目标函数对新安江模型参数敏感性和不确定性影响分析
DOI:
作者:
作者单位:

1.三峡大学水利与环境学院;2.南京水利科学研究院水文水资源与水利工程科学国家重点实验室

作者简介:

通讯作者:

中图分类号:

P333

基金项目:

湖北省教育厅科学技术研究重点项目(D20211205);国家自然科学基金项目(51609124)


Analysis of the effect of objective function on the sensitivity and uncertainty of Xin'anjiang model parameters
Author:
Affiliation:

1.College of Hydraulic and Environmental Engineering, China Three Gorges University;2.State Key Laboratory of Hydrology, Water Resources and Hydraulic Engineering, Nanjing Hydraulic Research Institute

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    参数率定目标函数的选择对于建立流域水文模型至关重要。为入探究不同目标函数对水文模型参数敏感性及径流模拟不确定性的影响,本文以福建省金溪池潭流域日径流模拟为例,采用SCEM-UA优化算法率定三水源新安江模型,选择纳什效率系数(NSE)、均方根误差(RMSE)和Kling-Gupta效率系数(KGE)作为率定目标函数,同时构建分别偏好NSE、RMSE、KGE三者的组合多目标函数F1、F2和F3共6个率定目标,比较不同率定目标下的径流模拟精度差异,采用Sobol全局敏感性分析方法定量对比分析不同率定目标函数下的新安江模型参数敏感性,最后基于GLUE方法分析丰枯水期和不同流量级别下的径流模拟不确定性。结果表明:(1)6个率定目标函数下,新安江模型的主要敏感参数识别结果一致,多目标函数F1,F2,F3在参数敏感程度及敏感性排序表现出一致性;(2)F1,F2,F3较单目标函数展现出更高的径流模拟精度,F3率定期和验证期决定性系数达0.85,径流总量相对误差为0.45%和7.12%,模拟精度最优,F1和F2次之;(3)以NSE为率定目标所得到的径流模拟不确定性结果,在不同水文时期和不同流量级别下表现出较其他目标函数更高的覆盖率和较小的不确定性区间。不同目标函数下均呈现枯水期参数不确定性最小,丰水期最大,且随流量级别的增大而径流模拟不确定性增加的规律。

    Abstract:

    The choice of the parameter rate setting objective function is crucial to the development of the basin hydrological model. In order to investigate the influence of different objective functions on the sensitivity of hydrological model parameters and uncertainty of runoff simulation, this paper takes the daily runoff simulation of Jinxi Chitan watershed in Fujian Province as an example, and uses the SCEM-UA optimization algorithm to rate the three-source Xin'anjiang model, and selects Nash efficiency coefficient (NSE), root mean square error (RMSE) and Kling-Gupta efficiency coefficient (KGE) were selected as the rate objective functions, and a combined multi-objective function F1, F2 and F3 with preferences for NSE, RMSE and KGE are also constructed for a total of six rate objectives to compare the differences in runoff simulation accuracy under different rate objectives.. And the Sobol global sensitivity analysis method was used to quantitatively compare and analyze the sensitivity of the Xin'anjiang model parameters under different rate objective functions. Finally, we analyzed the differences in runoff simulation uncertainties under different flow levels and abundant and dry periods based on the GLUE method. The results show that (1) the identification of the main sensitive parameters of the Xin'anjiang model under six rate-determined objective functions is consistent, and the multi-objective functions F1, F2, and F3 show consistency in parameter sensitivity and sensitivity ranking; (2) F1, F2, and F3 show higher runoff simulation accuracy than the single objective function, and the deterministic coefficients of F3 rate period and validation period reach 0.85, and the relative errors of total runoff are 0.45% and 7.12%, with the best simulation accuracy, F1 and F2. are the next best; (3) The uncertainty results of runoff simulation obtained with NSE as the rate objective show higher coverage and smaller uncertainty interval than other objective functions under different hydrological periods and different flow levels. The uncertainty of the parameters under different objective functions is the smallest in the dry period and the largest in the rich period, and the uncertainty of the runoff simulation increases with the increase of the flow level.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2023-05-29
  • 最后修改日期:2023-08-23
  • 录用日期:2023-08-23
  • 在线发布日期:
  • 出版日期: