俞双恩, 王宁, 于智恒, 王君. DRAINMOD-S模型参数的全局灵敏度分析[J]. 水科学进展, 2015, 26(2): 271-278. DOI: 10.14042/j.cnki.32.1309.2015.02.018
引用本文: 俞双恩, 王宁, 于智恒, 王君. DRAINMOD-S模型参数的全局灵敏度分析[J]. 水科学进展, 2015, 26(2): 271-278. DOI: 10.14042/j.cnki.32.1309.2015.02.018
YU Shuang'en, WANG Ning, YU Zhiheng, WANG Jun. Global sensitivity analysis of parameters in DRAINMOD-S[J]. Advances in Water Science, 2015, 26(2): 271-278. DOI: 10.14042/j.cnki.32.1309.2015.02.018
Citation: YU Shuang'en, WANG Ning, YU Zhiheng, WANG Jun. Global sensitivity analysis of parameters in DRAINMOD-S[J]. Advances in Water Science, 2015, 26(2): 271-278. DOI: 10.14042/j.cnki.32.1309.2015.02.018

DRAINMOD-S模型参数的全局灵敏度分析

Global sensitivity analysis of parameters in DRAINMOD-S

  • 摘要: 为有效进行DRAINMOD-S模型参数的优选,更好地理解参数变化对模拟结果的影响,开展了模型参数灵敏度分析。以南通市九龙垦区暗管排水脱盐试验为例,采用Morris全局定性分析方法检测了DRAINMOD-S模型模拟土壤剖面含盐量时侧向饱和导水率Ksat、水动力弥散系数D、地表最大蓄水深度Sm、相对不透水层深度Im、排水系数Dr及地下水初始埋深W 6个参数的灵敏度。结果表明:Ksat对模拟结果影响最为显著,DSmDr次之,而WIm影响最小;各个参数间的非线性作用存在差异,以Ksat最为显著。为保证模型模拟质量,对敏感性参数应提高现场测试精度,在模型运行时,对灵敏度大的参数应进行重点调整,同时也不可忽视非线性作用较强的参数,从而有效地指导模型的参数率定,提高模型的适用性。

     

    Abstract: In order to efficiently select the optimized parameters in DRAINMOD-S and figure out how the parametric variation influences the simulation results, sensitivity analysis of the parameters in the model was performed. Taking the pipe drainage desalting test as an example, which was conducted in Nantong Kowloon Reclamation Area, the Morris global qualitative sensitivity analysis was adopted to detect the the parameters sensitivity simulated in DRAINMOD-S of soil salinity in the profile against six parameters, namely, lateral saturated hydraulic conductivity Ksat, hydrodynamic dispersion coefficient D, maximum depth of surface water Sm, actual distance from surface to impermeable layer Im, drainage coefficient Dr and initial groundwater depth W. The results showed that Ksat has remarkable influence on simulation followed by D, Sm, Dr, while W and Im have little influence; nonlinear interactions among the parameters are different, and Ksat is the most significant. To guarantee the quality of the model simulation, accuracies of on-site measurement should be improved for the sensitive parameters and more emphases should be put on the sensitive parameters during the modelling. Meanwhile, the parameters which have strong nonlinear interaction should not be ignored. Therefore, model parameter calibration can be guided efficiently, and the applicability of the model will be improved.

     

/

返回文章
返回