Reducing uncertainty in local climate projections

Abstract

Planning for adaptation to climate change requires accurate climate projections. Recent studies have shown that the uncertainty in global mean surface temperature projections can be considerably reduced by using historical observations. However, the transposition of these new results to the local scale is not yet available. Here we adapt an innovative statistical method that combines the latest generation of climate model simulations, global observations, and local observations to reduce uncertainty in local temperature projections. By taking advantage of the tight links between local and global temperature, we can derive the local implications of global constraints. The model uncertainty is reduced by 30\% up to 50\% at any location worldwide, allowing to substantially improve the quantification of risks associated with future climate change. A rigorous evaluation of these results within a perfect model framework indicates a robust skill, leading to a high confidence in our constrained climate projections.

Revisiting the Potential to Narrow Model Uncertainty in the Projections of Arctic Runoff

“…The second method (KCC) was developed by Ribes et al. (2021) and Qasmi and Ribes (2022) and is based on Bayesian statistics where a prior distribution, π(x), of the forced response to anthropogenic forcings is first derived from raw model outputs and then constrained directly with observations of one or more variables (here both observed global mean surface temperature and reconstructed basin‐wide average runoff). The prior is estimated using a Generalized Additive Model (assuming the additivity of the model responses to individual forcings) and a simple Energy Budget Model (allowing us to diagnose the runoff response to volcanic eruptions; for more details, see Supporting Information S1 from Ribes et al., 2021).…”

“…After a brief description of the various datasets and statistical methods (Section 2), the L19 limitations are first shown by using two runoff definitions and two GCM generations over four northern high‐latitude river basins (Section 3). The L19 results are also compared to those obtained with the alternative Kriging for Climate Change (KCC; Ribes et al., 2021; Qasmi & Ribes, 2022) method developed at CNRM. Finally (Section 4), the reasons for the better behavior of KCC are briefly discussed and the method is also applied to constrain runoff projections over the aggregated “Arctic” basin (i.e., river basins whose outlet is in the Arctic ocean or neighboring seas).…”