Parametric uncertainty in complex environmental models: a cheap emulation approach for models with high-dimensional output

In order to understand underlying processes governing environmental and physical processes, and predict future outcomes, a complex computer model is frequently required to simulate these dynamics. However there is inevitably uncertainty related to the exact parametric form or the values of such parameters to be used when developing these simulators, with \emph{ranges} of plausible values prevalent in the literature. Systematic errors introduced by failing to account for these uncertainties have the potential to have a large effect on resulting estimates in unknown quantities of interest. Due to the complexity of these types of models, it is often unfeasible to run large numbers of training runs that are usually required for full statistical emulators of the environmental processes. We therefore present a method for accounting for uncertainties in complex environmental simulators without the need for very large numbers of training runs and illustrate the method through an application to the Met Office's atmospheric transport model NAME. We conclude that there are two principle parameters that are linked with variability in NAME outputs, namely the free tropospheric turbulence parameter and particle release height. Our results suggest the former should be significantly larger than is currently implemented as a default in NAME, whilst changes in the latter most likely stem from inconsistencies between the model specified ground height at the observation locations and the true height at this location. Estimated discrepancies from independent data are consistent with the discrepancy between modelled and true ground height.

[1]  J. Moncrieff,et al.  Quantifying methane and nitrous oxide emissions from the UK and Ireland using a national-scale monitoring network , 2015 .

[2]  Serge Guillas,et al.  Dimension reduction for emulation: application to the influence of bathymetry on tsunami heights , 2016 .

[3]  David J. Thomson,et al.  The U.K. Met Office's Next-Generation Atmospheric Dispersion Model, NAME III , 2007 .

[4]  Thomas J. Santner,et al.  Algorithms for Generating Maximin Latin Hypercube and Orthogonal Designs , 2011 .

[5]  Christophe Andrieu,et al.  A tutorial on adaptive MCMC , 2008, Stat. Comput..

[6]  A. Manning,et al.  Estimation of trace gas fluxes with objectively determined basis functions using reversible jump Markov Chain Monte Carlo , 2016 .

[7]  Serge Guillas,et al.  Dimension Reduction for Gaussian Process Emulation: An Application to the Influence of Bathymetry on Tsunami Heights , 2016, SIAM/ASA J. Uncertain. Quantification.

[8]  Richard J. Beckman,et al.  A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output From a Computer Code , 2000, Technometrics.

[9]  Yuhong Yang Can the Strengths of AIC and BIC Be Shared , 2005 .

[10]  Jens Mühle,et al.  Characterization of uncertainties in atmospheric trace gas inversions using hierarchical Bayesian methods , 2013 .

[11]  K. Fukumizu,et al.  Gradient-Based Kernel Dimension Reduction for Regression , 2014 .

[12]  N. Cressie,et al.  Spatio-temporal bivariate statistical models for atmospheric trace-gas inversion , 2015, 1509.00915.

[13]  Astrid Maute,et al.  Expert Knowledge and Multivariate Emulation: The Thermosphere–Ionosphere Electrodynamics General Circulation Model (TIE-GCM) , 2009, Technometrics.

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

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

[16]  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.

[17]  Thomas J. Santner,et al.  The Design and Analysis of Computer Experiments , 2003, Springer Series in Statistics.

[18]  David C. Woods,et al.  Emulation of Multivariate Simulators Using Thin-Plate Splines with Application to Atmospheric Dispersion , 2015, SIAM/ASA J. Uncertain. Quantification.