Rainfall- Runoff Modeling Using HBV Model and Random Forest Algorithm in Bazoft Watershed

Estimation of runoff in a catchment area is important from various aspects such as dam reservoir management, water resources management, flood regulation, and erosion control in river banks and bed. In the present study, a conceptual model of HBV and an intelligent model of Random Forest (RF) were used to simulate the rainfall- runoff process in Bazoft watershed at the Landi hydrometric station during the period of 2010 to 2017. In order to evaluate the performance of models, the statistical criteria, including Correlation coefficient (r), Root Mean Squares Error (RMSE), Nash-Sutcliffe efficiency coefficient (NS), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE) were used. Comparing the results of HBV and RF models revealed that the RF model outperformed the HBV. Thus, the RF model with r=0.95, NS=0.82, MAPE=9.59, MAE=0.25, and RMSE=0.39 m3/s was selected as the top model which might be used as a new choice to predict runoff in Bazoft watershed.