Assessing the consistency of CORDEX multidomain projections in overlapping regions worldwide

CORDEX users are confronted with multiple sources of climate change information in regions where multiple domains overlap. Assessing the consistency of these sources (particularly the consistency of the climate signals) and understanding potential conflicts is a relevant problem with practical implications. For instance, this knowledge will guide on the best use of CORDEX to produce worldwide information merging the results provided by the different CORDEX domains. Two main approaches have been followed in the literature: 1) Mosaic of overlapped domains: The results from different domains are overlaid producing a mosaic where each region is covered by a single domain; this is the procedure typically followed in CORDEX-CORE (Teichmann et. al., 2021), using the domain which is best suited for each region. 2) Grand ensemble (Spinoni et al., 2020): Pooling together all available simulations across domains for each gridbox. This approach maximizes the information but leads to a heterogeneous ensemble with varying size and members across regions which may create spatial artifacts (e.g. border effects).

A preliminary analysis by Legasa et al. (2020) quantified the changes/uncertainty related to the choice of domain in the Mediterranean area, using the Europe and Africa CORDEX domains. They showed that the variability of the climate change signal from the grand ensemble was mostly determined by the models, and less so by the domain choice. Therefore, there is some evidence (at least, in the Mediterranean) that the grand ensemble approach could be appropriate to enlarge the ensembles for specific regions by pooling multi-domain simulations.

Here we extend this analysis by considering all regions where the worldwide CORDEX dataset domains overlap. The new subcontinental climatic regions used in the IPCC AR6 (Iturbide et al., 2020) are used to aggregate the results. We show that, in these areas, precipitation and near-surface air temperature biases and, especially, future climate change projections are systematically similar for simulations performed with the same GCM-RCM pair over different overlapping domains. This consistency provides higher confidence in the regional results (particularly when there are no physical reasons –e.g. different parameterizations– explaining the differences) and supports the use of a grand ensemble, pooling all available simulations in overlap areas covered by small individual CORDEX ensembles.

References

Iturbide, M., et al. (2020) An update of IPCC climate reference regions for subcontinental analysis of climate model data: definition and aggregated datasets. Earth System Science Data, 12, 2959–2970, https://doi.org/10.5194/essd-12-2959-2020

Legasa, M.N., et al. (2020) Assessing multidomain overlaps and grand ensemble generation in CORDEX regional projections. Geophys. Res. Lett. 47,  https://doi.org/10.1029/2019GL086799 

Spinoni, J., et al. (2020). Future global meteorological drought hot spots: A study based on CORDEX data. Journal of Climate, 33 (9), pp. 3635-3661, https://doi.org/10.1175/JCLI-D-19-0084.1

Teichmann, C., et al. (2021). Assessing mean climate change signals in the global CORDEX-CORE ensemble. Climate Dynamics, 57 (5-6), pp. 1269-1292, https://doi.org/10.1007/s00382-020-05494-x

Acknowledgement

The authors would like to thank the Copernicus Climate Change Service (C3S) for funding part of this research. J.F. and A.S.C acknowledge project CORDyS (PID2020-116595RB-I00). J.M.G. and M.I. acknowledge project ATLAS (PID2019-111481RB-I00) funded by MCIN/AEI/10.13039/501100011033 and A.S.C and E.C. acknowledge project IS-ENES3 funded by the EU H2020 (#824084).