摘要
基于E-FAST全局定量敏感性分析方法的思想,提出一个针对分布式物理水文模型——流溪河模型的参数敏感性分析方法,包括参数简化及取值范围确定、参数采样及采样次数的确定、敏感性评价指标的选取、参数敏感度计算4个部分。针对我国4个不同规模的流域(黄龙带水库流域、流溪河水库流域、长湖水库流域和新安江水库流域)开展了分布式物理水文模型参数敏感性分析,并采用4种不同的评价函数进行了对比计算。结果表明,流溪河模型饱和含水率和田间持水率是高度敏感参数,饱和水力传导率、土壤层厚度、土壤特性系数、边坡和河道糙率为敏感参数,其他参数为不敏感参数。
Liuxihe Model is a physically based distributed hydrological model proposed for river basin flood forecasting,which has 12 different model parameters in every cell.Due to the huge model parameters,it is difficult to optimize all the model parameters,and the parameter sensitivity analysis is needed for simplifying the parameter numbers.This paper,based on the E-FAST algorithm that has been widely used in quantitative sensitivity analysis,presented a parameter sensitivity analysis framework for Liuxihe model,which has 4 steps including parameter simplification and ranging,parameter sampling,sensitivity analysis evaluation function determination and sensitivity index calculation.Four river basins in Southern China including Huanglongdai River Basin,Liuxihe River Basin,Changhu River Basin and Xinanjiang River basin with basin areas ranging from several hundred to several thousand square kilometers were chosen as the study cases to validate this method,and 4 different evaluation functions were tested.The results showed that the water content at saturation condition and water content at field condition are highly sensitive model parameters,soil hydraulic conductivity,soil thickness,soil porosity characteristics,hill slope Manning’s coefficient and river channel Manning’s coefficient are sensitive parameters,while the other parameters including potential evaporation,evaporation coefficient,underground water recession coefficient and water content at wilting condition are less sensitive parameters.
出处
《热带地理》
北大核心
2012年第6期606-612,632,共8页
Tropical Geography
基金
广东省科技计划项目(2011A030200013)
国家科技支撑计划项目(2012BAK10B06)
广东水利科技创新项目(2009-16)
