Prediction of Monthly Reference Evapotranspiration in Homs Using Artificial Neural Network and Fuzzy Inference System

Accurate estimating and predicting of hydrological phenomena plays an influential role in the development and management of water resources, preparing of future plans according to different scenarios of climate changes. Evapotranspiration is one of the major meteorological components of the hydrologic cycle and from the most complex of them, and the accurate prediction of this parameter is very important for many water resources applications. So, this research goals to prediction of monthly reference evapotranspiration (ET0) at Homs meteostation, in the middle of Syrian Arab Republic, using Artificial Neural Networks (ANNs), and Fuzzy Inference System (FIS), depending on available climatic data, and comparision between the results of these models. The used data contained 347 monthly values of Air Temperature (T), Relative Humidity (RH), Wind Speed (WS) and Sunshine Hours (SS) (from October 1974 to December 2004). The monthly reference evapotranspiration data were estimated by the Penman Monteith method, which is the proposed method by Food and Agriculture Organization of the United Nations (FAO) as the standard method for the estimation of ET0, and used as outputs of the models. The results of this study showed that feed forward back propagation Artificial Neural Networks (FFBP-ANNs) pridected successfully the monthly ET0 using climatic data, with low values of root mean square errors (RMSE), and high values of correlation coefficients (R), and showed that the using of the monthly index as an additional input, improves the accurate of prediction of the artificial neural networks models. Also, the results showed good ability of Fuzzy Inference Models (FIS) in predicting of monthly reference evapotranspiration. Sunshine hours are the most influential single parameter for ET0 prediction (R= 97.71%, RMSE = 18.08 mm/month) during the test period, sunshine hours and wind speed are the most influential optimal combination of two parameters (R= 98.55%, RMSE = 12.49 mm/month) during the test period. The results showed high reliability for each of the artificial neural networks and fuzzy inference system with a little preference for artificial neural networks which can add the monthly index in the input layer, and there for improve the presicion of predictions. This study recommends the using of artificial intelligence techniques in modeling of complex and nonlinear phenomena which related of water resources.