The Comparison of Two Empirical Runoff Yield Models and Three Physical Models

Abstract

We used the particle swarm optimization algorithm to make comparisons between five two-parameter runoff models, two empirical models, LCM and SCS; and three physical models, GA, GAML and GAF, in calculating rainfall-runoff yield. The comparison results show that the qualified rate and coefficient of determination of the LCM, SCS, and GAF models were higher than 83% and 0.89, respectively. GA and GAML models can simulate the mean value of data; however, they performed poorly in describing the relationship between rainfall and runoff yield. The qualified rate and coefficient of determination of the GA and GAML models were lower than 50% and 0.50, respectively. In conclusion, LCM and SCS models can be used to calculate the runoff yield of small ungauged watersheds. Small watersheds with gauged data can use GAF model, which has high accuracy and physically meaningful parameters, to calculate runoff yield.