Estimation of missing annual discharge

Flood is one of the well-known facts which endanger the lives and human resources around the world. Thus, accurate estimation of flood discharge in every region can lead to more precise hydraulic structures with adequate capacity to avoid the above problems. Usually, the estimation of flood capacity in any station required sufficient data. However, the lake of sufficient and long-term hydrological data in many situations is a major threat to the start new projects. Thus, it is necessary to develop new methods for different circumstances and situations to estimate the required data for the target station. In this study artificial neural network has been applied to the reconstruction of annual discharge of hydrometric stations in Urmia Lake Basin and the results have been compared with those of normal ratio method to introduce the best technique for this study. It was shown that neural network provides the best approximation based on the root mean square of the estimated errors (RMSE), the percent of volume error (VE), and the correlation coefficient (R²).