Rainfall studies for the highlands of Eritrea

Abstract

English: Long-term rainfall records of the highlands of Eritrea were examined to reveal the short and long-term patterns. Monthly and annual spectra have been analysed to provide a preliminary assessment in terms of homogeneity, spell length, water balance, drought and spatial variability. Three methods of spatial interpolation were evaluated for mapping continuous spatial rainfall values for mean annual rainfall totals. Interpolation methods examined in this study include inverse distance weighted (!DW), Spline and Kriging. A statistical assessment of the continuous surface created indicates that there is little difference between the estimating ability of the three spatial interpolation methods, but with Kriging performing better overall than Spline or !DW. Monthly means provide little information on many properties of the rainfall that are relevant to the wide variety of rainfall-related activities and decision making. A stochastic rainfall generator model based on a first-order Markov chain was developed for generation of artificial daily rainfall data from monthly totals. Intensive statistical testing of the performance of the model reveals that the model can be used with confidence to generate a pseudo daily rainfall dataset for the peak rainy months (July-August), for the highlands of Eritrea. A deterministic model was developed to investigate how the global rainfall predictors correlate to the two peak rainy months (July-August), which contribute 65% of the total rainfall in the highlands of Eritrea. The main aim of looking at these relationships is to develop a simple statistical model for forecasting rainfall amount. In a preliminary step, in order to identify the most influential rainfall predictor a correlation matrix and stepwise regression of 11 predictors with different lags were analysed. The influence of the southern Indian Ocean Sea Surface Temperatures was identified as the most influential predictor for the highland of Eritrea. The model was validated using the jack-knife skill test, Chi-squared and deviation based test giving a promising result.

Afrikaans: Die langtermyn reënvalverslag van Eritrea se Hooglande is ondersoek om beide die kort- en langtermyn patrone aan te dui. Maandelikse en jaarlikse spektra is geanaliseer om 'n voorlopige beraming in terme van homogeniteit, durasie lengte, waterbalans, droogte en ruimtelike veranderlikheid te voorsien. Drie metodes van ruimtelike interpolasie is geëvalueer om kontinue ruimtelike reënvalwaardes vir gemiddelde jaarlikse reënval totale te karteer. Interpolasie metodes ondersoek tydens hierdie studie sluit in "Inverse distance weighted" (IDW), Spline en Kriging. 'n Statistiese analise van die kontinue oppervlak wat geskep is, dui min verskille tussen die ramingsvermoë van genoemde drie ruimtelike interpolasie metodes, maar Kriging het oor die algemeen beter gevaar as Spline of IDW. Maandelikse gemiddeldes voorsien min inlgiting oor talle kenmerke van reënval relevant tot die wye verskeidenheid reënval-verwante aktiwiteite en besluitneming. 'n Stogastiese weergenererende (gebaseer op 'n eerste-orde Markov ketting) model is ontwikkel vir generasie van kunsmatige daaglikse reënvaldata vanaf maandelikse totale. Dit blyk uit intensiewe statistiese toetsing van die werkverrigting van die model dat genoemde model met vertroue gebruik kan word om 'n pseudo daaglikse reënval datastel vir die Hooglande van Eritrea te genereer. 'n Deterministiese model is ontwikkel om ondersoek in te steloor hoe die globale reënval voorspellers verwant is aan die twee spits reënmaande (Julie-Augustus). Laasgenoemde twee maande dra tot 65 % van die totale reënval in die Hooglande van Eritrea by. Die hoofdoel met die ondersoek van hierdie verhoudings is om 'n eenvoudige statistiese model vir voorspelling van reënval hoeveelheid te ontwikkel. In 'n poging om die sterkste reënvalvoorspeller te identifiseer, is as voorlopige stap, 'n korrelasie matriks en stapsgewyse regressie van 11 voorspellers met verskillende sloertye geanaliseer. Die invloed van die suidelike Indiese Oseaan See Oppervlak Temperature is geidentifiseer as die sterkste voorspeller vir die Hooglande van Eritrea. 'n Model is gevalideer deur gebruik te maak van die "Jack-knive" vaardigheidstoets, Chi-kwadraat en afwykingsgebaseerde toetse. Die resultate was belowend.