Application of a Sampling Based on the Combined Objectives of Parameter Identification and Uncertainty Analysis of an Urban Rainfall-Runoff Model

作者：Zhao Dongquan^*; Chen Jining; Wang Haozheng; Tong Qingyuan

来源：Journal of Irrigation and Drainage Engineering, 2013, 139(1): 66-74.

DOI：10.1061/(ASCE)IR.1943-4774.0000522

• Combined objective
• Parameter identification
• Rainfall-runoff model
• Latin hypercube sampling (LHS)
• Storm water management model (SWMM)

Urban rainfall-runoff modeling is important to analyze the health of urban water systems. The Storm Water Management Model (SWMM) developed by the U. S. Environmental Protection Agency (USEPA) is one of the most widely used rainfall-runoff models. In this paper, SWMM was integrated with a sampling based on a combined objective, aiming at (1) reducing the uncertainty of model parameters distribution in the sampling-based method, and (2) reducing uncertainty bounds of model outputs to obtain more accurate predictions. The Latin hypercube sampling (LHS) method was selected as the sampling strategy, and the Nash-Sutcliffe (NS) coefficient, correlation coefficient (R), percent bias (%BIAS), root mean square error (RMSE), and a combination with the previous four objectives were used to get behavioral parameter sets, respectively. A case study was made in an experimental catchment of Macau, China, to test the efficiency of different objectives on reducing uncertainty of the rainfall-runoff model. Results indicated that the combined objective method could reduce the uncertainties of both the model parameters and the predictions. DOI: 10.1061/(ASCE)IR.1943-4774.0000522.

• 出版日期2013-1
• 单位清华大学