison between the
measured and predicted SWC. Table 7.4 illustrates the measure of dispersion between the
measured and predicted SWC at the different horizon layers from the four inflow rate
treatments. For simplicity the time intervals of 0, 45 and 455 hours from the onset of
redistribution were assumed to be applicable to all four inflow rate furrow treatments.
Changes in SWC were indicated by the prefix (-) if representing a loss in volumetric SWC
from the flow domain while the prefix (+) represented a gain.
167
SWC (mm mrn') SWC (mm mm:') SWC (mm mm')
0.18 0.22 0.26 0.30 0.34 0.18 0.22 0.26 0.3 0.34 0.18 0.22 0.26 0.3 0.34
o o o
100 100 lOO
E 200 200 200
..E_, 300 Sm 300 35 m 300.p.t.:. 400 400 400
il)
Cl 500 500 500
600 600 600
700 700 700
800 800 ..L.._ ____ 800 ..L_ -l.J..._ __'
0.18 0.22 0.26 0.3 0.34 0.18 0.22 0.26 0.3 0.34 0.18 0.22 0.26 0.3 0.34
o o o
100 40L min-I 100 100
E 200 200 200E 300 Sm 300 55 m 300
:;; 400
p.. 400 400
il) 500 500 500
Cl 600 600 600
700 700 700
800 _.__--------' 800 .J..._ -l 800 .J..._ .-l
0.18 0.22 0.26 0.3 0.34 0.18 0.22 0.26 0.30 0.34 0.18 0.22 0.26 0.3 0.34
o o o
100 80L mi,n-I 100 100\
,.E-.._200
200 200
E 300 Sm 300
300
'-' 55 m 85 m
-5 400 400 400
0..
CS 500 500 500
600 600 600
700 700 700
800 -'-- -...J 800 _.__------ 800 ..L_ _-l
0.18 0.22 0.26 0.3 0.34 0.18 0.22 0.26 0.3 0.34 0.18 0.22 0.26 0.3 0.34
o o o
100 160 L min-I 100 lOO
,.-.._200 200 200
E8 300 5 m 300 55 m 300 85 m
:;; 400 400 400
p.. 500 500 500
il)
Cl 600 600 600
700 700 700
800 _.__ -l 800 800 ..L_ _...J
~ M: initial -- P:initial
---!r-- M:47 hrs ------ P:47 hrs
- o£)- M:4SS hrs - - - P:4SS hrs
Figure 7.4 Measured and predicted soil water content (SWC) for the initial and selected time
intervals during redistribution at different furrow distances and soil profile depths for the 20,
40,80 and 160 L min-I inflow rates.
168
Along the 20 L min-I inflow rate furrow, changes between the measured and predicted SWC
during the 455 hour redistribution period were variable in space and time. Changes in
volumetric SWC, with respect to layers thickness of 300 mm for the A and B-horizons and
250 mm for the C-horizons, were -17, -9 and -4 mm for the respective A, Band C-horizons at
the 5 m furrow section. Predicted volumetric SWC for the corresponding horizons was -16,
+2 and -3 mm. Dispersion between the measured and predicted SWC was 1.0, 3.9 and 0.8
from the respective A, Band C-horizons. At the 35 m furrow section changes in measured
volumetric SWC were -19,-4 and +3 mm while from the predicted was -24, -1 and -5 mm,
from the respective A, Band C-horizons. At the 55 m furrow section, measured volumetric
SWC changed by -25, +6, and +1 while the predicted changed by -31, 0 and -4 mm, from the
respective A, Band C-horizons. Corresponding dispersion between measured and predicted
SWC at 35 and 55 m furrow sections for the profile horizons ranged from 0.3 to 1.6 and 0 to
2.6, respectively.
In the 40 L min-I inflow rate furrow, after the 455 hour of redistribution measured volumetric
SWC at the 5 m furrow section had changed by -7 mm from the A and B-horizons and -1 mm
from the C-horizon. The predicted volumetric SWC recorded changes of -7, -4 and -3 mm
from the respective A, Band C-horizon. Corresponding changes from the measured and
predicted volumetric SWC at the 35 furrow section were -18, -3 and -1 mm and -33, + Il and
-8 mm, respectively. At the 85 m furrow section changes from the measured and predicted
volumetric SWC amounting to -2.5, +9, and -8 mm and -22, +14 and -8.3 mm, respectively
were observed corresponding to the A, Band C-horizons. Measure of dispersion between the
measured and predicted SWC ranged from 0.6 to 2.0, 1.9 to 7.1, and 1.5 to 25.2 from the 5,
55 and 85 m soil profile horizons, respectively.
From the 80 L min-I inflow rate furrow, noticeable changes in volumetric SWC after the 455
hour redistribution at the 5 m furrow section were amounting to -10, -0.24 and -0.23 from the
measured while from the predicted were -28, +1.5 and -8.3 mm from the respective A, Band
C-horizons. After 47 hours gains of 1.5 and 1 mm from the measured SWC were noticed
from the Band C-horizons, respectively. At the 55 m furrow section at the of the
redistribution period the volumetric SWC has changed by -26, -0.4 and +2.2 mm from the
measured while the predicted by -32, +2 and 8 mm, from the respective A, Band C-horizons.
After 47 hours of redistributions the measured volumetric SWC had recorded changes of -l l ,
+3 and +5 mm from the corresponding profile horizons.
169
Table 7.4 Measure of dispersion between the measured and predicted soil water content (mm
mm-I).
Horizons
Inflow rate Distance (m) A B C
20 5 0.98 3.90 0.75
35 1.31 0.31 1.55
55 1.22 0.00 2.60
40 5 0.91 0.64 1.95
55 1.89 3.81 7.05
85 8.80 1.48 25.15
80 5 2.67 6.17 36.28
55 1.22 4.36 3.55
85 1.03 4.72 24.41
160 5 0.85 0.61 0.49
55 0.95 0.78 1.64
85 1.49 0.19 16.67
In the 85 m furrow point, the measured and predicted volumetric SWC at the 455 hour of
redistribution recorded changes amounting to -13, -0.13 and +0.1 mm, and -13, +0.6, and -3
mm, corresponding to the A, Band C-horizons, respectively. Measure of dispersion at the 5,
55 and 85 m furrow sections for the profile horizons ranged from 2.7 to 36.3, 1.2 to 4.4, and
1.0 to 24.4, respectively.
Noticeable changes in volumetric SWC after the 455 hours of redistribution from the 160 L
min-I inflow rate furrow were 14, +7, +2 mm, and 12, +5 and 1 mm from the measured and
predicted at the 5 m furrow section. At 47 hours the measured volumetric SWC recorded -6,
+5 and +2 while the predicted -10, +4 and -1 mm, for the respective A, Band C-horizons. At
the 55 m distance, by the end of the redistribution period the measured volumetric SWC
recorded changes of -15 mm in the A-horizon and +2 mm from the Band C-horizons.
Changes from the predicted values were -14, -2 and +3 mm from the respective A, Band C-
horizons. After 47 hours the measured volumetric SWC recorded -5, +0.3 and +3.4 mm from
the corresponding A, Band C horizons. At the 85 m furrow section, the measured volumetric
SWC had changed by -14, -5 and +1 mm from the respective A, Band C- horizons after the
455 hours of redistribution. The predicted had changed by -21, -1 and -17 mm from the
corresponding horizons. Dispersion between the measured and predicted SWC ranged from
0.5 to 0.9, 0.8 to 1.6 and 0.2 to 16.7 for the 5, 55 and 85 m furrow sections among the soil
profiles horizons.
170
7.3.3 Estimated effective K-coefficient during redistribution
Figures 7.5 to 7.8 illustrates the variation in the estimated K-coefficient from the measured
and predicted SWC for the 20, 40, 80 and 160 Lmin-I inflow rate treated furrows.
Redistribution at soil profile depths ranging from 0-600 and 0-850 mm were characterised
using average K-coefficient (Kerf). Homogeneity test calculated from measured and predicted
SWC were illustrated in Table 7.5.
The calculated Kerr from the measured and model prediction values showed considerable
variations along the 20 L min-I inflow rate furrow. At the 5 m furrow section it varied from
0.7 to 0.023 mm hour" and 25 to 4.4 mm hour" from the respective measured and predicted
values for the 0-600 mm depth. At 0-850 mm depth estimates from the measured values
ranged from 0.6 to 0.006 while from model predictions the Keff ranged from 6 to 0.6 mm
hour". At the 35 m furrow section from the 0-600 mm depth, the Keff values ranges from 0.2
to 0.009 mm mm-I and 0.6 to 0.06 mm hour' were observed from the measured and model
predictions, respectively. Variations at the 0-850 mm depth ranged from 0.05 to 0.02 mm
hour" from the measured and 2.4 to 0.06 mm hour-I form models predictions. At the 0-600
and 0-850 mm depths from the 55 m point the estimated Keff ranged from 0.15 to 0.005 mm
hour" and 0.04 to 0.001 mm hour" from the measured and from 0.31 to 0.002 mm hour-I and
0.02 to 0.0002 mm hour", consecutively. The Kefr estimated from measured and model
predictions was found to homogeneous for the 0-850 mm profile domain at 35 and 55 m
furrow sections.
Estimated Keff from the 40 L min-I inflow rate furrow varied between 0.13 to 0.01 mm hour"
and 0.63 to 8.3E-07 mm hour" at the 5 m section from the measured and predicted values
within the 0-600 mm depth. At the 0-850 mm depth it ranged from 0.33 to 0.007 mm hour"
from the measured and 1.41 to 0.0004 mm hour" from the model predictions. At the 55 and
85 m furrow sections, the estimated Keff from measured values within the 0-600 mm depth
ranged from 1.4 to 0.004 mm hour" and 0.002 to 0.00004 mm hour"; respectively while in
the 0-850 mm depth from 1.4 to 0.004 mm hour" and 0.001 to 0.00003 mm hour".
Corresponding to the consecutive furrow sections from model predictions was the Keff
ranging from 1.31 to 0.004 mm hour-I and 0.9 to 0.04 mm hour-I at 0-600 mm depth while at
the 0-850 mm depth it ranged from and 3.5 to 0.0003 mm hour" and 2.9 to 0.46 mm hour",
respectively.
171
lOO
10 5m ----------.:- :::-lilIJ
\.... 0.1 It.C1.
o::s [jJ0 t.t.
..s:::: 0.01
E I.~
ED.OOI
'-..../
~'t 0.0001
lE-OS
IE-06
100
ID 35 m ---~-- _--_---0.1\:.:..s. 0 d>t.o
..s:::: 0.01 oo
ê 0.001
'-..../
't 0.0001
~
lE-OS
IE-06
100 55 m
ID
\:.:..s.
, i
_g 0.1 , 00 _____ . I
ê ~.& -0.01 cjP---- ~----
~ 0.001 1$' ..
0)
~ 0.0001 --
lE-OS
IE-06
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34
Soil water content (mm mm')
o Measured: 0-600 mm t. Measured: 0-850 mm
-- Predicted:O- 600 mm Predicted: 0-850 mm
Figure 7.5 Effective K-coefficient from the furrow treated with 20L min-I inflow rate for the
0-600 and 0-850 mm infiltrated depths at the 5 m, 35 m and 55 m distances
172
100
10 !5m
0.1 lt~
\..
::l
-0= 0.01
[PJ ~ "
E 0.001
E
'-'
::: 0.0001
~" lE-OS
IE-06
100
10 55 m
,....._ -- IJ
\..
-g= 0.1 ~~ aD
~ 0.01 -----: @6. [jP
~ 0.001
~" I " 6 66
0.0001
lE-OS
IE-06
100 85 m
,....._ 10
\.. - -, --
::l ----
-0= ,0.1 ~- ,
E
0.01 ----- .E
'-'
:::
c 0.001 6~
~ ~li0.0001 ,:lih ~}
lE-OS , ,, I
IE-06 , ,
! i I
I
0.22 0.24 0.26 0.28 0.3 0.32 0.34
Soil water content (mm mm')
° Measured: 0-850 mm 6 Measured: 0-600 mm
-- Predicted: 0-600 mm - - - Predicted: 0-850 mm
Figure 7.6 Effective K-coefficient from the 40 L min-I inflow rate at 0-600 and 0-850 mm
infiltrated depths of the 5 m, 55 ill and 85 ill furrow distances.
173
100 ~--------------------------------------------------------
10
.... Sm
:o:l
...c:
E 0.1
'-E" 0.0 I
~"'f; 0.001
0.0001
lE-OS
IE-06 +---------~------~.---------~---------------------------~
100
10 55 m
...--,
l- l
::l
0
...c: 0.1
E
E 0.01
'-"
i::::
" 0.001~
0.0001
lE-OS
o o
o
1E-06
100
I- 10 85 m
o::l
...c:
E 0.1
'-E" 0.0 I
~"'f; 0.001
0.0001
lE-OS
IE-06 +-------~--------~~------~--------~------~.~------~
0.22 0.24 0.26 0.28 0.3 0.32 0.34
Soil water content (mm mrrr')
o Measured: 0-600 mm l:;. Measured: 0-850 mm
-- Predicted: 0-600 mm - - - Predicted:0-8S0 mm
Figure 7_7 Effective K-coefficient from the 80 L min-) inflow rate at 0-600 and 0-850 mm
infiltrated depths of the 5 m, 55 m and 85 m furrow distances.
174
100
~ 10 5m
...
::l
0
.J: 0.1
E
E 0.01 ~f
i::
c 0.001
~ ll~
0.0001 ~l!.
I
lE-OS I
I
lE-06 I
100
~ 10
\... 55 m
::l
0
.J: ~:
E 0.1
E 0.01 L~ 0
'-' 6[jJ
i::
~" 0.001 ~& »»
0.0001 ,.... .... ....
....
lE-OS ....
lE-06 ,
100
0
~ 10 -r85 m _______-:;:.
\... I
::l i ---- 0 ,,,'0
.J: 0.1 ~ in,,
E ~D '0.01 [JJ , ..... :E ",, i
'-' ,
i:: 0.001 JI»..,'
~ '" 0.0001 ___~.t:#'
lE-OS I
j
lE-06 i I
0.22 0.24 0.26 0.28 0.3 0.32 0.34
Soil water content (mm mm')
o Measured: 0-600 mm ~ Measured: 0-850 mm
-- Predicted: 0-600 mm - - - Predicted: 0-850 mm
Figure 7.8 Effective K-coefficient from the 160 L min-' inflow rate at 0-600 and 0-850 mm
infiltrated depths of the 5 m, 55 m and 85 m furrow distances
175
Table 7.5 Homogeneity test for the effective K-coefficient calculated from measured and
predicted soil water contents.
Bartiett's test anal ysis
Inflow
rate Distance Profile depth Computed Tabular
(L min") (m) (M) X2 x2: (0.05,2)
20 5 0-600 28.16 3.84
0-850 21.07 3.84
35 0-600 9.53 3.84
0-850 0.41 3.84
55 0-600 9.96 3.84
0-850 0.02 3.84
40 5 0-600 2.10 3.84
0-850 34.00 3.84
55 0-600 46.70 3.84
0-850 3.44 3.84
85 0-600 0.02 3.84
0-850 2.37 3.84
80 5 0-600 0.59 3.84
0-850 1.55 3.84
55 0-600 8.68 3.84
0-850 0.50 3.84
85 0-600 24.59 3.84
0-850 5.83 3.84
160 5 0-600 10.31 3.84
0-850 0.0003 3.84
55 0-600 29.28 3.84
0-850 0.00004 3.84
85 0-600 106.34 3.84
0-850 4.11 3.84
Bold = passed homogeneity test at 0.05 X2 Test
At 5m and 55m furrow length sections the estimated Keff passed the homogeneity test at the
0-600 mm and 0-850 mm soil profile flow domains, respectively. At the 85 m furrow length
section the measured and estimated Keff were found to be homogeneous at both 0-600 mm
and 0-850 mm profile flow domains.
Substantial variations in estimated Kerr were also observed in the 80 L min") inflow rate
furrow. At the 5 m furrow section at depths of 0-600 mm and 0-850 mm the measured values
ranged from 0.06 to 0.01 mm hour") and 0.04 to 0.008 mm hour', respectively. Estimates
from the models predictions ranged from 1.1 to 0.013 and 3.25 to 0.004 mm hour") for the
respective consecutive depths. The Ksrr at the 55 m furrow section from the 0-600 and 0-850
176
mm depths varied from 0.17 to 2.9 x 10.6 mm hour" and 0.05 to 0.004 mm hour" from the
measured values while from the model predictions it ranged from 0.46 to 0.002 mm hour-I
and 2.1 to 0.15 mm hour', respectively. In the 85 m furrow section the Keff within 0-600 mm
depth fell in the range of 0.35 to 0.02 mm hour' and 38 and 0.3 mm hour" for the measured
and models predictions at 0-850 mm depth was in the range of 0.3 to 0.02 and 0.0002 to
1.73E-05 for the consecutive soil profiles depths. The estimated Kerr for measured and model
predictions passed the homogeneity test at 0-600 mm and 0-850 mm profiles flow domains at
the 5 m furrow section and at the 0-850 mm profile domain for the 55 m section.
Variable spatiality was also noticeable at the 160 L min-I inflow rate furrow. The Keff at the 5
m section was in the range of 0.11 to 0.01 mm hour-I and 0.004 to 0.0001 mm hour-I for the
measured values from the respective 0-600 mm and 0-850 mm depths. Corresponding values
from model predictions for the respective consecutive depths ranged from 2.73 to 0.003 and
0.00003 to 1.7E-07 mm hour". At the 55 m and 85 m sections the estimated Kerr from the
measured values varied from 0.4 to 0.0003 mm hour" and 38 to 0.006 mm hour-I for the 0-
600 mm depth while for the 0-850 mm it ranged from 0.004 to 0.0003 and 0.09 to 0.0001 mm
hour", respectively. The predictions of Kefr for the respective consecutive furrow sections
ranged from 38 to 0.09 mm hour" and 16.4 to 0.5 mm hour" at the 0-600 mm depth and
0.0002 to 9.6E-06 mm hour-I and 12.5 to 7.02E-05 mm hour-I at the 0-850 mm depth.
However, the estimated Keff from measured and predicted values passed the homogeneity test
at depths of 0-850 mm from the 5 m and 55 m furrow sections.
7.3.4 The rate of soil water redistribution
In this paper the rate of redistribution is represented by the ratio of dKdefef in relation to
infiltrated depth (mm) as illustrated in Figure 8.9. For simplicity only the K-coefficient
calculated from the measured SWC were used, and the semi log linear regression functions
for vertical redistribution (Vz) and the corresponding coefficient of determinations (R2) were
summarized in Table 7.6.
A small linear variation between the rate of redistribution and infiltrated could be observed.
Infiltrated depths of 82 to 46 mm and 90 to 60 mm in the 20 L min-I inflow rate furrow
produced Vz that varied from 0.03 to 0.008 and 0.02 to 0.003 mm hour -I hours, respectively.
From the 40 L min-I inflow rate furrow, infiltrated depths of 93 to 36 mm and 102 to 37 mm
177
respectively, produced Yz that ranged between 0.07 to 0.001 and 0.06 to 0.004 mm hour",
respectively.
Table 7.6 Rate of vertical redistribution (Vz) expressed as a function of infiltrated depth (8z).
Inflow rate( L mm") DeJ2th(mm) Regression R2
20 0-600 Log Yz = 0.015 (8z)- 2.866 0.91
0-850 Log Yz = 0.028 (8z)- 4.368 0.77
40 0-600 Log Yz = 0.031(8z)- 4.375 0.95
0-850 Log Yz = 0.027(8z)- 4.484 0.80
80 0-600 Log Yz = 0.032(8z) - 4.208 0.84
0-850 Log Yz = 0.011(8z) - 2.827 0.99
160 0-600 Log Yz = 0.061(8z) - 5.626 0.65
0-850 Log Yz = 0.048(8z) - 6.682 0.80
In the 80 L min -I inflow rate furrow, infiltrated depths of 78 to 57 and 81 to 57 mm were
attained with corresponding Yz of 0.03 to 0.005 and 0.01 to 0.007 mm hour -I, respectively.
The 160 L min-I inflow rate furrow recorded infiltrated depths of 87 to 59 mm and 88 to 59
mm with respective Yz with range of 2 to 0.02 and 0.01 to 0.0002 mm hour". The R2 of the
Yz regression functions was within the range of 0.99 to 0.65 from the furrow sections
observed.
7.4 Discussion
Characterising soil water movement during redistribution in soils could be a difficult exercise
because of the interactions of gravity and matrix suction gradients across a flow domain that
could exhibit different physical properties. Monitoring changes in subsurface SWC and the
use of the HYDRUS-2D model to predict changes in SWC along short furrows provided
insight on how the layering of the horizons in the Tukulu soil affected redistribution.
Critical to the improvement of the model's prediction was parameter optimisation under field
scale conditions. Although no strict format was followed in selecting the hydraulic
parameters for optimization other than being cautious of not exceeding three at a time for
concerns of uniqueness (Hopmans et al., 2002; Simunek et al., 2009). The 8s and 8r were
among the most frequently optimised parameters irrespective of depth and inflow rate furrow
regime.
178
20 L min-I,
0.1
a:>
"0
""tt: 0.01 _--n---- ----......Q --/"u _--
~ '"
"0 4--------
... '" ... -
0.001
0.0001
40 L min-I
a:> 0.1 o t:.
~ ______,:;-~ - - - - - - - -
~'" 0.01 ~~------~--- t:.
"0
0.001 ~. - - - - -- - -- - -- - - - - - - _- - --
0.0001 +------.------~-------._------r------,_------l------_,----~
1
80 L min-I
0.1
a:>
~ 0.01
~'"
"0 0.001
0.0001 -I--------,-------,-------....--------T---------.---------.-------,------I
10
160 L min-I
1
a:>
"0 0.1
""tt: - :. ----0-----o
~'" 0.01 ~ :
"0 - -- - -- - --
t:.
0.001 _- _ - -- - -- t:.
s->::
0.0001 !
40 I30 60 70 80 90 100 110
Infiltrated depth (mm)
00-600 mm t:.0-8S0 mm
Figure 7.9 Relationship between the rate of redistribution and infiltrated depth at 0-600 and
0-850 mm profile depth for the 20, 40,80 and 160 L min-I inflow rates.
179
Although the 8s were fairly comparable with field and laboratory measured porosity of this
soil (Chimungu, 2009), the 8r exhibited remarkable inconsistencies that could be partly be
due soil mixing of the upper horizons with the C-horizons clay-rich soil during deep
ploughing or ripping operations. The general differences among parameters within the profile
and along the furrows were to be expected given the considerable spatial variability in this
soil. An explanation could be given for this but it was interesting to notice that the
optimization of a and n parameters at the same time either reduced the models prediction or
led to failure of the inverse solution to converge.
Agreement between the measured and predicted SWC was fairly satisfactory with the R2
varying from 0.88 to 0.52 of the regression of the measured versus the predicted. In most
cases the model's predictions overestimated the SWC, especially from the upper horizons.
This was to be expected because model's predictions do not consider entrapped or dissolved
air which lowers 8s under in situ conditions. Pronounced dispersion between observed and
predicted SWC at near the soil surface could be due to evaporation that was not accounted for
by the model's prediction while dispersion at deeper layers could be attributed to the model's
redistribution diffusive flux (Wohling and Mailhol. 2007).
Despite some noticeable discrepancies between the measured and predicted SWC, the
changes in SWC, which are an integral part of field experiments, on the XY line were
reflective of the redistribution process. The upper layers demonstrated a remarkable soil
water release to the lower horizons even though substantial levels of SWC were retained far
above the 8r as a result of hysteresis. However, the recorded gains from the lower layers were
less than the losses from the upper layers suggesting that SWC measurements from the single
neutron access tube could not capture the complete redistribution pattern across the furrow
section domain (Barger et al., 1999). Despite calibration of the neutron water meter at 150
mm depth from the soil surface the general lack of precision of this instrument near the soil
surface could have also contributed to the discrepancies in SWC measurements.
Nevertheless, the change in the volumetric SWC with time and space within the flow domain
was able show the variations in soil water storage in a meaningful and similar manner to that
of the zeroth moment described by Yeh et al. (2005) and Lazarovitch et al. (2009). Changes
in soil water storage were greatest in the orthic A-horizon and least in the underlying C-
horizon, irrespective of the inflow rate. Although spatial differences in physical and hydraulic
properties were a contributing factor, the infiltrated depth from this horizons and subsequent
hysteresis were the primary reasons. Noticeable changes in SWC at the C-horizons were
180
observed where deep infiltration was realised especially at the furrow inlet of the 20 and 40 L
min-I inflow rates and at the furrow end of the 80 and 160 L min-I furrows. Under these
circumstances it is reasonable to suggest that deep infiltration augmented gravity driven
redistribution a phenomenon that was also noted by Hillel (2004).
The actual permeability of the infiltrated domain during soil water redistribution was
illustrated by the variability of the Kerr. Overestimations of Kerr were to be expected given the
initial overestimates of SWC from model predictions, especially at the furrow inlets and ends.
For the upper 0-600 mm soil profile flow domain Kerr were generally higher from both
measured and predicted values indicating that this section of the profile was generally deeply
wetted compared to the 0-850 mm domain. The loose structure and the smooth transition
between of the A and B horizons supported the higher permeability is this domain. At the 0-
850 mm the Kerr exhibited a steeper gradient in most furrow sections suggesting that the
prismacutanic C-horizon at 600 to 850 mm depth could have played a major role in
augmenting strong matrix suction against gravity driven redistribution. Although spatial
variability was the reason for the inconsistencies in Kerr slope between the 0-600 and 0-850
mm profiles flow domains, a general linear relationship could be drawn between Ksrr and the
respective average SWc. Such linearity was also demonstrated by the Kerr values between 0-
600 and 0-850 mm profile domain at each furrow section. In the former the Kefr fell in the
range of 10 to 0.001 mm hour" while in the latter it was between the ranges of 1 to 0.0001
mm hour". This was not surprising since the permeability of the infiltrated depth decreased
with increasing profile because of the increase in water gradient and matrix suction at the
lower horizons of this soil. Inferences to the linearity between Kerr and infiltrated depth were
also made in other others including the work of Rubin (1967) and Yeh et al. (2004).
In this work the linear relationship between Kerr and infiltrated depth was extended to
determine the rate of redistribution (Vz) within a specified profile flow domain (0-600 or 0-
850 mm depth). This analogy found its validity on the convective term representing the rate
of change of the first spatial moments in the downward (z) direction as described by Yeh et
al. (2005). In this soil an infiltrated depth of 100 mm would be expected to augment a
redistribution rate of 0.1 to 0.01 mm hour". An infiltrated depth of 50 mm of SWC would fall
in range 0.01 to 0.001 mm hour" within the profile depth of 600 mm and 0.01 to 0.0001 mm
hour" for the 850 mm depth. For a soil with a restrictive underlying horizon, these rates of
redistribution indicated minimum drainage activity that is consistent with findings of and
Fraenkel (2008) and Chimungu (2009). Therefore, the low rates of gravity augmented
181
redistribution could be said to offer greater flexibility not only to control the frequency of
irrigation but also the desired infiltrated depth.
7.5 Conclusion
Soil water redistribution on the South African Tukulu, also referred as Cutanic Luvisols of
the World Reference Group, was characterised along 90 m distant furrows. A single-run of
flood irrigation was used for the subsurface soil water redistribution trial under different
inflow rate treatments. The predicted soil water content distribution predicted from the
parameter optimization of HYDRUS-2D matched fairly well with measured values. The
results provide support to use the HYDRUS-2D as a tool for designing and evaluating the
efficiency of furrow irrigation management systems. Despite discrepancies resulting from
field spatial variability the use of the geometric mean scheme to calculate the Keff provided
reasonable estimates of the average permeability of the different furrow profile sections
during redistribution. Vertical redistribution was estimated using the relationship between the
ratio d::ff and infiltrated depth and the results were able to discriminate redistribution at 0-
600 mm from that of 0-850 mm profile depths. The rate of redistribution in a deeply
infiltrated profile at depth up to 850 mm was found to be less than one order of magnitude
when compared to partially infiltrated profiles up to depths of 500 mm. This result is
consistent with the physical properties of the soil horizons layers of the Tukulu. The upper
strata is course structured that encourages deep vertical redistribution while the underlying
prismatic C-horizon is of low permeability and restricts percolation below the soil profile.
Thus furrow irrigation practices can be developed in the Tukulu soil type and Cambisols,
with similar layering composition, with confidence that deep subsurface redistribution shall
support soil water storage and minimum deep drainage losses.
Acknowledgements
Acknowledgements to the staff and management of Paradys experiment station of the
technical and hospitality they provided during this experiment. Special thanks to the Strategic
Academic Cluster for the water management in water-scarce areas of the University of the
Free State for the financial assistance in carrying out this experiment.
182
7.6 References
Abbasi, F., Jacques, D., Simunek, J. and van Genuchten, M. Th., 2003. Inverse estimation of
soil hydraulic and solute transport parameters from transient field experiments:
Heterogeneous soil. Am. Soc. OfAgric. Eng., 46: 1097-1111.
Aris, R., 1956. On the dispersion of a solute in a fluid flowing through a tube. Proc. R. Soc.
London, Ser., 235: 67-78.
Austin, c., 2003. Micro flood, a new way of applying water. http://waterright.com .•
22/0912011, 10.00 am (LT).
Baker, D. L., Arnold, M. E. and Scott, H. D., 1999. Some analytical and approximate Darcian
means. Ground Water, 37: 532-538.
Bargar, B., Swan, J. B. and Jaynes, D., 1999. Soil water recharge under un-cropped ridges
and furrows. Soil Sci. Soc. Am. J., 63: 1290-1299.
Barry, D. A, and Sposito, G., 1990. Three-dimensional statistical moment analysis of
Stanford/WaterlooBorden tracer data. Water Resour. Res., 26:1735-1747.
Brooks, R. H. and Corey, A. T., 1964. Hydraulic properties of porous media, Hydrology
paper no. 3. Civil Engineering Dep., Colarado State University, Fort Collins, USA.
Chimungu, J. G., 2009. Comparison of field and laboratory measured hydraulic properties of
selected diagnostic soil horizons. M.Sc. (Agric) Dissertation, University of the Free
State, Bloemfontein, South Africa.
Clemmens, A. J., Eisenhauer, D. E. and Maheshwari, B. L., 2001. Infiltration and Roughness
equations for surface irrigation: How form influences estimation, An ASAE Meeting
Presentation, Paper no. 01-2255, hg@asae.org: 08/08/2011, 15.30 pm (LT).
Fraenkel, C. H., 2008. Spatial variability of selected soil properties in and between map
Units. Msc (Agric) Dissertation, University of the Free State, Bloemfontein.
Gasto, J. M., Grifoll, J. and Cohen, Y., 2002. Estimation of internodal permeabilities for
numerical simulation of unsaturated flows. Water Resour. Res., 38: 1326-1340.
Gee, G. W., and Ward, A. L., 2001. Vadose zone transport field study: Status report, PNNL-
13679, Pac., Northwest National Library, Richland, Washington DC, USA.
Gomez, K. A., and Gomez, A.A., 1984. Statistical procedures for agricultural research, 2 ed.
John Wiley and Sons, New York, USA.
Hensley, M., Botha, J. J., Anderson, JJ ., van Staden, P. P. and du Toit, A., 2000. Optimising
rainfall use efficiency for developing farmers with limited access to irrigation water,
Water Research Commission report, 878, Pretoria, South Africa.
183
Hillel, D.H., 2004. Environmental soil physics, New York: Academic Press, USA.
Hopmans, J. W., Simunek, L, Romano, N. and Durner, W., 2002. Simultaneous
determination of water transmission and retention properties, Inverse Methods. IN:
Methods of Soil Analysis.Part 4.Physical Methods (J.H. Dane and G.c. Topp, Eds.),
Soil Science Society of America Book Series No. 5: 963-1008.
Kool, J. B., and Parker, J. C., 1988. Analysis of the inverse problem for transient unsaturated
flow, Water Resour. Res., 24: 817-830.
Kosugi, K., 1996. Lognormal distribution model for unsaturated soil hydraulic properties,
Water Resour. Res., 32: 2697-2703.
Lazarovitch, N., Warrick, A.W., Furman, A. and Simunek, J., 2009. Water content
distribution in drip irrigation described by moment analysis. Vadase Zone 1., 6: 116-
123.
Lazarovitch, N., Warrick, A.W., Furman, A. and Zerihum, D., 2009. Subsurface water
distribution from furrows described by moments analyses. 1. of Irrig. and Drain.
Eng., 135: 7-12.
Leconte, R., and Brissette, F. P., 2001. Soil moisture profile model for two-layered soil based
on sharp wetting front approach. J. of Hydrologic Eng., 6: 141-149.
Mathews, C. J., Knight, F. J. and Braddock, R. D., 2004. Using analytic solutions for
homogeneous soils to assess numerical solutions for layered soils. ARC Discovery
Indigenous Research Development Program, C.mathews@griffith.edu.au, 31/04/2011,
17.00 pm (LT).
Rubin, J., 1967. Numerical method for analyzing hysteresis-affected, post-infiltration
redistribution of soil moisture, Soil Sci. Soc. Am. Proc., 31: 13-20.
Schaap, M. G, Leij, F. J. and van Genuchten, M.Th., 2001. Rosetta: A computer program for
estimating soil hydraulic parameters with hierarchical pedotransfer functions. J.
Hydrol., 25: 1163-1176.
Scotter, D. R., Clothier, B. E. and Harper, E. R., 1982. Measuring saturated hydraulic
conductivity and sorptivity using twin rings. Australian Journal of Soil Research, 20:
295-304.
Simunek, J, Sejna, M. and van Genuchten, M. Th., 2009. The HYDRUS-2D software
package for simulating two-dimensional movement of water, heat and multiple
solutes in variably saturated media, Version 2.0. Rep. IGCWMC-TPS-53, p. 251, Int.
Ground Water Model, Cent., Colo. Sch. of Mines, Golden, CO.
184
Soil Classification Working Group, 1991. Soil Classification - A taxonomic system for
SouthAfrica, Memoirs on the Agricultural Natural Resources of South Africa No. 15.
Department of Agricultural Development, Pretoria.
Srivastava, R., and Yeh, T.C.J., 1991. Analytical solutions for one-dimensional transient
infiltration toward the water table in homogenous and layered soils. Water Resour.
Research, 27: 753-762.
Sudicky, E. A., 1986. A natural gradient experiment on solute transport in a sand aquifer:
Spatial variability of hydraulic conductivity and its role in the dispersion process,
Water Resour. Res., 22: 2069-2082.
van Genuchten, M.Th., 1980. A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils. Soil Sci. Soc. Am. J., 44: 892-898.
Warriek, A. W., 1991. Numerical approximations of Darcian flow through unsaturated soil.
Water Resour. Res., 27: 1215-1222.
Wildenschild, D., and Jensen, K. H., 1999. Numerical model1ing of observed effective flow
behaviour in unsaturated heterogeneous sands. Water Resour. Res., 35: 29-42.
Willmotti, C. 1., Ackleson, S. G., Davis, R. E., Feddema, J. J., Klink, K. M., Legates, D. R.,
Dennel, O. J. and Rowe, CM., 1985. Statistics for the Evaluation and Comparison of
Models. 1. of Geophysical Research, 90:8995-9005.
WohJing, T. H., and Maihol, J. C, 2007. A physically based coupled model for simulating ID
and surface -2D subsurface flow and plant water uptake in irrigation furrows, II:
model test and evaluation. 1. of Irrig. and Drain. Eng., 133: 543-553.
Wohling, T. H., Schrnitz, G. H. and Maihol, J. C., 2004. Modelling two-dimensional
infiltration from irrigation furrows. 1. Irrig. Drain. Eng., 130: 296-303.
World Reference Base for Soil Resources (1998). World soil resources report, 84, ISSS /
ISRIC FAO, Rome, Italy.
Yeh, T. CJ., Ye, M. and Khaleel, R., 2005. Estimation of effective unsaturated hydraulic
conductivity tensor using spatial moments of observed moisture plume. Water Resour.
Res., 41: W03014.
185
CHAPTERS
INTEGRA TING MICRO-FLOOD WITH IN-FIELD RAINWATER HARVESTING
:(i) SOIL WATER BALANCE PROCEDURE AND APPLICATION
Abstract
A procedure for estimating soil water balance (SWB) components for the newly integrated
micro-flood irrigation (MFI) with in-field rainwater harvesting (IRWH) was developed and
validated. Gains and losses from the dryland (DL), supplemental (SPI) and full irrigation (FI)
water regimes (WR) under IRWH with three (1 m, 2 m and 3 m) runoff strip widths (RSW)
were accounted for. Site specific data was used to estimate rainfall-runoff relationships and
deep drainage (DD). Plant canopy cover was considered during the estimation of evapo-
transpiration (ET), evaporation (Ev) and transpiration (T). Full, partial and no-shaded were
used describe certain sections of the experimental plots depending on the surface area under
canopy cover. Consistency of the procedure with soil hydrological and agronomic principles
led to its acceptance. It was then applied to determine the effect of the RSW and WR on the
soil water balance components over the production season. Net gains from rainfall were
comparably higher from the 2 m and 3 m RSW. The 1 m RSW had the least net gains from
rainfall but had consistently higher ET and T for all WR. Greater surface area covered by
crop canopy and accessible to plants roots from both twin and alternative rows was the
possible explanation. Irrigation increased ET, Ev and T irrespective of the regime.
Evaporation also increased with RSW irrespective of the WR. Deep drainage was only
affected by RSW and WR following major rainfalls (>24 mm) and irrigation especially late
in the season. Interactions were limited to change in soil water content, DD and T mainly
after the occurrence of rainfalls at short intervals or after an irrigation event. It was therefore
concluded that RSW and WR affected the SWB components and that the 1 m RSW offered
an optimum surface area for substituting evaporation losses by transpiration.
Key word: soil water regime, runoff strip width, dryland, irrigation, rain water harvesting,
186
8.1 Introduction
Although the concept of integrating rainwater harvesting and irrigation is rather ancient, the
integration of micro-flood (MFI) with in-field rainwater harvesting (IRWH) carries the
prospects of transforming runoff farming in the Free State Province of South Africa.
Common to runoff farming technologies is the harvesting of rainwater in the form of runoff
from a larger area and concentrating it to a smaller area for immediate or later productive use
(Owes et al., 1999; Hensley et al., 2000: Prinz and Malik, 2002: Anschutz et al., 2003).
Micro-flood and IRWH use different forms of runoff and their integrated effect on the soil
water balance (SWB) is yet to be established.
Improving soil water storage and availability for crop plants is the common goal of MFI and
IRWH. Micro-flood irrigation makes use of available irrigation water to supplement rainfall
through the application of small inflow rates to relatively short furrows (Austin, 2003). On
the other hand IRWH, designate some relatively larger section of the field to serve as mini-
catchments (runoff area) for the harvesting of in-field runoff that get stored in the soil profile
of the cropped area (Hensley et aI., 2000). The two practices appear to be compatible and
well adapted to improve rainfall effectiveness and productive potential of difficult soils,
especially those with a dry soil water regime.
Despite the fact that improving soil water regime by in-situ or in-field rainwater harvesting
and supplemental irrigation is a well established practice in runoff farming, the necessary risk
management procedures needs to be developed. One area of concern with integrated soil
water management practices is the transformation of the soil water balance that if not well
monitored could increase the risk of high evaporation, water logging, flooding that could
result soil erosion and overtopping of the runoff storage structures, and rising of the water
table. Traditionally, IRWH has guided against soil erosion and flooding risk related factors
by earmarking specific soil types for in-field runoff harvesting. Although, ex-field runoff has
been reduced to almost zero but high evaporation losses continues to be a challenge. Given
that irrigation also increases evaporation especially flood methods, this condition would be
expected to be made worse by the introduction of MFI. The use of organic mulches on IRWH
alone has suppressed evaporation by 15 % on clay soils and 31 % on sandy soils (Botha,
2006) compared to bare soils. Nevertheless, a total of 79 % of annual rainfall is reported to
have been lost to evaporation from IRWH fields (Botha, 2006).
187
One of the agronomic aspects that have become popular in recent time i the role of plant
canopy cover in suppressing evaporation. Given that the size of the canopy increa es with of
annual crops, growth, various agronomic practice have been shown to encourage early
canopy cover and hence, reduce evaporation as well a runoff losses. Thi included higher
plant density (Maqbool et aI., 2006), fertiliser application (Stroonijder, 2009) and narrow row
spacing (Eberbatch and Pala, 2005), and irrigation (Zhang et aI., 1998). The last two are
implicated to the IRWH and MFI through the respective strip width of the basin and runoff
area and provision of micro-flood supplemental irrigation. On the traditional IRWH design
the basin and runoff area constitutes of a respective standard strip width of 1 m and 2 m
(Figure 8.1). These strip widths were selected on the basis of mechanical convenience and
optimising in-field runoff under the semi-arid conditions of the province. However
optimising the surface area under canopy cover may require changes on the standard
dimensions but without violating the acceptable runoff to basin area urface area ratio (Oweis
et al., 1999). Aware of this predicament, Mzezewa et al. (2011) opted to encourage canopy
cover in the basin area by including a legume crop a an intercrop.
. 1m .... 2m
Runoff water accumulates
in basins and percolates
beyond the evaporation zone
Figure 8.1 A diagrammatic representation of the in-field rainwater harvesting technique
(Hensley et al., 2000).
Developing a mass con ervation procedure of soil water gained and lost for the newly
integrated MFI-IRWH management practice was the primary contribution of thi to study.
Previously, the soil water balance (SWB) components for IRWH was estimated without
con idering the role of canopy cover on crop water use. This could be under tood given that
188
most early studies focused in getting the in-field rainfall-runoff relationships and seasonal
crop water use and efficiency parameters correct. Given the growing pressure on efficient use
of agricultural water resources in semi-arid areas (Stroosnijder, 1987), integrating MFI with
lRWH make it necessary to account for every aspect that influences the soil water net gains
and losses, including that of the crop canopy cover. On this regard the specific objectives of
this study was firstly; to develop a procedure for quantifying SWB components for the
integrated MFI-IRWH management practice and secondly, to determine the effect of runoff
strip width (RSW) and water regime (WR) on the SWB components over the production
season.
8.2 Procedures for solving IRWH soil water balance
8.2.1 Soil water storage
Soil water storage was defined by the drainage upper limit (DUL) and lower limit (DLL)
boundary conditions. The DUL in our field studies was defined as the volumetric SWC when
drainage flux approaches 0.001 mm hour" while the DLL was the regarded to be the lowest
volumetric SWC recorded from each soil profile horizon during field trials. The difference
between the average profiles DUL and DLL depicts the profiles available water capacity
(PAWC). Under field conditions SWC normally fluctuates between this range illustrating
different levels of soil water deficit. Characterising the experimental site were DUL and DLL
values of 0.299 and 0.177 mm mmo) in the areas where the underlying horizon was of a
prismatic clay structure and the profile of 850 mm depth. On areas with dolorite C-horizon at
400 to 650 mm depth, the applicable average profiles DUL and DLL values were 0.28 to 0.12
mm mm", respectively. Differences between initial and final SWC during the production is a
function of soil water recharging and withdrawal processes. The latter is represented by deep
drainage, runoff (R), and evapo-transpiration (ET) and the former in profiles without water
table contribution by rainfall (P) and irrigation (1). Since runoff is converted from losses to
gains by IRWH, the soil water balance components influencing soil water storage could be
expressed as
LlSWC = (I + P) - ET - DD (8.1)
Where I1SWC is change in soil water storage and DD being deep drainage.
189
8.2.2 Harvested runoff
Rainwater harvested as in-field runoff or run-on for the different runoff strip widths were
estimated using linear regression equations summarized in Table 8.1, developed on site
according to the procedure described by Hensley et al. (2000). Runoff coefficient from
rainfall-runoff functional relationships was almost equal to zero at rainfalls amounting to 9.2,
22.1 and 24.5 mm from the respective 1,2 and 3 m RSW.
Table 8.1 Regression functions of rainfall (P) and runoff (Rin) relationships from different
runoff strip widths.
Runoff strip width (m) Linear regression
Rin= 0.373P - 7.13 0.98
2 Rin= 0.405P - 8.959 0.94
3 Rin= 0.513P - 12.549 0.99
Although rainfall received over the experimental area was assumed to be the same, the
infiltrated depth varied between the runoff and basin strips. The run-off generated over the
runoff strip and collected in the basin strip was accounted for by partitioning rainfall events
according to the runoff-basin area ratio.
8.2.3 Irrigation
According to the BEWAB+ (van Rensburg and Zerizqhu, 2008) irrigation scheduling model
an amount of 625 mm of water would be required for a 120 day maturing maize cultivar to
attain a maize yield of 10000 kg ha". Given that more than 50% of the long term average
rainfall is received between the months of November to March, a seasonal rainfall of around
300 mm was expected. Therefore, the balance was provided by means of the micro-flood
irrigation following a fixed schedule for FI and SPI produced by the BEWAB+ irrigation
software. For the 30 m furrow run the 40 L min-I inflow rate from previous test showed that
advance times between 20 to 30 minutes could attain infiltrated depths of 30 mm. The
designated 40 L min-I inflow rate at 100 % efficiencies could supply 0.9 m3 or 900 litres of
water on the 30 m furrow run at a cut-off time of 23 minutes.
190
8.2.4 Deep drainage
Soil water (8) lost to deep drainage (DD) was estimated using the internal drainage curves
determined from earlier field studies on the same site. Due to spatial variation, the prismatic
and dolorite rock were the two predominant C-horizons across the experimental area. The soil
water content-drainage relationship from the prismatic and dolorite horizons assumed
exponential functions, respectively written as:
(8.2)
(8.3)
Accumulative drainage over the production season from each plot was presented as the total
drainage component of the soil water balance.
8.2.5 Evapo-transpiration
Evapo-transpiration is the component of the soil water balance that is generally unknown and
constitutes of evaporation and transpiration. The soil water balance expression (8.1) could be
re-arranged to become:
ET = (I + P) - DD - ilSW C (8.4)
Because the IRWH involves different tillage and row systems for the runoff and basin strips,
the calculations of ET treated these areas as separate subsystems. The contributions of
evaporation and transpiration to ET from the runoff (ETr) and basin strips (ETb) were
considered important in assessing the effects of different runoff strips on IRWH under
different soil water regimes. In the basin area ETb was mainly a function of crop water use
and canopy cover. During seed germination and seedling establishment that lasted about 2
weeks after planting, canopy cover or crop water use was considered to be zero, hence
evaporation constituted 100 % of ETb. At full canopy cover contributions from evaporation
were assumed to be 10 % and after cob filling was completed increased to 50 %. Canopy
cover as a function of crop height was observed to reach maximum at 7 and 8 weeks for the
irrigated and rain-fed plots, respectively. Transpiration (Tg) and evaporation (EVb) from the
basin area was related to ET b using the expression:
Tb = {JETb (8.5)
EVb = (1- {J) ETb (8.6)
191
Where P is the parameter that ranged from 0 to 0.9 (Figure 8.2), was considered to be zero at
the initial stage and 0.9 at full canopy cover. Partial canopy cover as for the 2 and 3 m RSW;
P ranged from 0 to 0.5 from establishment to the mid-vegetative stage. In the runoff area ETr
was the function of the runoff strip width and the effect of crop canopy. When the crop was
fully grown three classes of canopy cover were used to describe three runoff strip widths used
in this study. A full canopy was used for the 1 m runoff strip width and thus it was treated in
a similar manner with the basin strip. In the 2 m runoff strip and the 1 m strips of the 3 m
runoff area that were adjacent to the basin crop rows were considered to be under partial
canopy cover. The central strip in the 3 m runoff area was not affected by canopy cover and
therefore represented the total evaporation (EvIOO%)A. n average of the Ev 100%from the same
block treatment provided reference evaporation (Evref) for the experimental site. Calculation
of the ETr from the 2 and 3 m runoff strips assumed the same soil water balance expression in
Eq. 8.4. At the plot scale the ETpwas calculated using the weighted average of the basin and
runoff ET written as:
(8.7)
Where N is number of 1 m width runoff strips and Wp is the plot total width (m). To
characterise the contributions of evaporation and transpiration at plot scale also required that
ETr be partitioned. At the initial crop stage ETr= EVr for all the N strip units. The P
parameter ranging from 1 to 0.5 (Figure 8.2) was used to relate the decreasing contribution of
evaporation from the partially shaded strips (Evrp) to ETr. At mid and late season Ev., was
estimated from the average EVrefusing the expression
Evrp = {J. Evre! (8.8)
Where, P assumed a constant coefficient of 0.5. Subtracting from ETr yielded the
transpiration component (Trp)from the partially shaded strips. Integrating Ev., and Trpon the
ETp for a 2 m runoff strip the expression in Eq. 8.7 could be re-written as:
(8.9)
(8.10)
For the 3 m runoff strip Eq. 8.10 becomes:
192
(8.11)
8.2.6 Statistics
A split plot analysis of variance was used to establish the effect of RSW and WR and its
interaction on SWB components under IRWH. The SAS 9 software was used in this
endeavour in combination with the procedures for least significant different (LSD) test
described by Gomez and Gomez (1984).
8.3 Results and discussions
8.3.1 Illustrations of the soil water balance procedure
Figure 8.2 and Table 8.2 show results that illustrates the soil water balance procedure over
the one production season from a 2 m RSW plot under supplemental irrigation. Integrating
MFI with IRWH presented the basin area (BA) with two water management functions; first,
to store rainfall plus run-on (Rn) produced in the runoff area (RA) and secondly, to distribute
uniformly the advance stream from supplemental micro-flood irrigation. Because of the
presence of a crop in the BA the procedure was able to discriminate between the gains and
losses from the BA, RA and the plot area (PA) during the major rainfalls and irrigation to
mitigate the dry spells.
Over the production period a total rainfall of 281 mm was received with major rainfall
occurring on the 7, 39, 60 and 99 DAP. The corresponding rainfall amounts were
respectively, 49 mm, 30 mm, 36 and 49 mm. The BA had a relatively steeper rise in PSWC
with the first two rainfalls recording an increase of 54 mm compared to the 42 mm from RA.
The 3rd and 4th rainfalls brought increases for profile soil water content (PSWC) of 37 and 51
mm from the BA and 16 and 24 mm from the RA, respectively. The additional contribution
from in-field runoff harvested from the RA as run-on into the BA over and above the rainfall
it received explains the differences in PSWC between the BA and RA. This was consistent
with the principle of in situ rainwater harvesting that runoff collected from a larger area gets
concentrated in a smaller area where crops grow to complement the short falls brought by
low rainfalls (Oweis et al., 1999; Hensley et al., 2000). At plot scale the corresponding
increases in PSWC for the same rainfall events were 46, 19 and 31 mm reflecting a weighted
average of the BA and RA. These gains were lower than those from the BA since in the 2 m
RSW the BA constituted only 33 % of the total plot area with the balance assigned to the RA.
193
Previous studies preferred to use the arithmetic mean to determine the plot water balance
components even though the proportional surface area from the BA and RA were different
(Botha, 2006; Joseph, 2007; Mzezewa et al., 2011). Overestimation in soil water studies as a
result of using the arithmetic mean due to its mathematical structure is well acknowledged in
literature (Yeh et al., 1985; Yeh and Harvey, 1990).
1 ~ BA: Dryland -. - RA: Dryland
0.9 A
0.8
0.7
~ 0.6
0.5
0.4
0.3
0.2
0.1
0
0 7 14212835424956637077 849198105112119
OAP
1 ~ BA; Irrigation - • - RA:irrigation
0.9 B
0.8
0.7
0.6
~ 0.5
0.4
0.3
0.2
0.1
0
0 7 14 21 28 35 42 49 56 63 7077 84 91 98105112119
OAP
Figure 8.2 Approximated ~ parameter representing the fraction of canopy cover during the
production season for the basin area (BA) and runoff area (RA) of the dryland (A) and
irrigation (B) plots.
194
Table 8.2 lilustration of the soil water balance procedure using estimated water processes over the production season from the basin, runoff and
plot area of in-field rainwater harvesting with a 2 m runoff strip under micro-flood supplemental irrigation.
Designated Da;rs after I2lanting
Area SWB (mm) 7 14 21 39 47 53 60 67 71 78 84 87 99 105 109 116
PSWC 179.5 185.2 177.0 122.9 128.1 141.1 152.1 151.2 132.9 133.3 121.5 148.2 173.2 142.8 168.9 177.1
~SWC -48.1 -5.6 -8.1 -54.1 5.2 13.0 11.0 -0.9 -18.3 0.5 -11.8 26.7 25.0 -30.4 26.1 8.3
Basin DD 0.02 0.01 0.03 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.12 0.18
ET 22.1 25.3 54.9 89.8 56.9 42.9 36.3 48.1 18.3 47.9 23.9 28.6 53.9 24.3 27.3 16.5
E 22.1 25.3 45.2 32.0 5.7 4.3 3.6 4.8 1.8 4.8 2.4 2.9 5.4 5.8 9.1 8.2
T 0.0 0.0 9.7 57.8 51.2 38.6 32.6 43.3 16.5 43.1 21.5 25.7 48.5 18.5 18.3 8.3
PSWC 161.6 173.2 159.3 152.0 140.3 147.1 156.2 141.0 129.8 123.0 126.8 129.3 147.9 133.7 139.1 154.3
~SWC -30.4 -11.6 -13.9 -7.3 -11.7 6.7 9.1 -15.2 -I 1.2 -6.8 3.8 2.5 18.6 -14.2 5.5 15.2
Runoff DD 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
ET 15.1 26.9 31.5 67.7 57.6 8.4 42.5 34.8 22.5 20.3 16.6 15.6 46.2 24.3 6.0 15.8
E 15.1 26.9 28.8 46.2 28.8 4.2 21.2 17.4 11.2 10.2 8.3 7.8 23.1 12.2 3.0 7.9
T 0.0 0.0 2.6 21.4 28.8 4.2 21.2 17.4 11.2 10.2 8.3 7.8 23.1 12.2 3.0 7.9
PSWC 167.6 177.2 165.2 142.3 136.3 145.1 154.8 144.4 130.8 126.4 125.0 135.6 156.3 136.7 149.0 161.9
~SWC -36.3 -9.6 -12.0 -22.9 -6.1 8.8 9.8 -10.5 -13.6 -4.3 -1.4 10.6 20.7 -19.6 12.3 12.9
Plot DD 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1
ET 12.4 17.4 28.8 52.5 38.2 17.1 26.2 27.7 13.6 22.7 13.5 14.7 33.4 16.2 I 1.1 10.8
E 12.4 17.4 24.7 26.1 14.3 5.1 11.0 10.3 5.8 7.7 4.9 4.7 13.4 8.0 5.1 7.3
T 0.0 0.0 4.1 26.4 23.9 12.0 15.2 17.3 7.8 15.1 8.6 10.1 20.0 8.2 6.0 3.5
PSWC= profile soil water content, ~SWC= change in soil water content, DD= deep drainage, ET= evapotranspiration, E= evaporation, and T=
transpiration
195
Stabilising soil water storage and crop water use is one of the primary goals for integrating
supplemental irrigation with in situ rainwater harvesting (World Bank, 2005; Rockstrom et
al., 2007). Supplemental irrigation was used to bridge the persistent dry spells of 25,21,39,
and 17 days that occurred during the periods of 14 to 39 DAP, 39 to 60 DAP, 60 to 99 DAP
and 99 to 116 DAP, respectively. The rain that occurred during these periods was too low to
adequately recharge the soil water and this was not strange given the low and erratic rainfalls
of the area (Botha, 2006). Supplemental irrigation maintained the PSWC at relatively higher
levels in the BA than that from the RA that only relied on rainfall, typical of IRWH under
dryland conditions. Increased soil water availability in the BA area was of paramount
importance given that crop water demand had increased over the production period reaching
critical levels between 47 to 105 DAP, a time corresponding to the vegetative growth and
reproductive crop growth phase (FAO, 2011).
Concentrating in-field runoff and introducing supplemental irrigation in BA was illustrated in
the procedure by increased water availability not only for total crop water use (ET) but also
for DD. In the first 21 DAP, ET was predominately in the form of Ev and recorded an
increase from 22 to 55 mm while DD ranged from 0.01 to 0.03 mm. For the same period DD
remained practically zero while ET increased from 15 to 32 mm in the RA. Given that the
soil was primarily bare and crop water demand at a very infant stage, the high Ev and low
DD captured by the procedure was reasonable. Rockstrom et al. (2007) recommended dry
planting and delaying water application early in the season to minimize high Ev and DD.
Throughout the season ET from BA remained higher than that from the RA irrespective of
rainfalls events with the exception of the 71, 105 and 116 DAP. The low ET for the BA
corresponding to these DAP can be attributed the short time interval of 4 days for 71 DAP
SWC measurement, and the general slow down in crop growth at the onset of the
reproductive stage (FAO, 2005). Decline in growth rate towards the end of the season could
be attributed to ET being lower than in the RA. Interestingly, the procedure was able to
capture the small amounts of DD at the later stages of the growing period while during the
mi-season it was nearly zero. This was an indication of that transpiration during the high crop
water demand kept the PSWC low enough to prevent significant DD losses. Similar findings
were found by Gobeze (2012) who recognised that DD during the growing season despite
regular irrigation DD was kept at negligible levels by maize transpiration in a well rooted soil
profile. The function of root water uptake on reducing DD was complemented by the
196
selection of soils suitable for IRWH with a restrictive layer in the B or C horizon (Hensley et
al.,2000).
Over and above the noticeable differences in ET between the BA and RA was the inverse
relationship between Ev and T during the production season. While Ev dominated ET in the
early stages of plant growth, the proportion of T increased remarkably from 39 to 99 DAP.
With an approximated average reference evapo-transpiration (ET0) of 6.5 mm per day for the
same period, the corresponding daily average increase in T that could be deduced from the
SWB table (Table 8.2) was an increase from 3 mm to 6 mm in the BA and 1 to 2.1 mm in
RA. On the other hand Ev decreased from a daily average of 6 to 1.3 mm in the BA and from
4 to 2.4 mm in the RA for the same period. The increase in T was consistent with crop
growth and crop water demand that is widely described using the crop coefficient (K};
depicted as the ratio of crop water use to ETo (FAO, 2011). In this study a ~ parameter was
used to represent the crop growth with respect to canopy cover; a primary factor in the
approximation of K;
The benefit of concentrating harvested runoff and supplemental irrigation water to where
crop plants grow was well illustrated in the overall water balance. Although at plot scale Ev
and shared an equal proportion of 46 % of the total available water, Ev from the BA was 50
% less than that of the BA indicating that much of the water concentrated in the BA was
channelled to T than Ev. In addition, ET from the BA was approximately 40 % greater than
that of the RA with corresponding T values of 254 mm more than the RA. This was
consistent with the 315 mm of water applied in the BA through supplemental irrigation.
Despite these increases in water storage DD was within negligible levels in agreement with
previous studies that soils with a restrictive layer were effective in reducing DD losses
(Hensely et aI., 2000; Hensley et aI., 2007). However prospects of lateral drainage occurring
from the BA to the RA could be supported by the ET in the RA that was 170 mm higher than
the seasonal rainfall. Given the restrictive DD conditions, this was not surprising because
water was concentrated only to a 1 m RSW and the remaining 2 m RSW constituting of the
the RA was virtually dry in most cases.
8.3.2 Effect of runoff strip width and water regime on the soil water balance
components
Table 8.3 summarizes the gains from rainfall and irrigation under supplemental and full
irrigation during the production period. Statistical summary pertaining to the effect of main
197
and sub treatment on soil water storage and losses at different stage of crop growth was
presented in Table 8.4, Table 8.5 (plant establishment), Table 8.6 (vegetative stage), Table
8.7 (reproductive) and Table 8.8 (ripening) stages. Although the RSW (1 m, 2 m and 3 m)
and WR (DL, SPI and FI) had variable effect on the SWB components over the production
season, some comparisons could be drawn.
Plant establishment stage:
The plant establishment phase lasted for 21 DAP and this was within the time frame of 15 to
25 days reported by Smith (2006). Early during this period water components started to be
affected by the RSW and WR main treatments while their interaction had an effect towards
the end of this stage. Predominately affected was the soil water storage and water use related
components. This was attributed to the 48.7 mm and 27 mm rainfalls received after the
respective 7 and 14 DAP and were of sizable amounts to recharge the soil profile. Affected
by RSW throughout this period was the PSWC and flSWC. The PSWC on the 7, 14 and 21
DAP recorded a respective range of 0.20 to 0.25 mm mm", 0.23 to 0.26 mm mm-I and 0.23
to 0.24 mm mm", The 1 m and 2 m RSW had comparably higher PSWC than the 3 RSW.
Improvement in PSWC among the RSW was expected to favour the 3 m RSW because of the
large surface area of its runoff or catchment area to harvest in-field runoff (Oweis et al.,
1999). But this was not the case because of the high roughness of the field surface especially
at this early stage when mechanical activities from basin preparation and fertiliser application
are still new.
Change in SWC was also affected by RSW with the 1 m and 2 m RSW recording equally
high net gains of 33 mm on the 7 DAP and equally low net gains of -10 mm on the 21 DAP.
The 3 m RSW had a high net gain of 23 mm on the 14 DAP; a period that was also
characterised by a relative increase in DD and Ev from all RSW. Water application from the
FI on the 14 DAP could have exacerbated this situation; hence FI had the highest net gain of
22 mm. Despite the increases in DD following the rainfall and subsequent irrigation there
were no significant differences suggesting that the soil profile was initially dry at planting, a
common phenomenon in semi-arid areas (Stroosnijder, 1987). Rockstrom et al. (2007)
described dry planting as a conservative strategy to minimise Ev and DD, given that the soil
is predominately bare and crop water demand is still at infant stage. This was supported by
the fact that ET was predominately in the form of Ev and increased with water availability an
observation that was well established in literature (FAO, 2011).
198
Table 8.3 Summary of the analysis of variance depicting the effect of runoff strip width (RSW) and water regime (WR) on selected soil water
balance SWB components.
Days after pl_antin_g
SWB components Establishment Vegetative Reproducti ve Ripening
(mm) Factors 7 14 21 39 47 53 60 67 71 78 84 87 99 105 109 116
Profile SWC RSW * * * ns ns ns * ns ns * ns ns * ns ns ns
WR ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns
RSW x WR ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns
RSW * * * ns ns ns * ns ns ns * ns ns ns * ns
flSWC WR ns * * * * * * * * * * * * * * *
RSW x WR ns ns * ns ns ns ns ns ns ns ns ns ns ns * ns
RSW ns ns ns ns ns ns ns ns ns ns ns ns * * ns *
Deep drainage WR ns ns ns * ns ns ns * * * ns ns * ns ns *
RSW x WR ns ns ns ns ns ns ns ns ns ns ns ns * ns ns ns
RSW ns ns ns ns * * * * * ns * ns * * ns ns
Evapo-transpiration WR ns ns * * ns ns ns ns * ns * * ns ns * ns
RSWx WR ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns
RSW ns ns ns ns * * * * * * * * * * * ns
Evaporation WR ns * * ns ns * ns ns ns ns ns ns ns ns ns ns
RSWx WR ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns
RSW * * * * * * * * * * * * * *
Transpiration WR ns * * * * ns * * * * * ns * ns
RSW x WR * ns ns ns * ns ns ns ns ns ns ns ns ns
*= Significant at 0.05% probability level, ns= not significant, -=data not available.
SWC: soil water content.
199
The plant establishment stage was concluded with RSW and WR interactions on t1SWC and
T. From the 14 to the 21 DAP is relatively a period of a dry spell and for that reason
supplemental and full irrigation were applied. At this period plant canopy cover was
approximated at 18 % compared to the 10 % approximated by FAO (2011). Interaction effect
on t1SWC shows that the 2 m and 3 m RSW were affected by irrigation in the same manner
while net losses from the 1 m RSW were comparable for all WR. Dryland had the highest net
gain of 16 mm coming from the 3 m RSW, although it could be too early to attribute this
strong interaction to the in-field runoff potential of this RSW, but it could be taken as an
indication that net gain from in-field runoff could exceed that from irrigation provided
sufficient rainfall is available. Interaction on T ranged from 6.7 mm to 1.2 mm with the
highest and lowest coming from the DL with the 2 m and 3 m RSW. The strong involvement
of the DL especially from both interactions could be attributed to the substantial amount of
rainfall received at the onset of this growth stage. Although irrigation was introduced towards
the end of this period, the strong interaction of 5.8 and 5.1 mm from the respective SPI and FI
confirmed the potential contribution of supplemental irrigation could have on transpiration
(World Bank, 2005).
Vegetative growth stage:
Soil water components were widely affected by the RSW and WR mam effects and
interaction effect coming late during this growth stage that lasted about 46 days. Rainfalls of
29.1 mm, 17.1 mm, 10.9 mm and 36 mm received the respective 39, 47, 53 and 60 DAP were
the major gains for the DL. Water applied at plot scale from the FI was a depth of 45 mm, 15
mm, and 30 mm on the 39, 53 and 67 DAP, respectively. On the other hand, 15 mm SPI was
applied on the 47, 53 and 67 DAP. Despite the fairly even rainfall distribution, the first
significant rainfall occurred after 25 days of dry spell, a reminder of the erratic nature of
rainfall in the area (Botha, 2006).
Changes in soil water content were the most frequently affected component by the RSW and
WR main effects. On the 39 DAP t1SWC was affected by WR factors recording the highest
net loss of -21 mm followed by DL with -12 mm.The higher losses from SPI reflect the
nature of deficit irrigation (Bennie et al. 1998) mainly because the crop establishment phase
has just passed. Deep drainage was affected by WR on the 39 and 67 DAP and both cases the
FI recording highest drainage of 4.1 and 2.6 mm, respectively with the lowest conling from
the DL although it was comparable with SPI. The relatively low activity in DD despite the
200
frequent rainfall and irrigation could be suggesting that water uptake for T increased with
water availability throughout this growth phase.
This stage was concluded with RSW and WR interactions on L1SWCand T. From the 14 to
the 21 DAP is relatively a period of a dry spell and for that reason supplemental and full
irrigation were applied. At this period plant canopy cover was approximated at 18 %
compared to the 10 % approximated by FAO (2011). Interaction effect on L1SWCshows that
the 2 m and 3 m RSW were affected by irrigation in the same manner while net losses from
the 1 m RSW were comparable for all WR. Dryland had the highest net gain of 16 mm
coming from the 3 m RSW, although it could be too early to attribute this strong interaction
to the in-field runoff potential of this RSW, but it could be taken as an indication that net gain
from in-field runoff could exceed that from irrigation provided sufficient rainfalls are
available. Interaction on T ranged from 6.7 mm to 1.2 mm with the highest and lowest
coming from the DL with the 2 m and 3 m RSW. The strong involvement of the DL
especiall y from both interactions could be attributed to the substantial amount of rainfall
received at the onset of this growth stage. Although irrigation was introduced towards the end
of this period, the strong interaction of 5.8 and 5.1 mm from the respective SPI and FI
confirmed the potential contribution supplemental irrigation could have on transpiration
(World Bank, 2005).
Evapo-transpiration, Ev and T were also affected by RSW for the greater part of this phase
with only T affected as early as on 39 DAP. Transpiration was constantly higher at all times
from the 1 m RSW while it was comparably similar from the 2 and 3 m RSW. In contrast Ev
ranged from 23 to 3 mm, 29 to 9.4 and 30 to 11 mm from the respective 1 m, 2 m and 3 m
RSW. The increase in Ev with increasing RSW could be explained by the surface area
available for direct radiation and that which is covered by the plant canopy (Eberbach and
Pala, 2005). Most plants from twin rows interlock leaves during the vegetative stage (FAO,
2011) and this is more pronounced on narrow rows of 0.5 to 1 m wide (Onyango, 2009 ). In
this study this was illustrated by a linear relationship between plant height and canopy cover
and it was estimated to reach full canopy cover after the 47 DAP. To this effect the
proportional contribution of Ev on ET was rapidly reduced to 10 % while that of T was
increased up to 90 % depending on the surface area under full canopy cover. The 1 m RSW
had 90 % of its surface area under full canopy cover while 2 and 3 m RSW had 50 and 33%
under partial canopy cover. This explains why Ev from the wider RSW was comparably
higher than from the 1 m RSW.
201
Table 8.4 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance (SWB) during plant establishment growth stage.
Factors RSW x WR
RSW (m) WR Im 2m 3m LSD( t :S 0.05) CV (%)
DAP SWB 2 3 DL SP FI DL SP FI DL SP FI DL SP FI RSW WR RSW x WR RSW WR
PSWC 0.25a 0.24" 0.20b 0.24 0.23 0.22 0.26 0.24 0.24 0.24 0.26 0.23 0.22 0.19 0.20 QOO ns M 6.72 30.38
!':!.SWC 32.7" 33.0a 19.0b 26.2 29.2 29.3 30.1 33.1 34.8 31.7 33.5 33.7 16.7 21.0 19.3 4.3 M M 12.5 23.8
7 Dr 0.8 0.9 1.1 1.3 0.7 0.9 0.6 0.6 1.2 1.5 0.5 0.7 1.8 0.9 0.6 ns ns ns 92.8 123.1
ET 15.0 14.3 15.5 15.4 14.9 14.5 16.8 14.9 13.3 13.8 14.8 14.3 15.6 15.0 16.0 ns ns ns 15.2 39.7
E 15.0 14.3 15.5 15.4 14.9 14.5 16.8 14.9 13.3 13.8 14.8 14.3 15.6 15.0 16.0 ns ns ns 15.2 39.7
T
PSWC 0.26a 0.26a 0.23b 0.24 0.25 0.26 0.26 0.24 0.27 0.25 0.27 0.26 0.23 0.23 0.25 0.02 ns ns 5.28 32.77
!':!.SWC 11.2b 12.7b 23.la 13.3b 11.7b 22.0" 9.8 4.0 19.9 14.1 8.1 16.116.123.2 30.0 3.8 6.1 ns 19.6 63.7
Dr 3.0 1.9 2.3 1.4 2.5 3.3 0.8 4.0 4.1 2.2 1.1 2.5 1.3 2.4 3.2 ns ns ns 88.6 118.0
14 ET 18.7 19.1 17.4 16.1 16.9 22.3 17.6 19.5 19.0 16.0 17.9 23.5 14.6 13.2 24.3 ns 4.4 ns 14.4 39.5
E 18.7 19.1 17.4 16.1b 16.9b 22.3" 16.2 19.5 19.0 11.8 17.9 23.5 14.1 13.2 24.3 ns 4.4 ns 14.4 39.5
T
PSWC 0.24a 0.24a 0.23h 0.24 0.23 0.24 0.25 0.23 0.25 0.25 0.25 0.23 0.24 0.21 0.23 0.03 ns ns 6.14 32.39
21 !':!.SWC -I Da _lOa _3b la _12b _12b _101x: _91x: _91x: -4b _12" _15c 16" _12" _13" 4.3 3.3 7.4 -45.2 -69.4
Dr 1.8 2.3 1.5 1.5 1.5 2.5 1.0 1.6 2.9 2.8 1.1 2.8 0.8 1.7 1.9 ns ns ns 36.8 87.6
ET 23.5 24.9 22.9 16.8 27.2 27.3 18.5 25.9 26.1 18.2 28.1 28.6 13.7 27.7 27.3 ns ns ns 20.7 40.2
E 18.9 17.0 16.1 12.9 19.4 19.7 15.7 20.1 20.9 10.4 20.1 20.4 12.5 18.1 17.7 ns ns ns 22.8 42.8
T 4.6a 4.9" 2.7b 3.6 4.4 4.2 2.8" 5.8"b 5.lb 6.7" 3.9" 4.0c 1.2d 2.I"d 3.4cd ns 1.8 19.7 32.3
Dl = dry-land; SP= supplemental irrigation; FI= full irrigation; ns = not significant
202
Table 8.5 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance (SWB) during plant establishment growth stage.
Factors RSW x WR
RSW (m) WR 1m 2m 3m LSD( t s 0.05) CV (%)
OAP SWB 2 3 DL SP F1 DL SP FI DL SP FI DL SP FI RSW WR RSW x WR RSW WR
PSWC 0.25" 0.24a 0.20b 0.24 0.23 0.22 0.26 0.24 0.24 0.24 0.26 0.23 0.22 0.19 0.20 0.02 ns ns 6.72 30.38
DoSWC 32.7" 33.0a 19.0b 26.2 29.2 29.3 30.1 33.1 34.8 31.7 33.5 33.7 16.7 21.0 19.3 4.3 ns ns 12.5 23.8
7 Dr 0.8 0.9 1.1 1.3 0.7 0.9 0.6 0.6 1.2 1.5 0.5 0.7 1.8 0.9 0.6 ns ns ns 92.8 123.1
ET 15.0 14.3 15.5 15.4 14.9 14.5 16.8 14.9 13.3 13.8 14.8 14.3 15.6 15.0 16.0 ns ns ns 15.2 39.7
E 15.0 14.3 15.5 15.4 14.9 14.5 16.8 14.9 13.3 13.8 14.8 14.3 15.6 15.0 16.0 ns ns ns 15.2 39.7
T
PSWC 0.26" 0.26" 0.23b 0.24 0.25 0.26 0.26 0.24 0.27 0.25 0.27 0.26 0.23 0.23 0.25 0.02 ns ns 5.28 32.77
DoSWC 11.2b 12.7h 23.1" 13.3h 11.7b 22.0" 9.8 4.0 19.9 14.1 8.1 16.1 16.1 23.2 30.0 3.8 6.1 ns 19.6 63.7
Dr 3.0 1.9 2.3 1.4 2.5 3.3 0.8 4.0 4.1 2.2 1.1 2.5 1.3 2.4 3.2 ns ns ns 88.6 118.0
14 ET 18.7 19.1 17.4 16.1 16.9 22.3 17.6 19.5 19.0 16.0 17.9 23.5 14.6 13.2 24.3 ns 4.4 ns 14.4 39.5
E 18.7 19.1 17.4 16.1b 16.9h 22.3" 16.2 19.5 19.0 11.8 17.9 23.5 14.1 13.2 24.3 ns 4.4 ns 14.4 39.5
T
PSWC 0.24a 0.24" 0.23b 0.24 0.23 0.24 0.25 0.23 0.25 0.25 0.25 0.23 0.24 0.21 0.23 0.03 ns ns 6.14 32.39
21 DoSWC -10" -10" _3b 1a _12b _12b -101x: _91x: _91x: _4h -12" -15" 16" _12" -13" 4.3 3.3 7.4 -45.2 -69.4
Dr 1.8 2.3 1.5 1.5 1.5 2.5 1.0 1.6 2.9 2.8 1.1 2.8 0.8 1.7 1.9 ns ns ns 36.8 87.6
ET 23.5 24.9 22.9 16.8 27.2 27.3 18.5 25.926.1 18.228.1 28.6 13.7 27.7 27.3 ns ns ns 20.7 40.2
E 18.9 17.0 16.1 12.9 19.4 19.7 15.7 20.120.910.4 20.1 20.4 12.5 18.1 17.7 ns ns ns 22.8 42.8
T 4.6" 4.9" 2.7b 3.6 4.4 4.2 2.8" 5.8"h 5.1h 6.7" 3.9" 4.0" 1.2d 2.1"d 3.4cd 1 ns 1.8 19.7 32.3
Dl = dry-land; SP= supplemental irrigation; FI= full irrigation; ns = not significant
203
Table 8.6 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance in the vegetative growth stage.
Factors RSWx WR
RSW WR Im 2m 3m LSD( t s 0.05) CV (%)
OAP SWB I 2 3 0 SP FI 0 SP FI 0 SP FI 0 SP FI RSW WR RSW.WR RSW WR
PSWC 0.22 0.22 0.22 0.23 0.20 0.24 0.22 0.20 0.26 0.22 0.21 0.23 0.23 0.19 0.24 ns ns ns 7.9 33.4
i'lSWC -13.0 -14.3 -4.9 -I I.9h _21.3" LOa -19.7 -21.2 2.0 -11.4 -28.5 -3.1 -4.6 -14.1 4.0 ns 3.8 ns -81.3 -59.0
39 Dr 2.2 2.3 2.0 1.1b 1.4h 4.la 0.9 2.1 3.6 1.5 1.5 3.8 0.8 0.6 4.8 ns 1.3 ns 51.0 79.0
ET 61.1 57.2 51.1 44.7 52.3 72.4 48.4 57.3 77.7 41.0 56.6 73.9 44.7 43.1 65.5 ns 12.4 ns 19.7 36.2
E 22.8 29.1 29.8 27.4 23.9 30.4 24.6 16.8 27.0 24.1 29.5 33.7 33.6 25.4 30.5 ns ns ns 25.7 38.7
T 38.3a zs.r' 21.3" 17.2b 28.4" 42.0a 23.8 40.4 50.7 16.8 27.1 40.2 I 1.1 17.8 35.0 7.0 6.4 ns 19.6 36.1
PSWC 0.21 0.21 0.21 0.21 0.20 0.22 0.20 0.20 0.23 0.22 0.21 0.22 0.22 0.19 0.22 ns ns ns 9.2 34.1
i'lSWC -8.7 -5.6 -5.4 _7.0h 0.3a -13.0" -10.4 2.6 -18.3 -5.2 -2.4 -9.3 -5.5 0.7 -11.4 ns 2.9 ns -49.3 -73.8
Dr 1.2 0.7 0.3 0.6 1.4 0.2 1.4 2.0 0.2 0.1 1.7 0.2 0.4 0.4 0.2 2.2 2.8 3.7 53.1 68.9
47 ET 34.9" zs.r" 26.6ah 24.3 29.5 32.9 26.2 37.1 41.4 21.4 27.8 26.2 25.3 23.5 31.0 9.6 ns ns 27.1 37.1
E 5.6 b 10.6ah 15.6a 12.9 8.6 10.4 9.3 2.8 4.8 1l.2 10.8 9.9 18.1 12.2 16.5 6 ns ns 45.8 57.2
T 29.3" 16.2h 13.5b II.4h 25.0a 22.5a 16.9 34.3 36.6 10.3 22.0 16.4 7.2 18.7 14.5 6.2 4.1 ns 25.8 34.8
PSWC 0.22 0.23 0.23 0.21 0.22 0.25 0.19 0.22 0.24 0.21 0.22 0.25 0.22 0.21 0.26 ns ns ns 8.0 37.1
53 i'lSWC 2.7 4.3 4.3 _6.6b 9.9a 8.0a -10.0 9.2 8.9 -4.1 9.6 7.4 -5.8 10.8 7.8 ns 3.2 ns 46.9 64.3
Dr 1.7 1.7 1.5 0.0 2.4 2.4 0.02 3.1 1.9 0.03 2.0 3.0 0.1 2.1 2.2 ns ns ns 65.1 83.9
ET 25.6" 20.0b 24.6b 20.1 24.9 25.2 20.9 28.8 27.0 14.9 21.0 24.2 24.5 24.8 24.4 3.2 ns ns 11.3 39.4
E 21.9h 12.8b 12.6a 10.3 18.7 18.4 15.8 26.6 23.3 7.1 14.7 16.6 7.9 14.8 15.2 2.8 3.5 ns 14.6 36.3
T ze.i- 16.5b 11.6b 14.9b 19.2a 20.2a 22.9 26.3 29.1 12.7 15.8 20.9 9.0 15.5 10.5 4.7 3.6 ns 21.4 32.9
PSWC 0.21h 0.23a 0.22a 0.2 0.23 0.24 0.17 0.22 0.25 0.20 0.25 0.23 0.22 0.21 0.24 0.01 ns ns 5.0 34.7
i'lSWC 10.3h 12.2h 16.9a 19.7" 9.6b io.i' 16.6 4.8 9.5 18.1 9.1 9.3 24.4 14.9 11.5 3.8 2.7 ns 23.6 34.4
Dr 0.7 0.8 0.8 0.5 0.7 1.2 0.01 0.6 1.5 0.7 0.2 1.4 0.7 LI 0.6 ns ns ns 71.4 84.0
60 ET 32.4a 24.7b 21.7b 23.1 29.2 26.5 25.3 41.0 30.8 22.2 26.6 25.3 21.7 20.0 23.4 3.6 ns ns 11.1 34.5
E 3.2b 10.3a 12.0" 8.5 8.2 8.8 2.5 3.0 4.0 10.5 10.5 9.9 12.4 11.0 12.5 2.8 ns ns 26.9 62.1
T 29.2a 14.4b 9.7b 14.6b 21.0" 17.7ab 22.8b 38.0" 26.9b 11.7"d 16.1" 15.4" 9.3"d 8.9" 10.9"d 4.9 2.8 6.5 22.7 26.2
PSWC 0.21 0.22 0.21 0.20 0.20 0.23 0.17 0.20 0.25 0.21 0.21 0.22 0.21 0.19 0.22 ns ns ns 6.4 35.0
i'lSWC -2.6 -5.3 -7.9 -2.1 a _8.8" _4.9b -3.2 -6.7 2.2 -1.6 -7.7 -6.6 -1.4 -12 -10 ns 2.8 ns -91.9 -86.6
67 Dr 1.0 1.6 1.3 0.3h 1.1h 2.6a 0.1 1.1 1.9 0.5 0.8 3.4 0.4 1.3 2.3 ns I ns 42.9 56.3
ET 28.7a 25.9h 22.2b 21.4 26.9 28.5 24.3 28.6 33.3 23.1 24.2 30.4 16.8 27.9 21.8 5.1 ns ns 18.9 32.6
E 2.6h 9.4a 10.5a 6.6 7.7 8.4 1.4 2.3 4.2 10.4 8.4 9.5 7.8 12.4 11.4 3.2 ns ns 35.3 47.8
T 26.la 16.5b 11.6b 14.9 19.2 20.2 22.9 26.3 29.1 12.7 15.8 20.9 9.0 15.5 10.5 4.7 ns ns 21.4 32.9
D = dry-land; SP = supplemental irrigation; PI = full irrigation; ns = not significant
204
Table 8.7 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance in the reproductive growth stage.
Factors RSW x WR
RSW WR Im 2m 3m LSD( i s 0.05) cv (%)
OAP SWB 2 3 DL SP FI DL SP FI DL SP FI DL SP FI RSW WR RSW.WR RSW WR
PSWC 0.21 0.22 0.20 0.20 0.20 0.23 0.17 0.2 0.25 0.21 0.21 0.22 0.21 0.19 0.22 ns ns ns 6.4 35.0
I1SWC -10.5 -9.7 -9.8 _12.8b _12.6b _4.5" -14.2 -14.6 -2.5 -12.5 -12.2 -4.6 -11.8 -11.0 -6.4 ns 2.5 ns -38.5 -41.1
71 Dr 0.3 1.0 0.4 0.2" o.i' l.4b 0.1 0.1 0.8 0.4 0.05 2.5 0.2 0.1 1.0 ns 5.9 ns 45.6 66.6
ET 17.3" 12.0b 1O.9b 11.2 13.9 15.1 12.9 18.8 20.2 11.7 12.1 12.0 8.8 10.9 12.9 3.8 ns ns 23.5 39.0
E 1.6h 4.7" 5.4" 3.4 4.0 4.4 0.7 1.4 2.7 5.3 4.7 4.2 4.1 5.8 6.4 1.9 ns ns 38.9 76.8
T 15.7" 7.2b 5.5h 7.8 10.0 10.6 12.2 17.3 17.5 6.5 7.4 7.8 4.7 5.2 6.5 3.1 ns ns 26.4 31.0
PSWC 0.18" 0.20b 0.19b 0.18 0.17 0.21 0.16 0.17 0.23 0.21 0.19 0.20 0.19 0.16 0.20 0.01 ns ns 4.9 32.9
6SWC -14.4 -14.1 -16.4 -31.0b -5.5'1 _8.4" -32.6 -3.2 -7.4 -30.7 -3.6 -8.0 -29.9 -9.7 -9.7 3.9 ns ns -36.1 -42.6
Dr 0.59 0.76 0.31 0.12b 0.22b 1.32" 0.14 0.32 1.30 0.20 0.02 2.05 0.01 0.32 0.60 ns 0.6 ns 54.1 69.5
78 ET 23.2a 17.4"b 17.0b 15.9 20.5 21.1 16.6 24.3 28.6 15.7 17.0 19.4 15.4 20.3 15.4 6.1 ns ns 26.6 48.0
E 2.1 b 6.9" 8.9" 6.2 6.4 5.3 1.7 2.0 2.5 7.8 6.4 6.5 9.1 10.9 6.8 2.4 ns ns 32.4 82.1
T 21.1 a 10.5b 8.1 h 9.7b 14.1a 15.9" 15.0 22.3 26.1 7.9 10.6 12.9 6.3 9.4 8.6 5.3 3 ns 32.5 36.9
PSWC 0.18 0.20 0.19 0.18 0.17 0.21 0.16 0.17 0.23 0.21 0.19 0.20 0.19 0.16 0.20 ns ns ns 7.0 33.6
84 6SWC _8.9h _5.9"h -4.3" -15.7b -2.1 a -1.3" -16.6 -4.3 -5.8 -14.0 -2.2 -1.6 -16.6 0.3 3.4 2.8 2.3 ns -48.6 -58.3
Dr 0.06 0.14 0.13 0.14 0.04 0.15 0.12 0.05 0.01 0.27 0.02 0.13 0.02 0.06 0.3 ns ns ns 67.8 75.0
ET 18.4a 12.7b lUb 10.8b 15.7" 16.3" I 1.1 21.2 22.7 10.5 14.2 13.6 10.8 11.7 12.6 4.2 3.1 ns 24.0 36.2
E 1.7b 5.5" 6.6a 5.1 4.2 4.5 1.3 1.6 2.1 6.1 5.4 5.2 8.0 5.7 6.2 1.7 ns ns 29.2 52.7
T 17.2" 7.5b 5.8b 7.3b 11.5" I 1.8" I 1.5 19.6 20.6 5.5 8.8 8.4 4.9 6.0 6.4 3.5 2 ns 28.1 32.0
PSWC 0.20 0.20 0.19 0.18 0.19 0.22 0.16 0.19 0.24 0.19 0.20 0.21 0.19 0.18 0.21 ns ns ns 7.4 34.4
6SWC 6.7 6.1 3.1 _O.4e 10.9" 5.4b -0.6 13.2 7.5 2.7 10.4 5.4 -3.3 9.3 3.3 ns 2.2 ns 71.7 67.7
Dr 0.2 0.2 0.1 0.2 0.3 0.1 0.1 0.4 0.1 0.3 0.3 0.02 0.04 0.2 0.1 ns ns ns 50.7 62.2
87 ET 11.0" 8.3h 7.6b 5.7b 10.8" 10.3" 5.6 14.3 13.0 5.3 9.7 9.9 6.3 8.4 8.1 2.4 1.8 ns 22.1 33.1
E 1.1h 3.7" 4.2" 2.6 3.2 3.2 0.6 1.1 1.8 3.2 3.9 3.8 4.1 4.5 3.9 1.7 ns ns 47.8 51.2
T 9.8a 4.6b 3.4b 3.1 b 7.6" 7.2a 5.lb 13.3" 11.2" z.r- 5.7b 6.1 b 2.2e 3.9bc 4.2bc 1.3 1.2 2.7 17.6 33.1
PSWC 0.20b 0.23" 0.22· 0.22 0.21 0.22 0.21 0.18 0.21 0.24 0.23 0.23 0.23 0.21 0.23 0.01 ns ns 5.3 33.6
I1SWC 19.5 21.0 20.3 31.2" 17.5b 12.1' 32.3 15.2 11.0 30.4 18.7 13.9 30.9 18.6 11.5 ns 3.7 ns 21.5 29.8
99 Dr 0.9b l.4b 3.6" O.4e 1.8b 3.7" o.or= 1.7bcd 0.9bcd 0.8'd o.z= 3.2bc 0.5ed 3.4b 7.0" 1.2 1.2 2.7 51.9 67.4
ET 42.7" 35.3"b 33.6b 32.3 37.9 41.5 31.1 46.5 50.6 32.7 35.5 37.7 33.1 31.7 36.1 8.0 ns ns 17.7 35.8
E 5.1 b 14.5" 17.8" 12.1 10.9 14.3 3.1 3.7 8.4 15.2 13.1 15.4 18.1 15.9 19.3 5.7 ns ns 37.6 66.3
T 37.7" 20.8b 15.9h 20.2h 27.0" 27.1" 28.0 42.8 42.2 17.5 22.4 22.4 15.0 15.8 16.8 6.6 4.2 ns 21.7 28.1
PSWC 0.21 0.21 0.21 0.22 0.19 0.22 0.19 0.19 0.24 0.22 0.20 0.21 0.23 0.18 0.21 ns ns ns 6.1 31.8
IDS 6SWC -6.7 -9.4 -12.3 -0.9" -18.2e _9.3b 6.9 -17.7 -9.2 .7 -19.1 -10.0 -10.2 -17.7 -8.8 ns 4.9 ns -66.2 -86.1
Continued 205
Dr 0.1 b 0.5" 0.6" 0.3 0.2 0.6 0.02 0.2 0.1 0.5 0.2 1.0 0.5 0.4 0.8 0.4 ns ns 66.2 77.8
ET 20.4· 17.3b 16.4b 20.5 16.5 17.2 19.9 19.1 22.1 20.6 15.7 15.7 2l.0 14.5 13.8 2.8 ns ns 12.8 3l.2
E 4.5b 7.4" 8.5" 7.0 6.7 6.8 3.4 3.7 6.4 8.3 7.4 6.6 9.2 9.0 7.4 2.8 ns ns 33.3 50.5
T 14.0· 8.lb 5.9b 7.8 9.8 10.4 10.9 15.4 15.6 6.8 8.4 9.1 5.7 5.5 6.4 2.3 ns ns 20.5 28.1
D = dry-land; SP= supplemental irrigation; FI= full irrigation; ns=not significant
Table 8.8 Effect of runoff strip width (RSW) and water regime (WR) on soil water balance in the ripening growth stage.
Factors RSW x WR
RSW(m) WR Im 2m 3m LSO( t :5 0.05) CV (%)
OAP SWB 2 3 DL SP FI DL SP FI DL SP Fl DL SP FI RSW WR RSW.WR RSW WR
PSWC 0.21 0.21 0.21 0.21 0.20 0.22 0.18 0.20 0.24 0.21 0.22 0.21 0.23 0.19 0.21 ns ns ns 7.3 33.0
L1SWC _2.6h _2.6h 4.4· -10.5" 10.4" _0.6b -18.6c 12.0· _l.2d .rs.r 1l.6a _l.2d 5.2°C 7.5·h 0.5ed 3.3 2.3 5.3 -50.5 -69.3
109 Dr 0.4 0.3 0.6 0.2 0.8 0.3 0.03 1.2 0.1 0.2 0.3 0.4 0.5 0.8 0.5 ns ns ns 50.9 71.2
ET 1l.2" 8.5b 8.7b 8.7 9.3 10.4 8.8 13.7 1l.2 8.4 6.6 10.5 8.9 7.7 9.6 2.2 ns ns 19.0 32.5
E 3.6a 4.8ab 5.9a 5.0 4.1 5.3 3.5 3.8 3.5 5.2 3.6 5.7 6.3 5.0 6.5 1.4 ns ns 24.1 37.1
T 7.6· 3.7b 2.8b 3.7 5.2 5.2 5.2 9.9 7.6 3.2 3.1 4.9 2.6 2.7 3.1 1.3 2.7 ns 22.7 38.6
PSWC 0.22 0.32 0.22 0.22 0.22 0.23 0.20 0.22 0.25 0.23 0.23 0.22 0.23 0.20 0.22 ns ns ns 6.3 32.8
L1SWC Il.l 9.9 10.2 15.4· 9.0b 6.8b 16.7 8.5 8.2 13.9 9.5 6.3 15.7 9.1 5.9 ns 2.4 ns 30.5 38.0
Dr 0.15b 0.69a 0.74" 0.25b O.44b 0.90" 0.0 0.3 0.1 0.5 0.2 l.4 0.2 0.9 1.1 0.5 0.35 ns 57.4 68.8
116 ET 16.6 15.6 13.1 13.8 15.4 16.1 14.2 18.5 17.2 16.8 14.0 16.0 10.5 13.7 15.1 ns ns ns 30.5 37.6
E 7.8 9.4 10.6 9.5 8.6 9.7 7.1 7.7 8.7 10.2 8.4 9.6 1l.3 9.7 10.8 ns ns ns 28.8 42.0
T 8.8a 6.2b 4.2b 6.0 6.8 6.4 7.1 10.8 8.6 6.6 5.6 6.4 4.2 4.1 4.3 2.4 ns ns 22.3 33.3
D = dry-land; SP= supplemental irrigation; FI= full irrigation; ns=not significant
206
The surface area from the 1 m RSW appears to have not only increased the area shaded by
plant canopy but also the accessibility of plant roots to soil water storage (Onyango, 2009).
Given the high surface area under full canopy cover from the 1 m RSW, the corresponding
high net losses from the soil water storage could only be attributed to the higher ET. This
appears to favour the introduction of irrigation on IRWH because the increased water
availability gets directed towards photosynthesis rather than unproductive losses (DD and
Ev). Dryland had the lowest ET and T among the WR a reflection of the water stress
environment confronting IRWH (Botha, 2006). Interaction was on the 60 DAP for the T
component with an interaction ranging from 38 to 8.9 mm. The SPI had the strongest and
lowest interaction from the respective 1 m RSW and 3 m RSW, illustrating the potential of
the SPI in stabilising the plant growth when integrated with IRWH.
Reproductive growth stage:
Soil water components were variably affected by the RSW and WR main effects. Effect from
the RSW and WR interaction on T and DD was seen on the 87 and 99 DAP. This stage lasted
about 45 days with flowering occurring as early as the 60 DAP falling under the vegetative
stage. Despite the high crop water demand required for grain development, there were three
rainfalls with corresponding amount of 12.1 mm, 10.3 mm and 48.5 mm on the 84, 87 and 99
DAP, respectively. Apparently a 24 day dry spell occurred during the flowering period and a
net loss of -31 mm from the soil water storage was reflected from the DL water regime. This
was undesirable given that water stress conditions at flowering could directly reduce grain
yield approximately up to 40% (Westgate et al., 2004).
Mitigation of dry spell included SPI with two applications of 15 mm on the 78 and 87 DAP
and Fr with 15 mm water application for the 71, 78 and 105 DAP as well as 7.5 mm on the 87
DAP. Showing more consistency with the irrigation events were improved changes on óSWC
and ET as the result of the main effect from WR. The Fr continued to record highest net gains
and ET, while DL had the least and improvements were consistent with rainfall events.
Affected by RSW throughout this stage were ET and its constituents. The 1 m RSW had ET
ranging from 43 to 11 mm, the 2 m RSW ranging from 8 to 35 mm and the 3 m RSW ranging
from 8 to 34 mm. The nearly similar ET range between the 2 and 3 m RSW is not surprising
given that both these RSW experienced partial shading from the spatially arranged runoff
area. The evidence of spatiality was shown Ev that ranged from 1 to 5 mm, 4 to 15 mm, and
4.2 to 18 mm from the 1 m, 2 m, and 3 m RSW, respectively. Although ET decreased with
207
increasing RSW the opposite is shown by the Ev range suggesting that T was higher at
narrow RSW. This was confirming earlier observations in this study that the I m RSW
suppressed Ev and enhanced T. Increasing canopy cover and concentration of crop roots
within the accessible near the basin area are the possible reasons (Magbool et al., 2006;
Grena and Hess, 1994).
Rare to these observations was a RSW and WR interaction on DD captured on the 99 DAP
just after the 49 mm rainfall. The DD losses from the interaction ranged from 7 to 0.01 mm
with the strongest interaction obtained from the FI of the 3 m RSW and the lowest from the
DL of the 1 m RSW. This was consistent with the constant recharge of soil water storage
from the FI and the capacity of the 3 m RSW to harvest sizeable in-field runoff. On the other
hand the 1 m RSW had the lowest DD losses and the reason could be twofold. Firstly, the
high abstraction of soil water from the plants roots concentrated over a small area (Eberbatch
and Pala, 2005) and secondly the low capacity of the RSW surface area to harvest adequate
in-field runoff could be the reason (Rockstrom et al., 2007). The SPI had DD interaction
ranging from 3 to 0.2 mm confirming the conservative advantage of recharging the soil
profile while reserving adequate rainfall storage space (Bennie and Hensley, 2001)
Ripening stage:
Runoff strip width and WR affected soil water components including an interaction on the
~SWc. Although this stage persisted beyond the 120 DAP, soil water measurements were
discontinued on the 116 DAP following a rainstorm with strong winds that lodged the crop.
Gains at this stage were from rainfalls of 8.5 mm and 24 mm from the respective 109 and 116
DAP. The first rain occurred just after the SPI and risk of DD was well taken care off given
the low and infrequent applications from SPI. This stage is characterised by grain
development including grain filling and dry matter accumulation that requires sustainable
supply of water (Westgate et al., 2004).
Apart from the RSW and WR interaction affecting ~SWC, ET and its constituents were
affected by RSW on the 109 DAP. Because of the interaction, the ~SWC ranged from 12 to -
19 mm with the highest net gain and losses coming from SPI of the 1 m RSW and DL of the
1 m RSW. The high net losses from DL were expected given the water stress conditions
encountered by dryland crop production. The strong interaction on SPI could be attributed to
the fact that 8.5 mm rainfall had just occurred after the SPI was applied and inadequate to
offset the abstraction that occurs in the 1 m RSW. Corresponding to the same period was the
208
highest Ev recorded from the 3 m RSW of 6.8 mm and lowest from 1 m RSW of 3.6 mm.
Transpiration was 7.6, 4.8 and 4.5 mm from the respective 1 m, 2 m and 3 m RSW.
Transpiration was also affected by RSW having a range of 3.7 to 8 mm with the highest from
the 1 m RSW. Corresponding ET followed the same trend with a range from Il to 8.8 mm.
Following a 2 consecutive month period being affected by the main effects of RSW and WR,
the change in ET on the 116 DAP to become unaffected by any of the main factors could be
considered to be a sign of slowing down by the crop plants and also of the atmospheric
evaporative demand at this late stage of the season (Kaisi, 2000; FAO, 2011). However, T
was affected until the 116 DAP with the highest T recorded from the 1 m RSW suggesting
that even though the crop could be approaching maturing water was still required for the
kernels to achieve physiological maturity (Westgate et al. 2004). Deep drainage was also
affected by RSW and WR foUow!ng the last rainfall of 24 mm suggesting that at this stage of
plant growth the soil profile water was no longer abstracted as it was during the vegetative
and reproductive stage. Although the amounts were relatively small with 1 mm being the
highest from the FI, suggesting that the soil profile from the PI was still having adequate soil
water reserves.
8.5 Conclusion
In this study a soil water balance was developed and applied to quantify the effect of the
integrated management practices of IRWH and MPI on the gains and losses over the
production season. The procedure considered the surface area available for in-field runoff
harvesting for the 1 m, 2 m and 3 m RSW. Also the surface area under canopy cover at
various crop growth stages was considered for the basin and runoff area. Available surface
area for in-field runoff was estimated to be 50, 67 and 75 % for the respective 1 m, 2 m and 3
m RSW with a corresponding canopy cover of 90, 50 and 33 % at mid season. The procedure
was consistent with the basic runoff farming and agronomic principles and therefore it was
regarded to be reasonable and fit for adoption.
The developed procedure was then applied to determine the effect of the runoff strip width (1
m, 2 m and 3 m) and water regimes (dryland, supplemental and full irrigation) on the soil
water gains and losses over the production season. At the plant establishment phase the soil
was initially dry, bare and with a high surface roughness. Therefore the influence of the RSW
surface area and canopy cover on soil water storage and evapo-transpiration was negligible;
hence evapo-transpiration was mainly in the form of evaporation. During the vegetative,
209
reproductive and ripening phases, net gain from the 2 and 3 m RSW were comparably higher
than from the 1 m RSW especially after the main rainfall not less than 24 mm. A net loss of
about -12 mm and more were associated with the dryland especially after dry spells of more
than 14 days.
Water application from either supplemental or full irrigation improved PSWC and net gain in
the same manner and that translated to higher evapo-transpiration, evaporation and
transpiration irrespective of the RSW. The 1 m RSW had superior evapo-transpiration and
transpiration despite the relatively low net gains. This phenomenon was attributed to the high
surface area fully shaded by the plant canopy and densely rooted as well. Rainfalls of not less
than 24 mm just after irrigation or at short intervals resulted on RSW x WR interaction.
While deep drainage was affected in a comparably similar manner between dryland and
supplemental irrigation, net water gain from the 2 m and 3 m RSW (dryland) was consistently
higher than that from irrigation.
Acknowledgements
We acknowledge the technical support rendered by the management and staff of Paradys
experiment station. Gratitude's to Mike Fair for his assistance regarding the use the SAS 9
software during the statistical analysis in this work.
210
8.6 References
Al-Kaisi, M., 2000. Crop water use or evapo-transpiration: Integrated crop management.
http://www.ipm.ipm.oastate.edu ,19111/2011. 10.30 am (LT).
Anschutz, J., Kome, A., Nederlof, M., de Neef, R., and van de Van T., 2003. Water
harvesting and Soil water retention. Sec ed. Agromisa Foundation. Wageningen,
Netherlands.
Araya, A. and Stroosnijder, L., 2010. Effects of tied ridges and mulch on barley (Hordeum
vulgare) rainwater use efficiency and production in Northern Ethiopia. Agricultural
Water Management, 97: 841-847.
Austin, c., 2003. Micro flood, a new way of applying waters. http://waterright.com.
22/09/2011, 10.00 a.m (LT).
Bennie, A.T.P. and Hensley, M., 2001. Maximising precipitation utilization In dryland
agriculture in South Africa- a review. J. Hydrol., 241: 124-139.
Bennie, A. T. P., Strydom, M. G. and Vrey, H. S., 1998. The use of computermodels or
agricultural water management on ecotope level. Water Research Commission report,
TT 102/98, Pretoria, South Africa.
Botha, J. J., 2006. Evaluation of maize and sunflower production in a semi-arid area using In-
field rainwater harvesting. Ph.D. (Agric) Dissertation, University of the Free State,
Bloemfontein.
Eberbatch, P. and Pala M., 2005. Crop row spacing and its influence on the partitioning of
evapo-transpiration by winter-grown wheat in Northern Syria. Plant and soil, 268:
195-208.
FAO, 2011. Chapter 5; Introduction to crop evapo-transpiration.http://www.fao.org.
20/11/2011, 10.30 am (LT);
Gobeze, L., 2012. Establishing optimum plant populations and water use of an ultra fast
maize hybrid (Zea Mays L) under irrigation. Ph.D. (Agric) Dissertation, University of
the Free State, Bloemfontein.
Gomez, K. A. and Gomez, A.A., 1984.Statistical procedures for agricultural research. 2nd ed.
John Wiley and Sons, New York, USA.
Hensley, M., Botha, J. J., Anderson, J. J., van Staden, P. P. and DuToit, A., 2000. Optimising
rainfall use efficiency for developing farmers with limited access to irrigation water.
Water Research Commission Report no. 878/1/00, Pretoria, South Africa.
211
Hensley, M., Le Roux, P. A. L., Du Preez, C. c., Van Huyssteen, C., Kotze, E. and Van
Rensburg, L. D., 2007. Soils; the Free State agricultural base. South African
Geographical Journal. 88: 11-21.
Iptas S. and Acar, A. A., 2006. Effects of hybrid and row spacing on maize forage yield and
quality. Plant and Soil Environment, 52: 515-522.
Joseph, F., 2007. Maize response to in-field rainwater harvesting on the Fort Hare/Oakleaf
Ecotope. MSc (Agric) Dissertation, University of the Free State, Bloemfontein, South
Africa.
Lascano, R. J., 1991. Review of models for predicting soil water balance. Soil water balance
in the Sudano-Sahelian Zone (proceedings of the Niamey Work shop) No. 199.
Maqbool, M. M., Tanveer, A., Ata, Z. and Ahmad, R., 2006. Growth and yield of maize (Zea
Mays L.) as affected by row spacing and weed competition durations. Pak. 1. Bot.,
38: 1227-1236.
Mzezewa, J. and van Rensburg, L. D., 2011. Effects of tillage on runoff from a bare clayey
soil on a semi-arid ecotope in the Limpopo Province of South Africa. Water SA 37: 1-
8.
Mzezewa, J., Gwata, E. T. and van Rensburg, L. D., 2011. Yield and seasonal water
productivity of sunflower as affected by tillage and cropping systems under dryland
conditions in the Limpopo Province of South Africa. Agricultural water Management,
98: 1641-1648.
Onyango, O. C., 2009. Decreased row spacing as an option for decreasing maize ( Zea may
L.) yield in Trans Nzoi district, Kenya. J. of Plant Breeding and Crop Science, 1: 281-
283.
Oweis, T., Hachum, A. and Kijne. J., 1999. Water harvesting and supplementary irrigation
for improved water use efficiency in dry areas. !WMl Contribution (No. 7), System
Wide Initiative on Water Management (SWIM). Colombo, Sri Lanka.
Prinz, D., and Malik, A. H., 2002. Runoff farming. Institute of Water Resources
Management, Hydraulic and Rural Engineering. Department of Rural Engineering,
University of Karlsruhe. Germany.
Stroosnijder, L., 1987. Soil evaporation; test of a practical approach under semi-arid
conditions. Netherlands Jour. of Agricultural Science, 35: 417-426.
Stroosnijder, L., 2009. Modifying land management in order to improve efficiency of
rainwater use in the African highlands. Soil and Tillage Research, 103: 247-256
212
Teh, C. B. S., Simmonds, L. P. and Wheeler, T. R., 2006. ModeJJing the partitioning of
evapotranspiration in a maize-sunflower intercrop. Malaysian Journal of Soil Science,
6. http://www.msss.com. 20/12/2011, 9.00 am (LT).
The World Bank, 2005. Shaping the future of water for agriculture; A source book for
investment in agricultural water management. Rural and Agricultural Development,
Washington D.C. USA.
van Rensburg, L. D. and Zerizghy, M,.G., 2008. The BEWAB+ computer program for
irrigation scheduling. Department of Soil, Crop and Climate Science, University of
the Free State, Bloemfontein, South Africa.
Westgate, M. E., Otegui, M. E. and Andrade, F. H., 2004. Physiology of corn plant. In: W.
Smith, C. Betrain, C. and J. Runge (eds). Corn; Origin, history, technology and
production. John Wiley and Sons, Inc., Hoboken, New Jersey.
Yeh, T. C. J., Gelhar, L.W. and Gutjahr, A. L., 1985. Stochastic analysis of unsaturated flow
in heterogeneous soils, 2. StatisticaJJy anisotropic media with variable xa. Water
Resources Research, 21: 457-464.
Yeh, T. and Harvey, D., 1990. Effective unsaturated hydraulic conductivity of layered sands.
Water Resources Research, 26:1271-1279.
Zhang, H., Oweis, T. Y., Garabet, S. and Pala, M., 1998. Water use efficiency and
transpiration efficiency of wheat under rain-fed conditions ad supplemental irrigation
in a Mediterranean-type environment.Plant and Soil, 201: 295-305.
213
CHAPTER9
INTEGRATING MICRO-FLOOD WITH IN-FIELD RAINWATER HARVESTING:
(ii) MAIZE YIELD AND WATER USE EFFICIENCY
Abstract
In this study micro-flood irrigation (MFI) and in-field rainwater harvesting (IRWH)
management practices were integrated for the first time. Its effect on maize yield and water
use efficiency (WUE) was tested using the dryland (DL), supplemental (SPI) and full
irrigation (FI) water regime (WR) treatments on the 1 m, 2 m and 3 m runoff strip width
(RSW) for IRWH. Four blocks with nine subplots were prepared to suit the complete block
3x 3 split plot factorial experiment. Each plot constituted of a standard 1 m basin strip and a
runoff strip that varied according to the RSW assignment. The basin and runoff strip were 30
m long and prepared using the respective basin and no till as required by the IRWH. Irrigated
plots received water from a 40 mm low pressured pipe calibrated to deliver 40L min -I. In
addition to the 281 mm seasonal rainfall a total depth of 105 and 173 mm irrigation was
applied per plot for the respective SPI and FI treatments. Seasonal changes in soil water
storage (~SWC), deep drainage (DD) and evapo-transpiration (ET) were determined.
Partitioning of ET into evaporation (Ev) and transpiration (T) was done using a simple
algorithm that relied on a ~ parameter for approximated crop canopy cover of the different
RSW. Results revealed that the 1 m RSW had the highest biomass and grain yields that were
respectively 19% and 32% above average. Correspondingly, seasonal ET and T were 10 %
and 44% above average. Total biomass and grain yield WUE based on T confirmed the 1 m
RSW as an optimum RSW although the 2 m RSW a comparable biomass WUE to become an
alternative. Shading from plant canopy aided in suppressing Ev and enhanced light
interception for photosynthesis as well as increased surface area that is well rooted for
optimum water and nutrient uptake were the main reasons. Introduction of irrigation through
MFI also improved crop yields and WUE with the FI producing biomass and grain increases
of about 200 % higher than DL. It was therefore concluded that the 1 m RSW and FI would
produce optimum maize yields and WUE for the newly integrated MFI-IRWH soil water
management system under the climate and soil conditions that prevails at Parady's
Experiment Farm of the University of the Free State, South Africa.
Key words: Dryland, irrigation, runoff strip width, in-field runoff, evapo-transpiration
214
9.1 Introduction
Increased innovation of conservational soil water management practices has made sustainable
food production possible especially under difficult climate and soil conditions. Among these
attractive technologies is the micro-flood irrigation (MFI) (Austin, 2003) and in-field
rainwater harvesting (IRWH) (Hensley et al., 2000). To exploit the water productivity of the
former and conservational potential of the latter, the two management practices have been
newly integrated in this study.
Micro-flood irrigation is a typical short furrow irrigation method developed to optimise water
productivity on soils with dry soil water regimes. Benefits of irrigation especially under
supplemental regime in stabilizing crop growth and yields under prolonged dry spells are
well covered in literature (Oweis et al., 1999; Rockstrom, 2000; Walker, 2003; Rodriguez et
al., 2004; World bank, 2005; Moult et al., 2009). However, the efficiency of flood and furrow
irrigation methods has always come under scrutiny given its poor distribution uniformity and
unproductive losses through deep drainage and furrow end runoff. To improve application
and distribution efficiencies developers of MFI (Austin, 2003) used small inflow rates to
match soil water basic intake with advance time. On conventional furrow systems this was
normally achieved by using very large inflow rates with cyclical or surge flows and
automated cutback systems (Walker, 2003; Rodriguez et al., 2004). Attainment of
distribution uniformity that were competitive to pressured systems as high as 95% by MFI
were recorded when a 40 L min-I inflow rate was used on a 20 m furrow run prepared on
sandy loam soils (Austin, 2003). High application and water use efficiencies from other
forms of micro-flood systems especially those integrated with rainwater harvesting were
reported in various East African and South Asian countries (Awulachew et al., 2005; World
Bank, 2005; Rockstrom et al., 2007).
Developed for soils that produce appreciable surface runoff and evaporation losses under
conventional tillage system, IRWH has become popular in soil water conservation especially
in South Africa (Botha, 2006; Manona and Baiphethi, 2008; Hensley et al., 2011; Mzezewa et
al., 2011). The soils that have impermeable surface crust formed by raindrop impact (Hensley
et al. 2007), low organic matter (Stroosnijder, 2009) and swelling clay either in the B or C-
horizons (Hensley et al., 2000) have a persistent dry soil water regime (Hensley et al., 2007).
However, IRWH was able to turnaround the negative influences of these soil characteristics
into high in-field runoff, deep infiltration and improved soil water regime for crop growth
215
and natural nutrient cycling as well. Crop yields and water use efficiency were respectively
up to 50 % and 106 % higher to conventional tillage (Botha, 2006; Mzezewa and van
Rensburg, 2011).
Despite the improvement in efficient water use by the MFI and IRWH, high evaporation
losses associated with irrigation (Annandale et al., 2011) and in situ rainwater harvesting
(Both et al., 2006) remains a challenge. Developers of MFI claims that the use of
unconventionally short furrows and small inflow rates as small as 20 L min-I reduce above
surface stream flow volumes but encourage sub-surface flow which minimise evaporation
losses. On IRWH fields evaporation losses as high as 75 % of annual rainfall has been
reported even though in-field runoff was encouraged to infiltrate deep into the profile below
the evaporation sensitive zone (Bennie et al., 1994; Hensley et al., 2000; Mzezewa et al.,
2011). Evaporation competes directly with transpiration that is related to crop growth and
yields. Therefore, soil water management practices that reduce the former would
simultaneously increase the latter (Stroosnijder, 1987). Even though this could be the case
most of the transpired water is used for cooling purposes and a negligible amount is retained
by the crop for crop growth (Schneekloth and Andales, 2009). Therefore, the amount of water
used to produce a kilogram of yield (food) becomes critical for evaluating soil water
management practices designated for semi-arid environments.
Different approaches of suppressing evaporation especially on IRWH have been attempted.
One ground breaking work was the testing of different mulching materials on the basin and
runoff area (Botha, 2006). Organic mulches had the most promising results with evaporation
suppressed by 15% on clay soils and 31% on fine sandy loam soils compared to bare soils. In
the absence of supplemental irrigation the conservational benefits of mulches were eroded by
dry spells that exceeded 14 days (Botha, 2006). Intereropping with a legume on IRWH was
also found to suppress evaporation and this was attributed to increased plant density
(Mzezewa et al., 2011). Application of mulches and increasing plant density are among the
cultural and agronomic practices that encourage ground cover (Magbool et al., 2006) and thus
substitute evaporation for transpiration (Grena and Hess, 1994).
Other well established agronomic practices includes narrow row spacing (Eberbatch and Pal a,
2005), irrigation (Zhang et al., 1998) and fertiliser application (Stroonijder, 2009). Despite
their equal significance in encouraging early crop growth and canopy cover the first two have
an important bearing in the integration of MFI and IRWH. Irrigation increases water
216
availability for crop water use, but if not properly applied evaporation could be exacerbated.
Pilbeam et al. (1995) reported a scenario where irrigation on a bean crop increased
evaporation by more than 55 %, while only 10 to 19 % of the applied water was transpired.
Transpiration and evaporation are expected to vary with row spacing because of the variation
in surface area shaded by the crop canopy cover (Kumar et aI., 1988). This is more pertinent
to the IRWH with two row spacing; the 1 m for basin strip and 2 m for runoff strip width
(RSW). The latter could be varied with change in seasonal rainfall and field runoff
coefficient. For dryland conditions the 2 m RSW was considered as standard for most soils
earmarked for IRWH especially from the areas receiving annual rainfall of 550 mm (Hensley
et aI., 2000). Given the erratic nature of semi-arid rainfalls varying the 2 m RSW to optirnise
crop water use would subject the surface storage structures into more risk than the use of
alternative methods such as mulches and intercropping. However the integrating of MFI with
IRWH may necessitate the establishment of optimum RSW for certain irrigation regimes. It is
against this background that the study specific objective was to determine the effect of RSW
and water regime (WR) on maize production.
9.2 Material and methods
9.2.1 Description of experimental site
The field experiment was carried out at Parady's Experiment Farm (29°13'24, 69"S, 26°12'40,
93" E, altitude 1422 m) of the University of the Free State, South Africa. The area has an
average annual rainfall of 550 mm, mainly from October to April. December and January are
the hottest months of the production season with respective average maximum and minimum
air temperatures of 30 and 14°C and 31 and 15 °C. The selected site had two predominant
soil forms, the Tukulu and Swartland (Soil Classification Working Group, 1991) that are
similar to the respective Cutanic Luvisols and Cutanic Cambisols of the World Reference
Base for Soil Resources (1998). The corresponding soil profile description is summarized in
Table 9.1. The site has been used for maize cultivation under conventional tillage for the past
decade.
9.2.2 Experimental design and layout
A complete block 3 x 3 split plot factorial experimental design was used. The two main
treatments were the runoff strip width (RSW) a special form of row spacing adapted for
IRWH plots and water regime (WR) management factors. The RSW had the 1 m, 2 m and 3
217
m as sub treatments while the WR constituted of the dryland (DL), and micro-flood irrigation
in the form of supplemental (SPI) and full irrigation (Fr).
Table 9.1 Summary of the physical and chemical characteristics of the Tukulu and Swartland
soil types found in the experimental area.
Soil Eh~sicalEroEerties
Soil forms Tukulu Swartland
Master Horizons A Bl C A Bl C
Coarse sand (0/0) 5.3 9.2 2.1 4.7 3.2 54.3
Medium sand (0/0) 9.3 8.8 3.8 7.6 5.3 4.6
Fine sand (%) 41.2 31 28.3 42 37.6 17.2
Very fine sand (0/0) 25.3 21 8.4 31.7 26.6 2.5
Coarse silt (0/0) 2.1 2 3 2 3 3
Fine silt (0/0) 4.6 2.5 6.5 2 3
Clay (0/0) 11.3 26.4 47.9 11.3 21.9 IS
Bulk density (kg
m") 1670 1597 1602 1670 1530 1450
Porosity (0/0) 34 33 32.4 35 39.9 41.6
Ks (mm hour -I) 36.1 40 9.6 23.5 42.8 76.5
Chemical EroEerties
pH (water) 5.8 6.1 6.5 5.5 5.8 6.8
Ca (c mol.ikg") 35.3 40.8 62.7 23 25.78 45
Mg (c mol.Jcg') 18.8 20.8 54.7 12.1 13.1 25.4
K (c mok.kg") 12.8 6.8 12.9 5.5 7.45 7.7
Na (c mol..kg") 2.4 2.5 3.4 0.7 1.52 2.3
Four blocks with nine subplots were prepared to suit the 3x3 split plot factorial. Basin tillage
was used to prepare the basin area (BA) of about 0.5 m width and 30 m long running on a
zero degree slope. Runoff strip width between alternative rows also referred as the runoff
area (RA) in this study was treated with no till to serve as mini-catchments for rainwater
harvesting. Each plot had five basins with the peripheral basins serving as plot boarders. A 15
m buffer zone was maintained between blocks with a 5 m strip between unit subplots. On
plots that received irrigation treatment the basins were levelled using a furrow ridger to
ensure a smooth advance flow. The basins to serve as furrows were also close ended to detain
runoff within plots and to comply with the zero ex-field runoff of IRWH.
218
9.2.3 Planting and agronomic practices
A 120 day yellow maize cultivar (PAN 6236 B) was used for the experiment. Plant density
and intra row spacing was determined by soil water regime and runoff strip width as outlined
in Table 9.2 To ensure accurate intra row spacing, planting on straight lines was carried by
hand using a calibrated string. Basal and top dressing fertilizers were applied as required for
maize production under varying soil water regimes.
Table 9.2 Intra-row plant spacing at planting time (mm).
Water regime Plant Runoff strip width (m)
density 1 2 3
Dryland 25000 363 266 200
Supplemental 40000 227 67 125
Full 70000 130 95 71
A non selective herbicide was used to control weeds before planting on the RA and BA.
Thereafter it was applied following a new growth of weeds appeared on the experimental
plot. The field trial was conducted only for one season.
9.2.4 Irrigation design and application
The short furrow irrigation used for the supplement and full irrigation water regimes was
designed and tested on previous chapters. Low pressure pipes of 40 mm diameter were used
to deliver 80 mm of water per minute to individual 30 m length furrows at a distribution
efficiency of 85-91 %. A 333 and 190 mm seasonal water depth was applied using a seven
day cycle for the full and supplemental water regime. Individual furrows were wetted for an
average time of 15 minutes for 16 weeks from the day of planting (25 November 2007).
According to the BEW AB+ (van Rensburg and Zerizghy, 2008) irrigation scheduling model
an amount of about 625 mm of water would be required for a 120 day maturing maize
cultivar to attain a maize yield of 10000 kg ha-I. Given that more than 50% of the long term
average rainfall is received between the months of November to March, a seasonal rainfall of
around 300 mm was expected. Therefore, the balance was provided by means of the micro-
flood irrigation using a fixed schedule for the FI and SPI produced by the BEW AB+
irrigation software. For the 30 m furrow run the 40 L min-I inflow rate from previous test
showed that advance times between 20 to 30 minutes could attain infiltrated depths of 30
mm. For FI the depth applied per plot was 173 mm and SPI was fixed at 60 % of FI. The
219
designated 40 L min-I inflow at 100 % application efficiency could supply 0.9 m3 or 900
litres of water on the 30 m furrow run at a cut-off time of 23 minutes.
9.2.5 Field measurements
Soil water content and climatic parameters were measured on site during the experiment.
Individual subplots were installed with two neutron access tubes. Each was installed in the
central position of the RA and BA at 15 m distance of plot length. Access tubes were
installed to depths corresponding to the effective root depth of individual subplot soil profile.
The neutron water meter (NWM) was used to measure the soil water content (SWC). Initial
intentions to measure SWC before and after rainfall and irrigation events proved not to be
practical and measurements were then taken after rainfall once drainage and the rapid phase
of Ev has occurred. Rainfall, temperature, and relative humidity as well as potential
evaporation data was obtained from the local automated weather station. Distribution of
seasonal rainfall or precipitation (P) including the accumulative (Ace.) depth of P, SPI+P,
and FI+P is illustrated in Figure 9.1. Soil water content and climatic data was used to develop
the soil water balance (SWB) functions.
400 _ P ~ Acc. P --+-- Acc. SPI+P -I!r- FI+P 480 ,-.,E
360 _,I:. 440 E~-«' '-"
320 "Jr 400 ..c;' ,,~ ....360 0..,-., 280 ...6 -4---~
Q)
E -0.,..6 - Il:'" ,ft'" 320 ....E
'-" 240 _"A- -Ir ,A" 280 .
Q
c..
)
e.
t: If' q __<V'
.;0:; 200 -~ ~.,J! ,,/.c..e. h_-J;--..(T -- 240 Q)>'5. 160 ,If ,. ,,' 200 .;:;
'u 120 160
ce
e / "A" ::s120 E0... 80 80 ::su
40 u40 -<
0 0
7 14 21 39 47 53 60 67 71 78 84 87 99 105 109 116
Days after planting
Figure 9.1 Seasonal precipitation distribution and its accumulative totals with supplemental
(SPI) and full irrigation (FI).
9.2.6 Estimation of soil water balance components
The change in SWC (~SWC) of the profile for a given period during the production season
was mathematically described as;
220
~swc = (P + I) - DD - ET (9.1)
Where, P is rainfall amount (mm), I is irrigation amount (mm), DD is deep drainage (mm)
and ET is evapo-transpiration (mm). Due to spatial variation of the soil profile layers DD
from the Tukulu prismatic and Swartland saprolite C-horizons were respectively estimated
using the exponential expressions, respectively written as:
DD= 4 x10-63exp44o.s.o (9.2)
DD= 2 x10-8exp67.21.0 (9.3)
Evapo-transpiration was determined from the re-arranging the expression (1) to become:
ET = (I + P) - Dd - ~SWC (9.4)
Where, ~SWC is difference in profile average SWC between two consecutive measurements.
Partitioning of ET into Ev and T components was carried using the extent of canopy cover
during the four growth stages (plant establishment, vegetative, reproductive and ripening)
proposed by the FAO (2011). Plant height measured at the end of plant establishment and
when plants has interlock leaves between neighbouring rows and was linearly translated to
the extent of canopy cover. A ~-parameter was used ranging from 0 to 0.9 for the BA and 0 to
0.5 for the RA especially from the 2 and 3 m RSW. The 1 m RSW was treated the same way
as the BA. Transpiration (T) and evaporation (Ev) was related to ET using the expressions:
T = f3TE (9.5)
Ev = (1 - f3)ET (9.6)
9.2.7 Crop yields and biomass
Above ground biomass and grain yield were the pnmary crop evaluative indicators to
treatments responses. Given that the integrity of the crop stand was affected by a rainstorm
during the ripening stage, a 9 m2 was demarcated per sub plot and the number of plants and
cobs were counted and sampled for yield and biomass analysis. Suckers and primary maize
stalks were separated and weighed. A representative of five cobs from the sample was oven
dried to 12% moisture content, shelled and seeds weighed. Grain yield was expressed as kg
ha" by equating the seed mass from the sampled plants to the corresponding plant density per
unit hectare. In the same manner the representative biomass weight was expressed in kg ha-I.
221
9.2.8 Water use efficiency
Water use efficiency (WUE) was calculated for total biomass (WUEb) and gram yield
(WUEg). To have a more meaningful on the productivity of the different management
practices biomass or grain yield was expressed as a ratio to seasonal T.
9.2.9 Statistical analysis
A split plot analysis of variance was used to establish the effect of RSW and WR and its
interaction on seasonal water components, maize yield and water use efficiency. The SAS 9
software was used in this endeavour in combination with the procedures for least significant
different (LSD) test described by Gomez and Gomez (1984).
9.3 Results
A summary of analysis of variance depicting the effect of RSW and WR on seasonal SWB
components, biomass and grain yield and WUE of biomass and grain yield are presented in
Table 9.3 and 9.4. Variable was the effect of RSW and WR on the water, yield and WUE
parameters. The profile SWC was not affected by the RSW and WR factors, while £l.SWC
was the only component affected by these factors interaction.
Water components:
Runoff strip width (RSW) and its interaction with WR affected the £l.SWC. Seasonal net gain
ranged from 17 to 37 mm with the amount gained decreasing with the RSW. Net gain
differed among the WR factors ranging from 7 to 40 mm, decreasing with water
unavailability. The effect of interaction on £l.SWChad the highest net gain of 51 mm from the
1 m RSW of the FI. The same RSW had a net loss of -24 mm but from the DL. The 2 and 3 m
RSW for the different WR had a £l.SWC that ranged from 11 to 38 mm with net gain
decreasing from FI to DL. Seasonal deep drainage was only affected by WR with the highest
amount of 26 mm recorded from FI and followed by the SPI and DL that had comparable
values of 15 and 9 mm, respectively. Evapo-transpiration had a significant respond to RSW
with the highest value of 399 mm from the 1 m RSW. The 2 and 3 m RSW had comparable
respective ET values of 348 and 338 mm. Evaporation was affected by RSW recording a
range of 118 to 203 mm, increasing with widening of the RSW. On the contrary, T decreased
with increasing RSW having a range of 135 to 281 mm. Effect of FI and SPIon T was
comparable higher with respective values of 230 and 219 mm while DL had 138 mm.
222
Table 9.3 Summary of analysis of variance depicting the effect of runoff strip width (RSW)
and water regime (WR) on seasonal soil water components, crop yield and water use
efficiency (WUE).
Seasonal parameters Factors End of season
Water components
RSW ns
Profile SWC WR ns
(mm mm") RSWx WR ns
RSW *
8" SWC (mm) WR *
RSWx WR *
RSW ns
Deep Drainage (mm) WR *
RSWxWR ns
RSW *
Evapo-transpiration (mm) WR ns
RSWx WR ns
RSW *
Evaporation (mm) WR ns
RSWx WR ns
RSW *
Transpiration (mm) WR *
RSWx WR ns
Crop yield
RSW *
Total Biornass WR *
(Kg ha") RSWx WR ns
RSW *
Grain yield WR *
(Kg ha-I) RSWx WR ns
Water use efficiency
RSW *
Total biomass WR *
(Kg ha-I mm") RSWx WR ns
RSW *
Grain yield WR *
(Kg ha-I mm") RSWxWR ns
* = significant at 5% probability level, ns= not significant
223
Table 9.4 Effect of runoff strip width and water regime on seasonal water components and
crop yield.
Grain Grain yield
~SWC DD ET Ev T Biomass yield Biomass WUE WUE
Treatments (mm) (mm) (mm} (mm) (mm) ( Kg ha") (Kg ha·l (Kg ha" mrn') (Kg ha·1) mm
A. Runoff strip width
(m)
I 17.3b 15.2 398.9a 117.6c 28 J.3a 15665" 7535" 38.2" 18.4a
2 2l.9b 17.0 347.7b I77.6b 170.lb 12892b 557i' 35.8"b 15.4b
3 37.2" 17.4 338.4b 203.la 135.3b 10791c 3986c 30.9b II.4c
LSD(T:50.05) 15.5 ns 24.0 24.5 38.2 1779 730 5.7 2.6
CV(%) 49.8 48.0 5.4 12.1 16.0 11.1 10.5 13.3 14.3
B. Water regime
DL 7.4c 8.7b 296.5 158.9 137.6b 7124c 301 IC 24.0b 10. IC
SPI 28.6b 15.4b 378.5 159.5 219.0" 12413b 5333b 32.7b 13.9b
FI 40.4a 25.5a 410.0 179.8 230.2" 19812" 8754a 48.2a 21.2a
LSD(T:50.05) 10.3 8.4 ns ns 36.5 3465 1332 9.3 3.6
CV(%) 66.4 84.1 34.8 43.2 30.8 43.6 38.5 43.9 39
C. Interaction
DL -24.0c 5.4 312.9 113.0 199.9 8650 4146 27.7 13.3
Im SPI 25.4"b 19.3 429.6 106.6 323.0 14377 6762 33.5 15.7
FI 50.6a 20.9 454.3 133.1 321.1 23967 11697 53.3 26.1
DL 10.5b 12.4 284.5 163.0 121.5 6895 2792 24.3 9.8
2m SPI 22.3ab 10.1 361.7 177.8 183.9 12366 5496 34.2 15.2
FI 33.0ab 28.5 396.9 191.9 205.0 19416 8442 49.0 21.3
DL 35.5ab 8.4 292.2 200.8 91.3 5826 2095 19.9 7.2
3m SPI 38.3ab 16.7 344.1 194.0 150.1 10495 3741 30.5 10.9
FI 37.6ab 27.2 378.9 214.5 164.4 16052 6124 42.3 16.1
LSD (T:50.05) 33.2 ns ns ns ns ns ns ns ns
~SWC = change in soil water content, 00= deep drainage, ET= evapo-transpiration, Eveevaporation,
Tetranspiration, DL= dryland, SPI =supplemental irrigation, FI= full irrigation, WUE= water use efficiency,
ns= not significant
Maize yield:
Seasonal biomass and grain yield were affected differently by the RSW and WR factors even
though the resulting interaction was insignificant. Biomass yield ranged from 15665 to 10791
kg ha-I, increasing with decreasing RSW. Water regime resulted in biomass that was high as
19812 kg ha" from the Fl, followed by 12413 and 7124 kg ha-1 for the respective SPI and
DL. Grain yield ranged from 7535 to 3986 kg ha-1 for the RSW with the highest from the 1 m
RSW. The WR had grain yield of 8754 kg ha-1 from the FI, 5333 kg ha-1 from the SPI and
3011 kg ha-1 from the DL.
224
Water use efficiency:
Seasonal WUE for biomass and grain yield were affected by RSW and WR differently. The
1 and 2 m RSW had comparably higher WUEb with respective mean values of 38 and 36 kg
ha-I mm-I. The 3 m RSW recorded the lowest value of 31 kg ha-I mm', although it was not
different from that of the 2 m RSW. The WUEb from the WR ranged from 48 to 24 kg ha-I
mrn' with the FI having a significantly higher mean value. The 1 and 2 m RSW affected the
WUEg in a comparable way with respective mean values of 38 and 36 kg ha-I mm-I. The 3 m
RSW had a value of 31 kg ha-I mm' different from the 1 m RSW but comparable with the 2
m RSW. The WUEg were all significantly different with FI, SPI and DL recording 21, 14 and
10 kg ha-I mm', respectively.
9.4 Discussion
Improving crop yields under limited water supplies is a desirable attribute for semi-arid soil
water management practices. The newly integrated MFI with IRWH had three WR (Dryland,
supplemental and full irrigation) and RSW (1 m, 2 m and 3 m) investigated for their effect on
maize production. Carried out on a single season of 200712008, total biomass and grain yield
were affected differently by the WR and RSW.
Traditionally the IRWH was developed to improve crop yields and water use efficiencies
under dryland conditions using the 2 m RSW as a standard row spacing (Hensley et al.,
2000). The DL maize yields including total biomass and grain yields fell below the range of
those reported by Hensley et al. (2000) under IRWH. Despite the differences in local climates
and soil types, the below average (51%) seasonal rainfall was attributed to the low yields
obtained from this study. However, the results from Hensley et al. (2000) were reflective of
the high productive potential of IRWH especially when compared to conventional systems
under optimum rainfalls.
Among the RSW the 2 m RSW recorded a total biomass that was 19% higher than from the 3
m RSW and 18% lower than from the 1 m RSW. It also had a grain yield that was 40%
higher and 26 % lower than from the respective 3 m and 1 m RSW. The crop yields showed a
decrease with an increase in RSW and the 2 m RSW being an intermediate RSW fell within
the average. This result reflects that the initial selection of the 2 m RSW for IRWH was
reasonable. Under the conditions that seasonal DD and runoff was insignificant, the total
water use (ET) loss from the soil plant system could give an insight on the differences on
225
yields among the RSW. Seasonal ET followed same trend as crop yields with the 2 m RSW
being 3% higher and 12% lower than the respective 3 and ] m RSW. When ET was
partitioned into Ev and T, the proportion of Ev to ET was 29%, 51% and 60% for the 1 m, 2
m and 3 m RSW, respectively. In comparison with results reported by Hensley et al. (2000)
for three consecutive seasons, the total Ev mainly from the 2 m RSW constituted a proportion
ranging from 67 to 84% of seasonal ET that was also within the same range with those
reported by Botha (2006). Given that Ev is defined as soil water loss from the surface directly
to the atmosphere, suggest that increasing RSW encouraged Ev by increasing the surface area
exposed to direct radiation; a sentiment that is well supported in literature (Sangoi et al.,
2001; Onyango, 2009; Acciaresi and Zulunga, 2006).
On the contrary, reducing the RSW implied that much of the surface area becomes shaded by
the crop canopy and hence more light get intercepted to promote photosynthesis rather than
evaporation (Eberbach and Pala, 2005). Maize plants are appreciated for their tendency to
interlock leaves much earlier during the vegetative growth phase (FAO, 2011). In this regard,
the direct relationship between photosynthesis and T was translated to higher crop yields
especially from the 1 m RSW. Seasonal T from the 1 m RSW increased by 65% and 100%
over and above that from the respective 2 and 3 m RSW. Reducing row spacing for the same
plant density was also acknowledged for improving water and nutrient uptake through
increased surface area accessible to crop plants roots (Onyango, 2009) a phenomenon that is
considered to have led to the higher yields from the 1 m RSW despite the low seasonal net
gains.
Introducing irrigation either by supplemental or full irrigation through the integration of MFI
with IRWH resulted to improved yields and water use efficiencies. This was expected given
that maize yields generally increase linearly with water application (Schneekloth and
Andales, 2009; Annandale et al. 2011). Total biomass from the SPI and FI was respectively
higher than DL by 74% and 200% and the corresponding grain yield that was 77% and 200%
higher. Similar yield trends on maize (Bello, 2008), wheat (Zhang et al., 1998) and Tomatoes
(Agele et al., 2011) have been reported. More insight could be shared by the seasonal ET and
its constituents whereby SPI and FI had an increase of about 28% and 38% respectively
higher than DL. In another study irrigated maize had a higher seasonal ET than rain-fed
maize by of 12% while for irrigated wheat it was 5% higher than rain-fed (Suyker and
Verma, 2009). But the question of how much of seasonal ET was translated to crop growth
was inversely shown by the proportion of Ev that was about 42% and 44% of seasonal ET
226
from the respective SPI and FI. This was not strange in irrigation given that water is usually
applied during dry spells when atmospheric evaporative demand was fairly high.
The question of water-use efficiency (WUE) then becomes important if water availability in
IRWH has to support plant growth and yields rather than unproductive losses (Ev and DD).
Given that plant growth and yield is directly a function of T, therefore a more meaningful
judgement of WUE is obtained in reference to T as a measure of crop performance
(Stroosnijder, 1987; Zhang et aI., 1998; Eberbach and Pala, 2005). Increasing RSW decreased
WUEb while increasing water availability had the opposite effect. The 2 m RSW compared
with both the 1 and 3 m RSW suggesting its ability to exploit the benefits of in-field runoff
and partial crop canopy cover to sustain a reasonable crop growth. This productive potential
of the 2 m RSW was also appreciated by Hensley et al. (2000). Full irrigation had the highest
WUEb despite significant DD losses that was approximately 10.1 mm and 17 mm higher than
that from the respective SPI and DL. Interestingly, the DL and SPI had comparable WUEb
and could be attributed to the similarities in DD losses (Table 9.4). Hens1ey et al. (2000)
noted that management practices that result to same DD usually have homogeneous WUE.
The World Bank (2005) pointed out that management practices that encourage some level of
water stress maximise the benefits of water applied. This could be realised through increasing
soil water storage capacity and rainfall capture by in-field runoff from DL and deficit
irrigation with SPI.
Regarding WUEg, significant differences between RSW and WR were observed that were
consistent with the variations in grain yield and corresponding net gains. The 1 m and 2 m
RSW had WUEg that was higher than the 3 m RSW by respectively 64% and 35% suggesting
that the amount of water required to produce a kilogram of dry matter was higher with
increasing RSW. This was justified by the high claims Ev from the RSW with greater surface
area exposed to direct radiation (Aberbach and Pal a, 2005; Onyango, 2009). In comparison
the WUEg from all the RSW were fairly lower than those recorded from Hensley et al.,
(2000). Different climatic conditions and the application of a different method (Tanner and
Sinclair (1983) method to partition Ev and T could be the reason. Despite being developed
under different climatic conditions it also rely on the transpiration efficiency that is generally
an insensitive to water stress and water availability (Morgan et al., 2003) which could lead to
overestimates when used in semi-arid areas. However the Ev/T ratio which is a critical
indicator of WUE (Zhang et al., 1998) was 42%, 105% and 150% for the 1 m, 2 m and 3 m,
reflecting an increase with RSW, justifying the decline in WUEg with widening of the RSW.
227
The same ratio (Ev/T) decreased with water availability confirming the 100% and 40% rise in
WUEg from the respective FI and SPIover and above that from the DL. Syker and Verma
(2009) also recoded similar results whereby the WUEg for DL lagged behind that from
irrigation by 25%.
However, prospects of achieving reasonably high yields and WUE from the DL and SPI
could be inferred from the RSW and WR interaction that affected the change in soil water
storage (.6SWC). From this interaction the net gains were apparently similar for all SPI
irrespective of RSW and these were comparable with the FI of the 2 and 3 m RSW and DL
from the 3 m RSW. The highest net gain from the 1 m RSW under FI was reflective of the
potential of this RSW of substituting Ev for T under full plant canopy cover that could shade
up to 90% of the surface area that is also concentrated with plants roots.
9.5 Condusion
Integration of MFI with IRWH was tested for its effects on maize yield and WUE using
different RSW (1 m, 2 m and 3 m) and WR (DL, SPI and FI). The effect of the RSW and
WR on seasonal water and yield components and biomass and grain yield WUE that was
based on T was reflective of the productive potential of the newly integrated management
practices.
The 1 m RSW had optimum total biomass and grain yields that were higher than the 2m and
3m RSW by 21 % and 45 % and 35 and 89 %, respectively. This was supported by the
seasonal high ET with an Ev/T ratio that increased with RSW with that from the 1 m RSW
being 31 % lower than the 3 m RSW. This was because the 1 m RSW had plant canopy
shading the ground much earlier to suppress Ev and had much of its surface area concentrated
with plants roots that enhanced water and nutrient uptake. Confirming the 1 m RSW as the
optimum RSW was the WUE from the biomass and grain yield that were respectively above
average by 9% and 22%. However, the 2 m RSW presented itself as viable alternative with a
competitive biomass WUE that was 2.3 % above average.
Introducing irrigation through MFI also improved yield and WUE by increasing soil water
storage net gains. Full irrigation had the highest biomass and grain yields that were about
200% higher than the DL while SPI had an average increase of 76 % for both yield
components. The benefits of FI were supported by the high biomass and grain WUE even
though the SPI also competed fairly with the T component falling in the range with the FI.
228
Nevertheless, the application of FI with the 1 m RSW could be said to produce optimum
yields and WUE for the newly integrated MFI-IRWH soil water management system.
Acknowledgements
Acknowledgements directed to the technical team at Kenilworth Experimental Station of the
University of the Free State for their assistant in drying sampled maize plants for dry matter
determination.
229
9.6 References
Acciaresi, H. A. and Zulunga, M. S., 2006. Effect of plant row spacing and herbicide use on
weed above ground biomass and corn grain yield. Planta Daninha Vicosa-MG, 24:
287-293.
Agele, S.O., Iremiren, G.O. and Ojeniyi, S.O., 2011. Evapo-tran piration, water use
efficiency and yield of rain-fedand irrigated tomato. Inter. J. of Agric. and Bio., 13:
469-476.
Annandale, J. G., Stirzaker, R. J. Singles, A., van der Laan, M. and Laker, M. C., 2011.
Irrigation scheduling research: South African experiences and future prospects. Water
SA, 37: 751-762.
Austin, C, 2003. Micro flood, a new way of applying waters. http://waterright.com.
22/09/211, 10.00 am (LT).
Awulachew, S. B., Merrey, D. J., Kamara, A. B., Koppen Van, B., de Vries, P. and Boelee
E., 2005. Experiences and opportunities for promoting mall cale/micro irrigation
and rainwater harvesting for food security in Ethiopia. International Water
Management Institute, Working Paper 98, Addis Ababa, Ethiopia.
Bello, W. B., 2008. The effect of rain-fed and supplemetal irrigation on the yield and yield
components of maize in Mekelle, Ethopia. Ethopian J. of Env, Studies and
management, 2: 1- 7.
Botha, J. J., 2006. Evaluation of maize and sunflower production in a semi-arid area using In-
field rainwater harvesting. Ph.D. (Agric) Dissertation, University of the Free State,
Bloemfontein, South Africa.
Eberbatch, P. and Pala M., 2005. Crop row spacing and its influence on the partitioning of
evapo-tran piration by winter- grown wheat in Northern Syria. Plant and Soil,
268:195-208.
FAO, 2011.Chapter 5; Introduction to crop evapo-transpiration. http://www.fao.org.
20/11/2011,10.30 am (LT).
Gomez, K. A. and Gomez, A. A., 1984.Statistical procedures for agricultural research. 2nd ed.
John Wiley and Sons, New York, USA.
Grena, A. K, and He ,T.M., 1994. Water balance and water use of pearl millet-cow pea
intercrops in north east Nigeria, Agricultural Water Management, 26: 169-185.
230
Hensley, M., Bennie, A. T. P., van Rensburg, L. D. and Botha, J. J., 2011. Review of plant
available water aspects of water use efficiency under irrigated and dryland conditions.
Water SA, 37: 771-779.
Hensley, M., Botha, J. J., Anderson, J. J., van Staden, P. P. and DuToit, A., 2000. Optimising
rainfall use efficiency for developing farmers with limited access to irrigation water.
Water Research Commission Report no. 878/1100, Pretoria, South Africa.
Hensley, M., Le Roux, P., Gutter, J. and Zerizghy, M. G., 2007. A procedure for an improved
soil survey technique for delineating land suitable for rainwater harvesting. WRC
Report No. TT 3IlI07.Water Research Commission, Pretoria, South Africa.
Kumar, K. K ., Dantas, R. C. D. and Bezerra, V. F. 1988. The effect of row spacing on
Evapo-transpiration patterns in a crop field. Department of Atmospheric Sciences,
Federal University of Paraiba, Campida Grande-PB.
Manona, S. and Baiphethi, M. N., 2008. Developing a land register and a set of rules from
application of in-field rainwater harvesting in three villages in Thaba Nchu. WRC
Report No.TT367/08. Water Research Commission, Pretoria, South Africa.
Maqbool, M. M., Tanveer, A., Ata, Z. and Ahmad, R., 2006. Growth and yield of maize (Zea
Mays L.) as affected by row spacing and weed competition durations. Pak. J. Bot., 38:
1227-1236.
Morgan, C. L. S., Norman, J. M. and Lowery, B., 2003. Estimating plant available water
across a field with an inverse yield model. Division S-6; Soil and Water Management
and Conservation, Soil Sci. Soc. Am. J., 67: 620-629.
Moult N.C., Lecler N. 1., and Smithers J. C., 2009. A catchment-scale irrigation systems
model for sugarcane; Model application. Water SA, 35: 29-36.
Mzezewa, J., Gwata, E. T. and van Rensburg, L. D., 2011. Yield and seasonal water
productivity of sunflower as affected by tillage and cropping systems under dryland
conditions in the Limpopo Province of South Africa. Agricultural water Management,
98: 1641-1648.
Onyango, O. C., 2009. Decreased row spacing as an option for decreasing maize (Zea may
L.) yield in Trans Nzoi district, Kenya. J. of Plant Breeding and Crop Science, 1: 281-
283.
Oweis, T., Hachum, A. and Kijne. J., 1999. Water harvesting and supplementary irrigation
for improved water use efficiency in dry areas. IWMI Contribution (No. 7), System
Wide Initiative on Water Management (SWIM). Colombo, Sri Lanka.
231
Pilbeam, C. J., Simmonds, L.P., Kavilu, A.W., 1995. Transpiration efficiencies of maize and
beans in semi-arid Kenya. Field Crops Research, 41:179-188.
Rockstrom, J., 2000. Water resources management in smallholder farms In Eastern and
Southern Africa: An overview. Phys. Chemo Earth, 25: 275- 283.
Rockstrom, J., Hatibu, N., Oweis, T. Y. and Wani, S., 2007. Managing water in rain-fed
agriculture, Part 4 In: Unlocking the potential of rain-fed agriculture, !WMl,
Colombo, Sri Lanka.
Rodriguez, J. A, Dïaz, A., Reyes J. A. and Pujols R., 2004.Comparison between surge
irrigation and conventional furrow irrigation for covered black tobacco cultivation in
a Ferralsol soil. Span J. Agric. Res., 2: 445-458.
Sangoi, L., Ender, M., Guidolin, A. F., de Almeida, M. L. and Heberle, P. C., 2001. Influence
of row spacing reduction on maize grain yieldin regions with a short summer.
Pesq.agropes bras., 36:861-869.
Schneekloth, J. and Andales, A., 2009. Irrigation: seasonal water needs and opportunities for
limited irrigation for Colorado crops. Department of Agriculture, Colorado State
University. Colorado, USA.
Soil Classification Working Group, 1991. Soil Classification - A taxonomic system for South
Africa. Memoirs on the Agricultural Natural Resources of South Africa Nr.IS.
Pretoria: Department of Agricultural Development.
Stroosnijder, L., 2009. Modifying land management in order to improve efficiency of
rainwater use in theAfrican highlands. Soil and Tillage Research, 103: 247-256.
Suyker, A. E. and Verma, S., 2009. Evapo-transpiration of irrigated and rain-fed maize-
soybean cropping systems. Agricultural and Forest Meteorology 149: 443-452.
Tanner, C. B. and Sinclair, T. R., 1983. Efficient water use in crop production: Research or
re-search? In: H.M Taylor et al., (ed.) Limitations to efficient water use in crop
production. ASA, CSSA, and SSSA, Madison, WI.
The World Bank, 2005. Shaping the future of water for agriculture; A source book for
investment in agricultural water management. Rural and Agricultural Development,
Washington D.C. USA.
van Rensburg, L.D. and Zerizghy, M. G., 2008. The BEWAB+ computer program for
irrigation scheduling. Department of Soil, Crop and Climate Science, University of
the Free State, Bloemfontein, South Africa.
232
Walker W.R., 2003. SIRMOD Ill; Surface Irrigation Simulation, Evaluation and
Designs.Guide and technical document. Department of Biological and irrigation
Engineering, Utah State University
World Reference Base for Soil Resources (1998). World soil resources report, 84, ISSS /
ISRIC FAO, Rome, Italy.
Zhang, H., Oweis, T. Y., Garabet, S. and Pala M., 1998. Water use efficiency and
transpiration efficiency of wheat under rain-fed conditions and supplemental
irrigation in a Mediterranean-type environment. Plant and Soil, 201: 295-305.
233
CHAPTERIO
PERSPECTIVE ON RESEARCH
10.1 Introduction
Sustainable soil water storage is a major challenge of cultivated fields in the semi-arid areas.
A number of soil water management practices have been developed by local and international
studies to improve soil water regimes especially for small scale farmers. Rainfall capture and
supplemental irrigation are among the common water management options. Intensively
researched in the Free State Province of South Africa is the in-field rainwater harvesting
(IRWH) that partition the field into basin and runoff area. Although clay soils have been
earmarked for IRWH on the basis of their high water holding capacity, erratic rainfall
distribution and high evaporation prevailing in the region undermines water storage potential
of these soils, especially when dry spells persist longer than 14 days. In search for solutions
this thesis integrated micro-flood irrigation (MFI) system with IRWH. In this endeavour three
fields of study were researched including soil hydraulic properties of three soil types (Tukulu
and Sepane also known as Cutanic Luvisols, and Swartland as Cutanic Cambisols), short
furrow irrigation and integration of MFI with IRWH of which soil water balance (SWB)
management was of primary importance.
10.2 Soil hydraulic properties
Soil hydraulic properties are physical functions and parameters that describe the soil water
systems. These include soil water content (SWC), flux rates, and soil water characteristic
curve (SWCC or 8-h) as well as hydraulic conductivity (K-8). Soil morphological attributes
such as depth, texture and structure has profound influence on water flow patterns especially
on layered soil profiles. In this thesis morphological and hydraulic properties were
investigated concurrently using laboratory and in situ procedures. In Chapter 3 pedological
aspect of three representative layered soil forms were characterised in relation to internal
drainage and it was shown that horizons with clay content ranging from 26 to 48%
contributed less than 5 mm to total drainage. This observation was supported by the 8-h
relationship with a very small slope and K-8 relationship dropping by several orders of
magnitudes at near the saturated soil water content.
For optimal description of 8-h and K-8 relationships especially for structured soils,
appreciated for the many challenges in measuring hydraulic properties either in either in situ
234
or laboratory conditions, application of hydraulic models is widely practiced. In Chapter 4
three parametric models predictions were tested by fitting to laboratory measured SWCC and
in situ determined K-8 curves. Model predictions showed high accuracy for the former
compared to the latter. Predictions on the K-8 relationship were showed to improve more than
one order of magnitude when simulations were performed with HYDRUS-ID software based
on soil profile layers initial and boundary conditions and optimization of model parameters.
Model performance were also showed to vary with soil profile layers an observation that was
consistent with observations in Chapter 3 and later in Chapter 5 that soil layers affected water
flow and soil water storage differently. The effect of layered soils on evaporation is dealt with
in Chapter 5 and clayey horizons were shown to augment a steep soil water gradient against
drying front progression. This phenomenon restricted evaporation losses an observation that
corresponds to the findings about deep drainages losses in Chapter 3 especially in soil
profiles where the restrictive layer occupied the underlying position.
10.3 Micro-flood irrigation designs and evaluation
Conservational concepts and principles advocated by the developers of MFI in designing and
management of short furrow irrigation were tested in Chapters 6 and 7. Having been shown
in Chapters 3 to 5 to have higher soil water storage capacity the Tukulu soil form was
selected for this exercise. In Chapter 3 four inflow rates (20, 40, 80 and 160 L min-I) were
compared with respect to irrigation and opportunity time, advance stream distance-time and
infiltrated depth for a single-run irrigation in a 90 furrow length. Description of surface
characteristics of furrow irrigation was based on in situ observations only. From the smallest
to the largest inflow rate the advance and recession times ranged from 2.7 to 0.4 hours and 7
minutes to an hour respectively, with the 20 L min-I failing to reach the furrow end at a 60 m
distance.
Infiltration in furrows is basically two dimensional and this phenomenon was successfully
described using HYDRUS-2D software simulations. Remarkable agreement between
measured and predicted SWC was shown at all furrow distances and profile depths.
Infiltrated depths were observed to be a function of opportunity time, and this was confirmed
by the high distribution uniformity (DU) from all inflow rates at shorter furrow distances.
Deep percolation ratio (DPR) was showed to be insignificantly low for all inflow rates. This
observation corresponds with findings in Chapter 3 for deep drainage. Vertical redistribution
(Vz) of infiltrated SWC was dealt with in Chapter 7. The rate of Vz for small (20 and 40 L
235
min-I) and large (80 and 160 L min-I) inflow rates was showed to be high at soil profile
depths of 0 to 600 nun and 0 to 850 mm, respectively. Simulated SWC during the 455 hour
redistribution was also showed to be accurate on drier soils compared to near saturated soils.
10.4 Integrating MFI with IRWH
The merging of MFI with IRWH was based on their handling of runoff, though in different
forms, at field or plot scale with the aim of improving soil water storage and crop water use.
The merged MFI-IRWH water management system was tested in a maize crop grown for a
single season described in Chapter 8 and Chapter 9. Runoff strip width (RSW) and water
regime (WR) were the two main treatments with each having three sub treatments combined
on a 3 x 3 split plot factorial with four replications in a complete randomised design
experiment. One of the three sub-treatments combinations represented the traditional design
for IRWH of 2 m RSW and dryland (DL). The other two sub treatments were the 1 m and 2
m for RSW and full irrigation (FI) and supplemental irrigation (SPI) for WR. Experimental
plots were conceptualised into two components, the basin area (BA) and runoff area (RA).
In chapter 9 a SWB procedure was developed to take into account the gains and losses in soil
water storage effected in the BA and RA. Site specific empirical functions including rainfall-
runoff relationships and soil water content-time relationship were used to estimate total
harvested rainfall and deep drainage (DD) losses for the BA and RA. A fixed irrigation
schedule for the FI and SPI was followed produced by the BEW AB+ irrigation software for
semi-arid areas. For optimal rainfall capture at plot scale IRWH uses basin tillage and closed
ended furrows for MFI. Consequently, the SWB procedure disregarded runoff (ex-field)
losses and considered evapo-transpiration (ET) and DD as the major losses against the gains
from rainfall and irrigation. Though DD increased with increased RSW and water availability
it was shown in Chapter 10 to be significantly small over the production season. This result
corresponded to the observations made in Chapters 3, 6 and 7 for the Tukulu soil form.
Evapo-transpiration was partitioned into evaporation (Ev) and Transpiration (T). The BA was
distinguished as the area where crops grow by considering the area covered by plant canopy
to be proportional to the T contribution to ET by a ~-parameter ranging from 0 to 0.9, while
in the RA it ranged from 0 to 0.5 over the production season. Tests of the SWB procedure
showed consistency with the water gains and losses during the different maize growth stages
as affected by the RSW and WR main treatments. In evaluating the productive potential of
the merged MFI-IRWH system the SWB procedure showed optimum net gains for the 2 m
236
and 3 m RSW, especially from rainfalls not less than 24 mm. It was also showed that because
of the large surface area not shaded from direct radiation from these RSW most of the net
gain from rainfall were directed to Ev. This observation was illustrated by the Ev/T ratio that
was 30% larger for the 3 m RSW compared to the 1 m RSW. Dry spells more than 14 days
corresponded to net losses of about -12 mm, irrespective of RSW. Irrigation increased ET
with FI and SPI corresponding to higher seasonal ET compared to DL by 113 mm and 82
mm, respectively. Though the increase in ET with water availability was expected it was
shown that this increase was translated into production differently by the RSW. Total
biomass and grain yields for the 1 m RSW were shown to be higher compared to the 2 m and
3 m RSW by 21% and 45% and 35 and 89%, respectively. The DL had a total biomass of
7124 kg ha' I and grain yield of 3011 kg ha-I. Total biomass and grain yields for FI and SPI
were higher compared to DL by 200% and 76%. Correspondingly, the Irn RSW had the
highest WUE based on T for both grain (18.4 kg halmm') and total biomass (38.2 kg ha-I
mm") yield compared to the 2 m and 3 RSW. Among water regimes the highest WUE for
total biomass (48.2 kg halrnm') and grain yield (21 kg ha-Imm") corresponded to FI.
10.5 Thesis contributions and conclusions
Reflection on the procedures and findings from the series of studies including soil hydraulic
characteristics, MFI designs and evaluation and merging the MFI system with IRWH,
important contributions can be highlighted and conclusions made.
In Chapter 3 to Chapter 5 the functional role of clay soils in soil water storage for a draining
and evaporating profile was demonstrated and quantified using pertinent hydraulic
parameters such as SWC, flux rates, hydraulic conductivity and 8-h relationships. Matric
suction range of 0 to -10kPa was shown to be appropriate for describing matric suction
gradients on draining layered profiles compared to the 0 to -33 kPa proposed by Ratliff et al.
(1983). Inspection of drainage curves of other similar studies on layered profiles indicated
that drainage upper limit was approached at near -10 kPa even though they have initially
employed matric suction range of Ratliff et al. (1983). To this effect it can be concluded that
the 0 to -10 kPa matric suction range provided accurate description of matric suction
gradients during internal drainage and thus recommended for applications in matters
pertaining to estimation of drainage upper limit (DUL). In addition, re-defining DUL in terms
of drainage flux (0.001 mm hour") that amounts to DD losses proportional to 1% of annual
237
rainfall (Chapter 3) over a 7 month fallow period provided an alternative approach of
evaluating water storage potential of clay soils.
The MFI concept of selecting inflow rate according to soil's basic infiltration rate was tested
and evaluated in Chapters 6 and Chapter 7 using 20, 40, 80 and 160 L min-I inflow rates.
Though all inflow rates produced higher distribution uniformity (DU) at short compared to
long furrow distances, and the 20 and 40 L min-I treatments were much easier to handle
especially for a 30 m furrow run. In comparison, the 40 L min-I treatment had a higher DU
and smaller advance times. Therefore the 40 L min-I and 30 m furrow length were considered
appropriate for designing the merged MFI system with IRWH.
In Chapters 8 and Chapter 9 the merged MFI-IRWH water management system was
successfully tested on a maize crop for a single season. A soil water balance procedure was
developed that articulated the dynamic aspects of the soil-plant-atmosphere continuum for the
merged system and showed reasonable sensitivity to changes in water availability and RSW.
Application of the SWB procedure in evaluating the merged MFI and IRWH indicated higher
crop yields and water use efficiencies for the 1 m RSW and full irrigation. Therefore it can be
concluded that the 1 m RSW and full irrigation is appropriate for the newly merged MFI with
IRWH water management system. Though tested only for a single season this technique is
ready to be used by small scale farmers located on clay soils and who have access to
irrigation water to improve food security at household levels in Africa.
238
Appendices
239
Appendix A Profile description of the Tukulu soil form.
Map/photo: 2926 Bloemfontein Soil form and Family: Tukulu Dikeni
Latitude -Longitude: 29() 13' 24.69" S/26 012.40.93" E Surface rockiness: None
Altitude: 1421.9m Occurrence of flooding: None
Terrain unit: Midslope Wind erosion: Slight wind
Slope: 1% Water erosion: None
Slope shape: Straight Vegetationlland use: Agronomic field crops
Aspect: South Water Table: None
Micro relief: None Described by: M. Hensley/ S. Mavimbela
Parent material Solum: Origin binary,aeolin, solid rock Date Described: 1/11/2006
Underlying Material: Sandstone (Feldspathic) Weathering of underlying material: Moderate physical, moderate chemical
Alternation of underlying material: Ferrugised
Horizon Depth (mm) Description Diagnostic horizon
A 0-300 Moist state; dry colour: reddish brown (5YR4/8); moist colour; reddish brown (5YR4/8); texture: fine sandy loam; Orthic
structure: apedal massive: consistence: friable; few fine pores; common roots; gradual transition
BI 300-600 Moist state; dry colour: reddish brown (5YR3/6); reddish brown (5YR3/6);texture: tine sandy clay loam; Neoucutanic Neocutanic
Structure: apedal massive becoming weak subangular blocky towards transition: consistence: friable; few fine pores;few
clay cutans; very few fine pores; common roots; clear, tonguing transition
Cl 600-850 Moists state; dry colour. Dark grey, olive and yellow mottles; moist colour: olive grey with yellow mottles; texture: clay; Unconsolidated
common distinct grey material, and yellow reduced iron oxide motties; few prominent black oxidised iron oxide mottles; with signs of wetness
structure: prismatic; consistence firm; few slickensides; common clay cutans; very few roots; gradual transition
C2 850- ~ 1350 Allows water to enter, hence the presence roots. lt chops out the profile relatively easily up to £~1350 m, getting harder soft rock
towards the bottom and probably very impermeable slightly deeper. Unconsolidated:
I
240
Appendix B Profile description of the Sepane soil form.
Map/photo: 2926 Bloemfontein Soil form and Family: Sepane Rambles
I
Latitude +Longitude: 29{) 13'41.47" S/26{) 12'37.86 E Surface rockiness: None
Altitude: 1414.3 m Occurrence of flooding: None
Terrain unit: Midslope Wind erosion: Slight wind
Slope: 1% Water erosion: None
Slope shape: Straight Vegetation/land use: Agronomic field crops
Aspect: South Water Table: None
Micro relief: None Described by: M. Hensley/ S. Mavimbela
Parent material Solum: Origin binary,aeolin, solid rock Date Described: 1111/2006
Underlying Material: Sandstone (Feldspathic) Weathering of underlying material: Moderate physical. moderate chemical
Alternation of underlying material: Ferrugised
Horizon Depth (mm) Description Diagnostic horizon
I
A 0-300 Moist state: dry colour: reddish brown (5YR4/8); moist colour; reddish brown (5YR4/8); texture: fine sandy loam; Orthic
structure: apedal massive: consistence: friable: few fine pores; common roots: abrupt transition
BI 300-700 Moist state; dry colour: dark grey (5YR3/4); reddish brown (2.5YR3/4);texture: fine sandy clay loam: Pedocutanic Pedocunatic
Structure: moderate to medium sub-angular blocky: consistence: slightly firm: few normal fine pores; many clay cutans: clay
Skins; common roots; abrupt transition
Cl 700-800 Moists state; dry colour. Dark greyish brown (2.5YR4/2): moist colour: grey (2.5T5/0); texture: sandy clay: common distinct grey Prismacunatic,
and yellow reduced iron oxide rnottles: abundance of red. grey and black mottles; structure: prismatic: consistence:
slightly firm; few slickenside's; common clay cutans: very few roots; abrupt transition.
I
C2 800- ~ 1000 Undisturbed, signs of calcium deposits. It chops out the profile relatively easily up to £~1350 m, getting harder Soft rock
towards the bottom and probably very impermeable slightly deeper. Unconsolidated:
241
Appendix C Profile description of the Swartland soil form.
~"
tJ {f~
Map/photo: 2926 Bloemfontein Soil form and Family: Swartland /Rouxville -
Latitude -i-Longitude: 29() 13'23.98"S/26° 12'44.29" E Surface rockiness: None
<i?
"_J ,
Altitude: 1421.9m Occurrence of flooding: None I,
i.
; _::o.
Id:
Terrain unit: Midslope Wind erosion: Slight wind ·l1
Slope: 1% Water erosion: None .:J c
Slope shape: Straight Vegetation/land use: Agronomic field crops [
Aspect: South Water Table: None <==
Micro relief: None Described by: M. Hensley/ S. Mavimbela
Parent material Solurn: Origin binary,aeolin, local colluvium Date Described: 1/1112006
Underlying Material: Sandstone (Feldspathic) Weathering of underlying material: Moderate physical, moderate chemical
Alternation of underlying material: Ferrugised
Horizon Depth (mm) Description Diagnostic horizon
A 0-200 Moist state; dry colour: reddish brown (5YR5/8); moist colour; reddish brown (5YRl8); texture: tine sandy loam; Orthic
structure: apedal massive: consistence: friable; few tine pores; common roots; clear smooth transition
BI 200-400 Moist state; dry colour: dark reddish brown (5YR3/6); reddish brown (5YR3/6);texture: fine sandy clay; Pedocutanic Pedocunatic
Structure: moderate coarse sub-angular blocky: consistence: friable; very few fine sesquioxide concretions; few roots;
smooth transition
Cl 400 ~ 700 Moist state; undisturbed; dolorite rock; black concretions; many reddish-yellow geogenie mottles, very few tine sesquioxides concretions; Saprolite
Non hardened free lime; strong effervescence, very few roots; not observed transition.
242