Application of artificial intelligence methods to estimate monthly pan evaporation using meteorological data

In this study, it is aimed to estimate monthly open surface evaporation. The monthly meteorological data, total precipitation, average temperature, minimum temperature, maximum temperature, average wind speed, average relative humidity, average vapor pressure, average atmospheric pressure, of three stations (Konya, Karaman, Aksaray) operated by Turkish State Meteorological Service on Konya Closed Basin in Turkey are used as inputs to the artificial intelligence techniques to estimate monthly evaporation. Artificial intelligence methods consisting of MLP-SCG, MLP-GDX, RBNN, GRNN, ANFIS, ε-DVR models are used for estimating the monthly evaporation of the stations. The results were compared with the results of FAO-Penman-Monteith, Priestley-Taylor, Meyer and Romanenko empirical equations. Nash-Sutcliffe efficiency coefficient (NSE), mean absolute error (MAE) and root mean square error (RMSE) were used for evaluating the applicability of developed models. According to the results of three stations, the most successful artificial intelligence methods were obtained in ε-SVR method at Karaman and Aksaray stations and MLP-SCG models in Konya station. For estimation of evaporation, the worst performance among artificial intelligence methods used was obtained at GRNN at Karaman and Konya stations and ANFIS at Aksaray station. The results of the empirical equations used show that the FAO-Penman-Monteith is more successful than the other empirical equations in all three stations. Comparison of the applied models reveals that the artificial intelligence methods perform better than the empirical equations for evaporation estimation. Keywords: Evaporation, Climate Change, Konya Closed Basin, Artificial Intelligence Methods.