Evaluation of Extended Kalman Filter-based Neural Network (EKFNN) model and Gene Expression Planning in Rainfall-Runoff Modelin

Simulation of the rainfall-runoff process is the most important step in water engineering and water resource management studies. Exploitation of surface water and underground water resources, river management and flood warning requires prediction of river and runoff discharges of the watershed. In this study, Extended Kalman Filter-based Neural Network (EKFNN) method was used for rainfall-runoff modelling. Then, the results were compared with the Gene Expression Planning method, which showed good performance in rainfall-runoff modelling in most recent studies. The data used in this study is related to daily runoff and rainfall of the rain gauge and hydrometric stations of Malayer plain which includes Peyhan, Marvil and Namyleh stations, during the period of 2001 to 2013. The results indicated that the EKFNN model was superior to GEP model in daily river flow modelling in Malayer plain. In addition, the speed of implementation of the Gene Expression Planning model was greater and was able to present results in a short time. Finally, EKFNN model was selected as the superior model for Malayer plain.