Simulating rainfall-runoff process with a new combined artificial intelligence

Today, the rainfall-runoff process is one of the most important and complex hydrological phenomena in the management of surface water resources to take appropriate measures in the event of floods and droughts. In order to simulate this process, a proper understanding of the behavior of the basin can play an effective role in model selection and time saving related to simulation. Therefore, to simulate the runoff process of Karkheh catchment located in Iran, statistical models and artificial intelligence approaches including Multivariate Linear Regression (MLR), Artificial Neural Network (ANN), Support Vector Regression (SVR) and Support Vector Regression-Wavelet (WSVR) were used on a daily time scale during the statistical period of 2010-2020. To evaluate the simulation performance, statistical indices including coefficient of determination (R2), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Nash-Sutcliffe Efficiency coefficient (NSE), and bias percentage (PBIAS) were employed. The results illustrated that the studied models had better performance in composite structures. Moreover, artificial intelligence models have less error and better performance than statistical models. The results demonstrated that the Wavelet support vector regression model had greater accuracy and less error than other models. Overall, according to the outcomes, the use of hybrid artificial intelligence models is effective in the runoff process and can be considered as a suitable and rapid solution in water resources management.

Simulating rainfall-runoff process with a new combined artificial intelligence Keywords:

simulation , Artificial intelligence , Karkheh , Water resources management