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

We develop a new efficient methodology for Bayesian global sensitivity analysis for large-scale multivariate data. The focus is on computationally demanding models with correlated variables. 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 datasets, where we use cross-validation to determine the optimal degree of sparsity. This method was combined with a robust adaptive Metropolis algorithm coupled with a parallel implementation to speed up the convergence to the target distribution. The method was applied to a multivariate dataset from the IMPRESSIONS Integrated Assessment Platform (IAP2), an extension of the CLIMSAVE IAP, which has been widely applied in climate change impact, adaptation and vulnerability assessments. Our empirical results on synthetic and IAP2 data show that the proposed methods are efficient and accurate for global sensitivity analysis of complex models.

[1]  Derek Bingham,et al.  Efficient emulators of computer experiments using compactly supported correlation functions, with an application to cosmology , 2011, 1107.0749.

[2]  Benjamin Auder.  Classification et modélisation de sorties fonctionnelles de codes de calcul : application aux calculs thermo-hydrauliques accidentels dans les réacteurs à eau pressurisés (REP) , 2011 .

[3]  Marina Vannucci,et al.  Variable Selection for Nonparametric Gaussian Process Priors: Models and Computational Strategies. , 2011, Statistical science : a review journal of the Institute of Mathematical Statistics.

[4]  Ian P. Holman,et al.  Climate change impact modelling needs to include cross-sectoral interactions , 2016 .

[5]  Marie-Luce Taupin,et al.  Sensitivity analysis of spatio-temporal models describing nitrogen transfers, transformations and losses at the landscape scale , 2017, Environ. Model. Softw..

[6]  Ian P. Holman,et al.  Assessing cross-sectoral climate change impacts, vulnerability and adaptation: an introduction to the CLIMSAVE project , 2015, Climatic Change.

[7]  Aki Vehtari,et al.  Variable selection for Gaussian processes via sensitivity analysis of the posterior predictive distribution , 2017, AISTATS.

[8]  Guang Lin,et al.  Full scale multi-output Gaussian process emulator with nonseparable auto-covariance functions , 2015, J. Comput. Phys..

[9]  A. Saltelli,et al.  Making best use of model evaluations to compute sensitivity indices , 2002 .

[10]  A. Saltelli,et al.  Importance measures in global sensitivity analysis of nonlinear models , 1996 .

[11]  J. Rougier,et al.  Lightweight Emulators for Multivariate Deterministic Functions , 2007 .

[12]  Zhiyi Chi,et al.  Approximating likelihoods for large spatial data sets , 2004 .

[13]  Fabrice Gamboa,et al.  Sensitivity indices for multivariate outputs , 2013, 1303.3574.

[14]  J. Rougier.  Efficient Emulators for Multivariate Deterministic Functions , 2008 .

[15]  Supplementary Material to “Variable Prioritization in Nonlinear Black Box Methods: A Genetic Association Case Study” , 2018 .

[16]  T. Klein,et al.  Global sensitivity analysis: A novel generation of mighty estimators based on rank statistics , 2020, Bernoulli.

[17]  Zhenzhou Lu,et al.  Sensitivity analysis method for model with correlated inputs and multivariate output and its application to aircraft structure , 2019, Computer Methods in Applied Mechanics and Engineering.

[18]  Ilya M. Sobol,et al.  Sensitivity Estimates for Nonlinear Mathematical Models , 1993 .

[19]  N. Hamm,et al.  NONSEPARABLE DYNAMIC NEAREST NEIGHBOR GAUSSIAN PROCESS MODELS FOR LARGE SPATIO-TEMPORAL DATA WITH AN APPLICATION TO PARTICULATE MATTER ANALYSIS. , 2015, The annals of applied statistics.

[20]  Sudipto Banerjee,et al.  Practical Bayesian modeling and inference for massive spatial data sets on modest computing environments † , 2018, Stat. Anal. Data Min..

[21]  Charles R. Johnson,et al.  Matrix analysis , 1985, Statistical Inference for Engineers and Data Scientists.

[22]  Conrad Sanderson,et al.  Armadillo: a template-based C++ library for linear algebra , 2016, J. Open Source Softw..

[23]  Hong Xiao,et al.  Sensitivity analysis of correlated inputs: Application to a riveting process model , 2016 .

[24]  Kaichao Zhang,et al.  Multivariate output global sensitivity analysis using multi-output support vector regression , 2019, Structural and Multidisciplinary Optimization.

[25]  D. Nychka,et al.  Covariance Tapering for Interpolation of Large Spatial Datasets , 2006 .

[26]  Baptiste Broto,et al.  Variance Reduction for Estimation of Shapley Effects and Adaptation to Unknown Input Distribution , 2018, SIAM/ASA J. Uncertain. Quantification.

[27]  Andrew O. Finley,et al.  Applying Nearest Neighbor Gaussian Processes to Massive Spatial Data Sets: Forest Canopy Height Prediction Across Tanana Valley Alaska , 2017 .

[28]  M. Trnka,et al.  Direct and indirect impacts of climate and socio-economic change in Europe: a sensitivity analysis for key land- and water-based sectors , 2015, Climatic Change.

[29]  Carl E. Rasmussen,et al.  In Advances in Neural Information Processing Systems , 2011 .

[30]  Sonja Kuhnt,et al.  Design and analysis of computer experiments , 2010 .

[31]  Zhenzhou Lu,et al.  Multivariate Global Sensitivity Analysis Based on Distance Components Decomposition , 2018, Risk analysis : an official publication of the Society for Risk Analysis.

[32]  Matti Vihola,et al.  Robust adaptive Metropolis algorithm with coerced acceptance rate , 2010, Statistics and Computing.

[33]  Fei Liu.  A Dynamic Modelling Strategy for Bayesian Computer Model Emulation , 2008 .

[34]  Marian Farah,et al.  Bayesian Inference for Sensitivity Analysis of Computer Simulators, With an Application to Radiative Transfer Models , 2014, Technometrics.

[35]  G. Henze,et al.  SAMPLING BASED ON SOBOL 0 SEQUENCES FOR MONTE CARLO TECHNIQUES APPLIED TO BUILDING SIMULATIONS , 2011 .

[36]  Zhenzhou Lu,et al.  Generalized sensitivity indices based on vector projection for multivariate output , 2019, Applied Mathematical Modelling.

[37]  T. Gneiting.  Compactly Supported Correlation Functions , 2002 .

[38]  G. Moreaux,et al.  Compactly supported radial covariance functions , 2008 .

[39]  R. Stocki,et al.  A method to improve design reliability using optimal Latin hypercube sampling , 2005 .

[40]  D. Rubin,et al.  Inference from Iterative Simulation Using Multiple Sequences , 1992 .

[41]  P. Harrison,et al.  Differences between low-end and high-end climate change impacts in Europe across multiple sectors , 2019, Regional Environmental Change.

[42]  A. Gelman,et al.  Rank-normalization, folding, and localization: An improved R-hat for assessing convergence Rank-Normalization, Folding, and Localization: An Improved (cid:2) R for Assessing Convergence of MCMC An assessing for assessing An improved (cid:2) R for assessing convergence of MCMC , 2020 .

[43]  T. Santner,et al.  Estimating sensitivity indices based on Gaussian process metamodels with compactly supported correlation functions , 2014 .

[44]  Andrew Gelman,et al.  General methods for monitoring convergence of iterative simulations , 1998 .

[45]  David Makowski,et al.  Multivariate sensitivity analysis to measure global contribution of input factors in dynamic models , 2011, Reliab. Eng. Syst. Saf..

[46]  Hoon Kim,et al.  Monte Carlo Statistical Methods , 2000, Technometrics.

[47]  William Becker,et al.  Why so many published sensitivity analyses are false: A systematic review of sensitivity analysis practices , 2017, Environ. Model. Softw..

[48]  Antony M. Overstall,et al.  Multivariate emulation of computer simulators: model selection and diagnostics with application to a humanitarian relief model , 2015, Journal of the Royal Statistical Society. Series C, Applied statistics.

[49]  Conrad Sanderson,et al.  RcppArmadillo: Accelerating R with high-performance C++ linear algebra , 2014, Comput. Stat. Data Anal..

[50]  A. V. Vecchia.  Estimation and model identification for continuous spatial processes , 1988 .

[51]  Mark Girolami,et al.  Semi-Exact Control Functionals From Sard’s Method , 2020, Biometrika.

[52]  A. O'Hagan,et al.  Probabilistic sensitivity analysis of complex models: a Bayesian approach , 2004 .

[53]  Douglas W. Nychka,et al.  Covariance Tapering for Likelihood-Based Estimation in Large Spatial Data Sets , 2008 .