Evaluation the Performance of Different Models of Artificial Neural Network in Estimating Evaporation Losses from Pan around the Shahid Rajaei Dam Lake

Evaporation is one of the main components of the water cycle in nature, which plays a key role in agricultural studies, hydrology, meteorology, reservoir operation, irrigation and drainage systems design, irrigation scheduling and water resources management. In this study, eight types of meteorological parameters as inputs for estimating evaporation from the pan by artificial neural network for four meteorological stations around Shahid Rajaei Dam were investigated. Meteorological data were collected for ten years from 4 stations around Shahid Rajaei Dam. The results of statistical criteria of the models, distribution diagram and daily evaporation rate were estimated and observations showed that the neural network method was able to estimate the daily evaporation in the four stations with good accuracy. However, the best structure of neural network models for stations of Soleiman Tangeh, Sari Office, Frime Sahra and Telamadreh with seven input variables, one hidden layer and 12, 8, 10 and 12 neurons, respectively, were selected according to MSE and R2 criteria. MSE and R2 criteria were selected. The correlation coefficients for daily data in Soleiman tangeh, Sari, Sahra and Telmadreh stations were extracted 0.88, 0.91, 0.92 and 0.89, respectively. Also, the results of monthly evaporation simulation showed that the artificial neural network method was able to calculate the monthly evaporation with correlation coefficients of 0.98, 0.98, 0.99 and 0.99 with 95% confidence level for Soleiman Tangeh, Sari office, Frime Sahra and Telamadreh stations, respectively.