Comparison of gene expression programming (GEP) and neuro-fuzzy methods for estimation of pan evaporation (case study: south Khorasan province)

Accurate estimation as one of the important elements of the hydrological cycle of evaporation play an important role in the development and management of water resources plays countries facing a water crisis. so far methods, and many empirical formulas in estimating the nonlinear process evaporation of the basin, which provide high accuracy and as well as access to all the input parameters problem or measure they need a lot of time and money.The aim of this study was to compare of ability of Gene expression programming and neuro-fuzzy methods for estimation of evaporation in South Khorasan province. For this purpose daily data collected from occurs six synoptic stations during years 1990-2010. Input parameters are daily mean temperature, relative humidity, max and min temperature, wind speed and sun shine. Finally, to evaluate of models and compare them criteria such as coefficient of determination (R2) and Mean Bias Error (MBE) and Root Mean Square Error (RMSE) and MBE is used. Compression of result in test period showed the GEP Model has better performance than the neuro-fuzzy model to estimate the daily evaporation. The best result of GEP is R2=0.79¡ RMSE= 1.44 and MBE=0.35 in Boshroie station and worst is R2=0.7¡ RMSE= 2.6 and MBE=1.2 in Birgand station. Also the results showed the main factor in estimation of evaporation is mean temperature in all station except that mine temperature is impact parameter.