QQ客服 官方微博 反馈留言

	An efficient global sensitivity analysis approach for distributed hydrological model
	Xia J.
	2012
关键词	response surface methodology sensitivity analysis support vector machines RSMSobol method Huaihe River Basin environmental-models uncertainty methodology parameters systems design
英文摘要	Sensitivity analysis of hydrological model is the key for model uncertainty quantification. However, how to effectively validate model and identify the dominant parameters for distributed hydrological models is a bottle-neck to achieve parameters optimization. For this reason, a new approach was proposed in this paper, in which the support vector machine was used to construct the response surface at first. Then it integrates the SVM-based response surface with the Sobol' method, i.e. the RSMSobol' method, to quantify the parameter sensitivities. In this work, the distributed time-variant gain model (DTVGM) was applied to the Huaihe River Basin, which was used as a case to verify its validity and feasibility. We selected three objective functions (i.e. water balance coefficient WB, Nash-Sutcliffe efficiency coefficient NS, and correlation coefficient RC) to assess the model performance as the output responses for sensitivity analysis. The results show that the parameters g(1) and g(2) are most important for all the objective functions, and they are almost the same to that of the classical approach. Furthermore, the RSMSobol method can not only achieve the quantification of the sensitivity, and also reduce the computational cost, with good accuracy compared to the classical approach. And this approach will be effective and reliable in the global sensitivity analysis for a complex modelling system.
出处	Journal of Geographical Sciences
卷	22
期	2
页	209-222
收录类别	SCI
语种	英语
ISSN号	1009-637X
内容类型	SCI/SSCI论文
源URL	[http://ir.igsnrr.ac.cn/handle/311030/26801]
专题	地理科学与资源研究所_历年回溯文献
推荐引用方式 GB/T 7714	Xia J.. An efficient global sensitivity analysis approach for distributed hydrological model. 2012.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。