城市内涝模型参数灵敏度分析与方法比较

Sensitivity Analysis of Urban Flooding Model Parameters Using Local and Global Approaches

频发的内涝灾害给城市运行和管理造成了严重影响,尤其是道路交通。针对道路上的典型内涝点建立城市内涝模型,以积水过程为目标函数计算暴雨情景下各参数的灵敏度,并比较局部灵敏度分析与全局灵敏度分析两类方法对参数灵敏度计算结果和排序的影响。结果表明,管道曼宁系数是最灵敏参数,其灵敏度远大于其他参数;在局部灵敏度分析中,步长越小则参数灵敏度的可区分性越大,且灵敏度随参数初始值变化而剧烈波动;全局灵敏度分析得到的参数灵敏度取值较稳定,且参数灵敏度排序也较为稳定。

Urban flooding events frequently occurring have serious negative impact on city operation and management, especially on road traffic. In this study, an urban flooding model in a typical waterlogged road area was established to evaluate the sensitivity of its parameters by using waterlogging depth as the objective function under rainstorm situation, and effects of local sensitivity analysis method ＆ global sensitivity analysis method on the results and ranks were compared. The results showed that pipeline Manning coefficient was the most sensitive parameter which was much more sensitive than the other parameters ; in the local sensitivity analysis method, the smaller the step size, the more distinguishable the parameter sensitivity, and the sensitivity fluctuated dramatically with variation of initial parameter values; parameter sensitivity values and their ranks obtained from the global sensitivity analysis method were more stable.