ICON Simulations Suggest Model-Dependent Cloud Feedbacks in Global Storm-Resolving Models

Climate feedbacks, key determinants of Earth’s temperature response to radiative forcing, remain uncertain, largely because of challenges in representing cloud and convective processes. Global storm-resolving models (GSRMs), which explicitly resolve deep convection, have recently emerged as a promising alternative to conventional general circulation models to reduce uncertainties in clouds, convection, and possibly also climate feedbacks. Here, we perform control and +4K sea surface temperature perturbation experiments at two horizontal resolutions to evaluate climate feedbacks in the ICOsahedral Non-hydrostatic model (ICON) GSRM with convection parameterization disabled and compare them with models from the Coupled Model Intercomparison Project phase 6 (CMIP6), and two independent GSRMs. The results show that ICON exhibits stronger radiative damping and lower climate sensitivity than most CMIP6 models, with net feedbacks near the lower end of the CMIP6 range. This response is partly due to more negative lapse rate and relative humidity feedbacks. Compared to other available GSRMs, ICON exhibits a different balance between shortwave and longwave feedbacks, even when net cloud feedbacks are similar, highlighting the sensitivity of climate feedbacks to model formulation and resolution across GSRMs. Spatially, ICON’s cloud feedbacks display characteristic regional contrasts driven by diverse known processes, including rising cloud tops, reduced tropical cloud anvils, enhanced land-sea warming contrasts, and increased low-cloud reflectivity in marine stratus regions and the extratropics. Finally, despite variations in spatial resolution and simulation length, ICON’s climate feedbacks and underlying processes remain remarkably consistent. These findings highlight the need for coordinated GSRM intercomparisons to better constrain real-world climate feedback estimates.