Unraveling the striking difference in causal low-cloud susceptibility to aerosol between GCM and observation

Aerosol-cloud interactions (ACI) remain to be the largest source of uncertainty in anthropogenic forcing to the Earth’s energy budget. While large-scale observation-based studies grapple with the challenge of distinguishing between correlation and causation, general circulation models (GCMs) struggle to accurately represent intricate microphysical processes involved in ACI. In this work, we adopt a novel data-driven approach to derive causal cloud susceptibility to cloud droplet number concentration perturbations for marine warm clouds based on ERA5 meteorological fields and CERES-MODIS observed cloud properties. We then apply the same approach on GCM outputs of meteorological and cloud fields. Striking differences are observed in cloud susceptibilities derived from GCM outputs compared to observational datasets using the same data-driven framework. We will show results from exercises aiming to elucidate the underlying factors contributing to the difference, which will provide important insights on (micro- and/or macro-physical) process-representation within climate models – crucial for reducing climate change uncertainty. Moreover, our approach showcases the potential of leveraging observational datasets to infer causal information on ACI, offering a promising avenue for climate model improvement.