Towards process-informed bias correction of climate change simulations

Abstract

Biases in climate model simulations introduce biases in subsequent impact simulations. Therefore, bias correction methods are operationally used to post-process regional climate projections. However, many problems have been identified, and some researchers question the very basis of the approach. Here we demonstrate that a typical cross-validation is unable to identify improper use of bias correction. Several examples show the limited ability of bias correction to correct and to downscale variability, and demonstrate that bias correction can cause implausible climate change signals. Bias correction cannot overcome major model errors, and naive application might result in ill-informed adaptation decisions. We conclude with a list of recommendations and suggestions for future research to reduce, post-process, and cope with climate model biases

Multivariate bias adjustment of high-dimensional climate simulations: the Rank Resampling for Distributions and Dependences (R<sup>2</sup>D<sup>2</sup>) bias correction

“…Hence, by such an approach, R 2 D 2 would be applied in a conditional process-oriented BC framework. Indeed, if the present study focused on the methodological aspects of the multivariate bias correction, it is worth keeping in mind that any application of a BC method should be performed with some physically based motivations: depending on their intrinsic skills to model specific features, some climate simulations cannot sensibly be corrected, especially in climate change context where artifacts of bias correction may appear while not visible in present climate evaluations (e.g., Maraun et al, 2017). So the development of BC methodologies allowing one to include some physics in the adjustment procedure is an important perspective of research, in order to have BC approaches not used as black boxes while they should be a support to increase the realism of the climate simulations based on physical knowledge.…”

Climate pattern-scaling set for an ensemble of 22 GCMs – adding uncertainty to the IMOGEN version 2.0 impact system

“…This does, though, assume such GCM bias removal is valid for the entirety of any transient simulation. Recent analyses appeal for more process information to be accounted for when attempting bias correction (Maraun et al, 2017). …”

Multivariate bias adjustment of high-dimensional climate simulations: The Rank Resampling for Distributions and Dependences (R<sup>2</sup>D<sup>2</sup>) Bias Correction

“…Bias adjustments of the whole distribution through quantile-mapping techniques have been quite popular since it allows adjusting not only on the mean and variance but also any quantile of the variable of interest. Hence, many variants have been proposed (e.g., Déqué, 2007;Michelangeli et al, 2009;Kallache et al, 2011;Tramblay et al, 2013;Vrac et al, 2016) and applied in different studies (e.g., 20 Oettli et al, 2011; Colette et al, 2012;Tisseuil et al, 2012;Vigaud et al, 2013). Nevertheless, usually, those approaches only work in a univariate context, which means that they are designed to correct independently one variable at a time, for one location (e.g., grid cell) at a time.…”

“…Indeed, depending on their intrinsic skills to model specific features, some climate simulations cannot sensibly be corrected, especially in 5 climate change context where artifacts of bias correction may appear while not visible in present climate evaluations (e.g., Maraun et al, 2017). So the development of BC methodologies allowing to include some physics in the adjustment procedure is an important perspective of research, in order to have BC approaches not used as "black boxes" while they should be a support to increase the realism of the climate simulations based on physical knowledge.…”