Multivariate sensitivity analysis for a large-scale climate impact and adaptation model

Oluwole Kehinde Oyebamiji; Christopher Nemeth; Paula A Harrison; Robert W Dunford; George Cojocaru

doi:10.1093/jrsssc/qlad032

Multivariate sensitivity analysis for a large-scale climate impact and adaptation model

Oluwole Kehinde Oyebamiji, Christopher Nemeth, Paula A Harrison, Robert W Dunford, George Cojocaru

Computer Science

Research output: Contribution to journal › Article › peer-review

Abstract

We apply a new efficient methodology for Bayesian global sensitivity analysis for large-scale multivariate data. A multivariate Gaussian process is used as a surrogate model to replace the expensive computer model. To improve the computational efficiency and performance of the model, compactly supported correlation functions are used. The goal is to generate sparse matrices, which give crucial advantages when dealing with large data sets. The method was applied to multivariate data from the IMPRESSIONS Integrated Assessment Platform version 2. Our empirical results on Integrated Assessment Platform version 2 data show that the proposed methods are efficient and accurate for global sensitivity analysis of complex models.

Original language	English
Pages (from-to)	770–808
Number of pages	39
Journal	Journal of the Royal Statistical Society Series C (Applied Statistics)
Volume	72
Issue number	3
Early online date	15 May 2023
DOIs	https://doi.org/10.1093/jrsssc/qlad032
Publication status	Published - Jun 2023

Bibliographical note

Funding:
The authors gratefully acknowledge the support of EPSRC and NERC grants EP/R01860X/1, EP/S00159X/1, EP/V022636/1, and NE/T004002/1.

Keywords

Bayesian methods
compactly supported correlation function
Gaussian process
robust adaptive MCMC
sensitivity analysis

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1093/jrsssc/qlad032

Cite this

@article{a71b34471a724264a7fafa808f8daf96,

title = "Multivariate sensitivity analysis for a large-scale climate impact and adaptation model",

abstract = "We apply a new efficient methodology for Bayesian global sensitivity analysis for large-scale multivariate data. A multivariate Gaussian process is used as a surrogate model to replace the expensive computer model. To improve the computational efficiency and performance of the model, compactly supported correlation functions are used. The goal is to generate sparse matrices, which give crucial advantages when dealing with large data sets. The method was applied to multivariate data from the IMPRESSIONS Integrated Assessment Platform version 2. Our empirical results on Integrated Assessment Platform version 2 data show that the proposed methods are efficient and accurate for global sensitivity analysis of complex models.",

keywords = "Bayesian methods, compactly supported correlation function, Gaussian process, robust adaptive MCMC, sensitivity analysis",

author = "Oyebamiji, {Oluwole Kehinde} and Christopher Nemeth and Harrison, {Paula A} and Dunford, {Robert W} and George Cojocaru",

note = "Funding: The authors gratefully acknowledge the support of EPSRC and NERC grants EP/R01860X/1, EP/S00159X/1, EP/V022636/1, and NE/T004002/1.",

year = "2023",

month = jun,

doi = "10.1093/jrsssc/qlad032",

language = "English",

volume = "72",

pages = "770–808",

journal = "Journal of the Royal Statistical Society Series C (Applied Statistics)",

issn = "0035-9254",

publisher = "Wiley",

number = "3",

}

TY - JOUR

T1 - Multivariate sensitivity analysis for a large-scale climate impact and adaptation model

AU - Oyebamiji, Oluwole Kehinde

AU - Nemeth, Christopher

AU - Harrison, Paula A

AU - Dunford, Robert W

AU - Cojocaru, George

N1 - Funding: The authors gratefully acknowledge the support of EPSRC and NERC grants EP/R01860X/1, EP/S00159X/1, EP/V022636/1, and NE/T004002/1.

PY - 2023/6

Y1 - 2023/6

N2 - We apply a new efficient methodology for Bayesian global sensitivity analysis for large-scale multivariate data. A multivariate Gaussian process is used as a surrogate model to replace the expensive computer model. To improve the computational efficiency and performance of the model, compactly supported correlation functions are used. The goal is to generate sparse matrices, which give crucial advantages when dealing with large data sets. The method was applied to multivariate data from the IMPRESSIONS Integrated Assessment Platform version 2. Our empirical results on Integrated Assessment Platform version 2 data show that the proposed methods are efficient and accurate for global sensitivity analysis of complex models.

AB - We apply a new efficient methodology for Bayesian global sensitivity analysis for large-scale multivariate data. A multivariate Gaussian process is used as a surrogate model to replace the expensive computer model. To improve the computational efficiency and performance of the model, compactly supported correlation functions are used. The goal is to generate sparse matrices, which give crucial advantages when dealing with large data sets. The method was applied to multivariate data from the IMPRESSIONS Integrated Assessment Platform version 2. Our empirical results on Integrated Assessment Platform version 2 data show that the proposed methods are efficient and accurate for global sensitivity analysis of complex models.

KW - Bayesian methods

KW - compactly supported correlation function

KW - Gaussian process

KW - robust adaptive MCMC

KW - sensitivity analysis

U2 - 10.1093/jrsssc/qlad032

DO - 10.1093/jrsssc/qlad032

M3 - Article

SN - 0035-9254

VL - 72

SP - 770

EP - 808

JO - Journal of the Royal Statistical Society Series C (Applied Statistics)

JF - Journal of the Royal Statistical Society Series C (Applied Statistics)

IS - 3

ER -

Multivariate sensitivity analysis for a large-scale climate impact and adaptation model

Abstract

Bibliographical note

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this