Evaluation of Three Evaporation Estimation Techniques In A Semi-Arid Region (Omar El Mukhtar Reservoir Sluge, Libya- As a case Study)

In many semi-arid countries in the world like Libya, drinking water supply is dependent on reservoirs water storage. Since the evaporation rate is very high in semi-arid countries, estimates and forecasts of reservoir evaporation rate can be useful in the management of major water source. Many researchers have been investigating the suitability of estimates evaporation rates methods in many climatic settings, infrequently of which were in an arid setting. This paper presents the modeling results of evaporation from Omar El Mukhtar Reservoir, Libya. Three techniques namely (artificial neural networks (ANN), Multiple linear regression (MLR) and response surface methods (RSM) ) were developed, to assess the estimation of monthly evaporation records from 2001 to 2009; their relative performance were compared using the coefficient of determination(E), mean absolute percentage error (MAPE%), and 95% confidence interval. The key variables used to develop and validate the models were: monthly (precipitation Rf., average temperature Temp., relative humidity Rh., sunshine hours Sh., atmospheric pressure Pa. and wind speed Ws.). The encouraging results approved that the models with more inputs generally had better accuracies and the ANN model performed superior to the other models in predicting monthly Evp with high E=0.86 and low MAPE%= 13.9 and the predicted mean within the range of observed 95CI%. In summary, it is revealed in this study that the ANN and RSM models are appropriate for predicting Evp using climatic inputs in semi-arid climate