Abstract
Evapotranspiration is an important component in many hydrological, ecological and agricultural studies. There are many available direct and indirect methods to determine the evapotranspiration rate. In this study, alternative model based on multilayer Artificial Neural Network (ANN) using the backpropagation algorithm was proposed to estimate evapotranspiration as referred to pan evaporation. The meteorological data used in this study were obtained from Al-Zahra and Al-Zawia stations which located on the coastal area of western Libya lie. The input data were consisted of mean temperature, mean relative humidity and mean of actual sunshine hours of consecutive years (1995, 1996, 1997 and 1999). The performance of the ANN model was evaluated against a set of data that never seen by the model during the training phase. The evaluation of ANN model was also performed against Blaney and Criddle, Radiation and modified Penman methods. The results showed that ANN forecasts were superior to the ones obtained by Blaney and Criddle and Radiation methods. Due to its little input data, ANN is considered to be more efficient as compared with the modified Penman method. However, this application of ANN as a fitting tool should be useful in evapotranspiration modeling. Keywords: Evapotranspiration Pan evaporation, Artificial neural networks, Backpropagation algorithm.