A system for drought monitoring and severity assessment
Lourens, Uys Wilhelm
MetadataShow full item record
English: The objectives of this study were: (i) to develop a near real-time crop-specific drought monitoring system that delimits drought stricken areas and assesses the severity of droughts in these areas, (ii) to produce products from the system which can be used for decision support by decision makers, and, (ii) to test the system for maize production using historical production seasons. Objectives (i) and (ii) An agricultural drought monitoring system was designed, which combined crop growth modelling and a Geographic Information System (GIS) . The use of crop models made it possible to assess the drought damage suffered by crops, in relation to their growth stage. As drought is a spatially related phenomenon, a GIS was used to present the geographic distribution of a drought situation. A grid based, spatially distributed, system was designed. The map units of the South African 1.:250 000 map series were used as the base units on which to present information. Each base unit was divided into cells covering an area of 2' by 2' minutes of latitude and longitude. There were thus 1.800 grid cells in one such unit. The models were run for each of these cells. The data inputs required by the crop models therefore had to be spatially distributed. Methods of creating spatially distributed weather data bases, were implemented or developed. Existing interpolation techniques were used to create the rainfall and temperature data bases. A technique developed for determining daily irradiance, from the Japanese Geostationary Meteorological Satellite, was adapted for use on METEOSAT data obtained over South Africa. A spatially distributed soil data base was also created. Maize was chosen as the crop to monitor in the initial evaluation of the system. Drought monitoring was undertaken at fortnightly intervals from the beginning of the crop production season. At each interval, observed weather data was used up to the present date, and the season completed with surrogate data. Three surrogate scenarios were used: a below normal rainfall year, a normal rainfall year, and, an above normal rainfall year. Surrogate data were created for each homogeneous climate zone (HCZ) within the study area. The HCZ within which the cell lay was determined and its data used to complete the season. A rainfall data generator, the accuracy of which had been proved, was used in establishing the surrogate data. The cumulative probability distribution function (CDF) of seasonal yield, was used as the norm against which to measure current season performance at the conclusion of each monitoring session. CDF's were established for all combinations of soil, climate, and planting dates used within the bounds of a particular 1:250 000 map unit. The yield simulated for each cell was compared with the appropriate CDF, and the probability range within which it lay, determined. A drought index value was assigned based on this comparison. The indices were: 1- Extreme Drought (CDF probability range 0- 10%), 2- Severe Drought (>10- 20%), 3- Moderate Drought (>20- 30%), 4- Mild Drought (>30- 40%), and, 5 - No Drought (>40 - 100%). Maps showing the distribution, and tables providing the extent of area classified, were produced. Objective (iii) The drought monitoring system was tested for three maize production seasons. The accuracy of the system was determined by comparing the average maize yield per magisterial district with measured yield data. Individual farm records were also evaluated. The system accurately portrayed the general maize production trends during a severe drought (91/92}, while an r 2 of 0.59 was obtained for the individual yields. The crop modelling approach to drought assessment takes the interaction of the soil, plant and atmosphere into account and is crop specific. The important influence of both the amount and timing of rainfall in relation to crop growth stages is therefore reflected in the drought index.