Uncertainty Analysis due to the Application of Different Infiltration Methods on the Performance of HEC-HMS model Using GLUE Algorithm

Quantifying the uncertainty contribution of important factors on the performance of rainfall-runoff models has always been one of the major challenges for researchers and hydrologists. The main problems of applying these models especially in calibration period are the large number of required parameters and the lack of physical understanding for some of them. This research addressed the uncertainty contribution of different infiltration methods (Green-Ampt, SCS-CN, Exponential, Smith-Parlange, Initial-Constant and Deficit-Constant) on the performance of HEC-HMS model using GLUE algorithm. Results showed that using each of infiltration methods imposes different uncertainty bounds on the simulated flood hydrograph by HEC-HMS. Findings indicate that SCS-CN and Smith-Parlange owing to have the higher P-factor (0.78 and 0.72) and lower ARIL (0.39 and 0.40) values, enforce minimum uncertainty on the model output. In addition, the mentioned infiltration methods have the fewer sensitive parameters and then performs better than other methods. In contrast, the uncertainty of applying Initial-Constant and Deficit-Constant methods for simulation of flood hydrograph is relatively high the smaller percentage of the observed data are within the 95% uncertainty bandwidth. Moreover, sensitivity analysis of the parameters of each of the infiltration methods using the nonparametric Kolmogorov–Smirnov (D) test showed that parameters with the sharp and peaked distributions indicate well-identifiable parameters, while flat and spread distributions indicate uncertain parameters. Overall, the outcomes of this study prove that GLUE algorithm has high ability to determine the optimal range of rainfall-runoff model parameters and the prediction uncertainty bandwidth.