More accurate quantification of model-to-model agreement in externally forced climatic responses over the coming century

Nicola Maher,Scott B Power,Jochem Marotzke

doi:10.1038/s41467-020-20635-w

Abstract

Separating how model-to-model differences in the forced response (UMD) and internal variability (UIV) contribute to the uncertainty in climate projections is important, but challenging. Reducing UMD increases confidence in projections, while UIV characterises the range of possible futures that might occur purely by chance. Separating these uncertainties is limited in traditional multi-model ensembles because most models have only a small number of realisations; furthermore, some models are not independent. Here, we use six largely independent single model initial-condition large ensembles to separate the contributions of UMD and UIV in projecting 21st-century changes of temperature, precipitation, and their temporal variability under strong forcing (RCP8.5). We provide a method that produces similar results using traditional multi-model archives. While UMD is larger than UIV for both temperature and precipitation changes, UIV is larger than UMD for the changes in temporal variability of both temperature and precipitation, between 20° and 80° latitude in both hemispheres. Over large regions and for all variables considered here except temporal temperature variability, models agree on the sign of the forced response whereas they disagree widely on the magnitude. Our separation method can readily be extended to other climate variables.

Highlights

Separating how model-to-model differences in the forced response (UMD) and internal variability (UIV) contribute to the uncertainty in climate projections is important, but challenging
We calculate the projected change in an individual ensemble member: the forced response both in each individual single model initial-condition large ensembles (SMILEs) and across the six SMILEs, which is an estimate of the response due to external forcing alone, and the uncertainty in projections due to model-tomodel differences in the forced response (UMD) and internal variability (UIV)
We find that for the mean-state projections, Uncertainty due to internal variability (UIV) is generally smaller and uncertainty due to model-to-model differences (UMD) is larger in the SMILE analysis (Supplementary Fig. 11a, b, e, f), likely due to the overestimation of internal variability in the CMIP5 analysis

Summary

Introduction

Separating how model-to-model differences in the forced response (UMD) and internal variability (UIV) contribute to the uncertainty in climate projections is important, but challenging. Reducing UMD increases confidence in projections, while UIV characterises the range of possible futures that might occur purely by chance Separating these uncertainties is limited in traditional multi-model ensembles because most models have only a small number of realisations; some models are not independent. We use six largely independent single model initial-condition large ensembles to separate the contributions of UMD and UIV in projecting 21st-century changes of temperature, precipitation, and their temporal variability under strong forcing (RCP8.5). We quantify how model-to-model differences and internal variability contribute to the uncertainty in projections of temperature and precipitation and their temporal variability, without relying on the assumptions of previous methods. Lehner et al.[13] recently demonstrated that the regional errors in the partitioning of uncertainty can be as large as 50% using the traditional approaches

Methods

Results

Conclusion