Comparing performance of different loss methods in rainfall-runoff modeling

Abstract

With respect to the effect of precipitation loss on runoff generation, different loss methods of Soil and Conservation Service (SCS), Green and Ampt (G.A.), Initial-Constant (I.C.), Deficit-Constant (D.C.), Constant Fraction (C.F.), exponential (Exp.) and Soil Moisture Accounting (SMA) have been compared by HEC-HMS event based on Rainfall-Runoff modeling in Roud Zard basin. The SMA method with max average Nush-Sutcliffe (N.S.) and min Peak Weighted Root Mean Square Error (PWRMSE) in calibration and verification was the best method in stream flow simulation. The SCS and Exp. methods with similar N.S. and PWRMSE were placed as second suitable methods in sub-daily (2 h) event simulation, and the G.A., C.F. and I.C. methods were the lasts. The comparison between simulated and observed key variables showed that SMA with Mean Absolute Error (MAE) of 0.05 in volume and 3.34 in peak flow simulation was the best in both calibration and verification, and in time to peak, in verification events. In volume simulation I.C. was the second and SCS and G.A. were the worst but in peak flow, SCS was second and others were similar. It could be concluded that SMA as a continuous infiltration method is preferred to the other methods for event based Rainfall-Runoff modeling.