More positive climate feedback with higher resolution: A multi-resolution GSRM study with ICON in comparison to conventional models and other GSRMs

Global storm-resolving models (GSRMs) represent a frontier in climate change research, but their application remains limited due to high computational cost, and inter-GSRM comparisons are almost nonexistent. Moreover, a potential sensitivity of climate feedback in GSRMs of the Earth’s atmosphere to model resolution hasn’t been studied, yet.

In this study, we conducted AMIP and AMIP+4K experiments using the GSRM ICON, where the AMIP+4K experiment imposes a globally uniform sea surface temperature increase of 4 K. Each experiment was performed at three different horizontal resolutions: 20 km, 10 km, and 5 km.

Results show that for the AMIP+4K experiment, the net climate feedback parameter as well as its shortwave and longwave components all become more positive with increasing resolution. The difference in net climate feedback parameter between 20 km and 5 km resolution is comparable in magnitude to the model spread of climate feedback parameter in CMIP6 AMIP+4K experiments. The resolution dependence of the shortwave feedback in AMIP+4K experiment originates in the extratropics while the dependence of the longwave feedback is a result of tropical processes.

Regarding comparison with conventional models, the climate feedback parameter of ICON at 10km and 5km resolution falls within the model spread of CMIP6 AMIP+4K. However, over the extratropical oceans, ICON at all resolutions exhibits clearly stronger negative feedback than any of the CMIP6 models. Furthermore, the climate feedback from ICON at 5 km resolution is very close to that of another GSRM, X-SHiELD, at 3.4 km resolution. Nevertheless, the shortwave and longwave components differ significantly between the two models, indicating that even without convection parameterization—a key source of uncertainty—there is still notable inter-model variability in the representation of climate feedbacks.