Computer model calibration with large non-stationary spatial outputs: application to the calibration of a climate model

TL;DR

This paper presents an efficient Bayesian calibration method for large, non-stationary spatial outputs, exemplified through climate model calibration, by employing basis representations and advanced statistical techniques.

Contribution

It introduces a novel calibration approach that handles large, non-stationary spatial data efficiently using basis decompositions and INLA-SPDE methods.

Findings

Improved calibration accuracy for large spatial outputs.

Effective modeling of non-stationary spatial variation.

Validated approach with synthetic and climate data.

Abstract

Bayesian calibration of computer models tunes unknown input parameters by comparing outputs with observations. For model outputs that are distributed over space, this becomes computationally expensive because of the output size. To overcome this challenge, we employ a basis representation of the model outputs and observations: we match these decompositions to carry out the calibration efficiently. In the second step, we incorporate the non-stationary behaviour, in terms of spatial variations of both variance and correlations, in the calibration. We insert two integrated nested Laplace approximation-stochastic partial differential equation parameters into the calibration. A synthetic example and a climate model illustration highlight the benefits of our approach.