SWMM模型在城市不透水区地表径流模拟中的参数识别与验证

为了研究城市不透水下垫面的降雨径流过程和污染负荷,以屋面为例,选择径流管理模型SWMM,采用独立场次实测数据,应用基于不确定性分析的HSY算法和Monte Carlo采样方法对模型中的水文水力和水质参数进行识别和验证.结果表明,地表不透水区径流模型中主要包含6个关键参数,分别为不透水区初损填洼深度(S-imperv)、不透水区曼宁系数(N-imperv),指数累积方程中的最大可能累积值(max buildup)、累积常数(rate constant),指数冲刷方程中的冲刷系数(coefficient)和冲刷指数 (exponent).水文水力参数的识别可以最小二乘法偏差作为目标函数,水质参数的识别可以场次污染负荷和污染物峰值浓度作为目标函数.参数识别结果为N-imperv0.012～0.025, S-imperv 0～0.7, max buildup 15～30, rate constant0.2～0.8, coefficient0.01～ 0.05, exponent1.0～1.2.参数的区域灵敏度由大到小排序为 coefficient、S-imperv、N-imperv、max buildup、exponent、rate constant.识别后的参数可以通过模型验证,但是在模拟一些雨型特殊的降雨径流污染物浓度曲线时,仍然存在一定的困难.

Parameter Identification and Validation of SWMM in Simulation of Impervious Urban Land Surface Runoff

The purpose of this paper is the application of storm water management model (SWMM) in simulating runoff hydrology and water quality. The study chose a roof as the typical impervious urban land surface, and monitored several rainfall-runoff events for parameter identification. We identified and validated hydrological and water quality parameters, using Monte Carlo sampling method and HSY algorithm, which are based on uncertainty analysis. Results show that impervious urban land surface runoff model includes 6 critical parameters, which are depression storage (S-imperv), Manning's n (N-imperv), maximum buildup possible (max buildup), buildup rate constant (rate constant), washoff coefficient (coefficient), and washoff exponent (exponent). Identification of S-imperv and N-imperv could use least square error as objectives, while others could use errors of event pollution load and peak concentration of pollutant as objectives. The identification results of the 6 parameters are N-imperv 0.012-0.025,S-imperv 0-0.7, max buildup 15-30,rate constant 0.2-0.8,coefficient 0.01-0.05, and exponent 1.0-1.2. Regional sensitivities of these parameters in non-ascending order are coefficient, S-imperv, N-imperv, max buildup, exponent, and rate constant. Identified parameters are able to be validated by SWMM model. However, current model structures still have some difficulties in simulating runoff pollutant concentration curves caused by some special rain patterns.