Estimating Actual Evapotranspiration in a Catchment Using Artificial Neural Networks with Minimum Climatic Data Case Study: Emame Representative Catchment

Actual evapotranspiration (ETa) is one of the major components in the hydrologic cycle and its accurate estimation is of paramount importance for many studies concerning hydrologic water balance and water resources planning and management. Evapotranspiration is a complex nonlinear phenomenon depending on several interacting climatological and crop factors. This paper determines the minimum climatic data required for estimating ETa in a representative catchment (Emame, Iran) using artificial neural networks (ANNs) technique. Four combinations of weather parameters were considered as input data and the resulting values of ETa were analysed and compared with those of water balance method. The study indicated that maximum and minimum air temperature, relative humidity, and wind speed are the minimum climatic data required for estimating ETa. With these input data, the root mean square error (RMSE) and the coefficient of determination (R2) for the comparison between observed and estimated ETa are 0.17 mm d-1 and 0.95, respectively,. Plotting measured values of ETa versus predicted values suggests that 82 percent of the values lie within a scatter of ±15%.