Surrogate-Based Bayesian Inverse Modeling of the Hydrological System: An Adaptive Approach Considering Surrogate Approximation Error

被引:38
作者
Zhang, Jiangjiang [1 ,2 ]
Zheng, Qiang [1 ,2 ]
Chen, Dingjiang [1 ,3 ]
Wu, Laosheng [4 ]
Zeng, Lingzao [1 ,2 ]
机构
[1] Zhejiang Univ, Inst Soil & Water Resources & Environm Sci, Coll Environm & Resource Sci, Hangzhou, Peoples R China
[2] Zhejiang Univ, Zhejiang Prov Key Lab Agr Resources & Environm, Hangzhou, Peoples R China
[3] Zhejiang Univ, Minist Educ, Key Lab Environm Remediat & Ecol Hlth, Hangzhou, Peoples R China
[4] Univ Calif Riverside, Dept Environm Sci, Riverside, CA 92521 USA
基金
中国博士后科学基金; 中国国家自然科学基金;
关键词
MONTE-CARLO-SIMULATION; HYDRAULIC CONDUCTIVITY; UNCERTAINTY ASSESSMENT; MECHANICAL BEHAVIORS; MARGINAL LIKELIHOOD; EXPERIMENTAL-DESIGN; DATA ASSIMILATION; WATER-RESOURCES; EFFICIENT; OPTIMIZATION;
D O I
10.1029/2019WR025721
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Bayesian inverse modeling is important for a better understanding of hydrological processes. However, this approach can be computationally demanding, as it usually requires a large number of model evaluations. To address this issue, one can take advantage of surrogate modeling techniques. Nevertheless, when approximation error of the surrogate model is neglected, the inversion result will be biased. In this paper, we develop a surrogate-based Bayesian inversion framework that explicitly quantifies and gradually reduces the approximation error of the surrogate. Specifically, two strategies are proposed to quantify the surrogate error. The first strategy works by quantifying the surrogate prediction uncertainty with a Bayesian method, while the second strategy uses another surrogate to simulate and correct the approximation error of the primary surrogate. By adaptively refining the surrogate over the posterior distribution, we can gradually reduce the surrogate approximation error to a small level. Demonstrated with three case studies involving high dimensionality, multimodality, and a real-world application, it is found that both strategies can reduce the bias introduced by surrogate approximation error, while the second strategy that integrates two methods (i.e., polynomial chaos expansion and Gaussian process in this work) that complement each other shows the best performance.
引用
收藏
页数:25
相关论文
共 82 条
[1]   Reservoir Optimization in Water Resources: a Review [J].
Ahmad, Asmadi ;
El-Shafie, Ahmed ;
Razali, Siti Fatin Mohd ;
Mohamad, Zawawi Samba .
WATER RESOURCES MANAGEMENT, 2014, 28 (11) :3391-3405
[2]  
[Anonymous], 1999, DOCUMENTATION USERS
[3]  
[Anonymous], 2024, Uncertainty Quantification: Theory, Implementation, and Applications
[4]  
[Anonymous], 1995, TECHNOMETRICS, DOI [DOI 10.1080/00401706.1995.10485913, 10.1080/00401706.1995. 10485913]
[5]  
[Anonymous], 2006, Gaussian processes for machine learning, DOI DOI 10.1142/S0129065704001899
[6]  
[Anonymous], 2009, DATA ASSIMILATION EN
[7]   A review of surrogate models and their application to groundwater modeling [J].
Asher, M. J. ;
Croke, B. F. W. ;
Jakeman, A. J. ;
Peeters, L. J. M. .
WATER RESOURCES RESEARCH, 2015, 51 (08) :5957-5973
[8]   Sparse polynomial chaos expansions and adaptive stochastic finite elements using a regression approach [J].
Blatman, Geraud ;
Sudret, Bruno .
COMPTES RENDUS MECANIQUE, 2008, 336 (06) :518-523
[9]   Adaptive sparse polynomial chaos expansion based on least angle regression [J].
Blatman, Geraud ;
Sudret, Bruno .
JOURNAL OF COMPUTATIONAL PHYSICS, 2011, 230 (06) :2345-2367
[10]  
Brooks S, 2011, CH CRC HANDB MOD STA, pXIX