Estimation of daily potential evapotranspiration of meadow using artificial neural networks

The ability of artificial neural networks (ANN) method was studied in this research in estimation of daily potential evapotranspiration(ETp) with grass reference crop and its comparison with Penman-Monteith method. The five years daily climatic data of Ekbatan station, Hamadan, Iran, including maximum and minimum temperature, maximum and minimum relative humidity, wind speed and sunny hours were input data and ETp was set as output data. The best ANN architecture was selected on the basis of minimizing the root mean square error (RMSE). The architecture of one hidden layer with one processing element gave the least RMSE of 0.7 mm/day and determination coefficient of 0.84. Whereas, the RMSE and determination coefficient of Penman-Mantieth were found 1.2 mm/day and 0.84 respectively. Based on these results, it could be concluded that the ANN predicted ETp better than Penman-Mantieth method. The sensitivity analysis also showed that the minimum temperature and maximum relative humidity had the most and the least effect on input respectively.