Assessing CESM2 Clouds and Their Response to Climate Change Using Cloud Regimes
CESM2 has a very high climate sensitivity that is associated with strong cloud feedbacks. This study uses a clustering method to compare the model’s clouds to observations and characterize how the clouds change with warming.
Using an improved cloud regime methodology exposes the biases in CESM2’s present-day clouds. Essentially the model is unable to distinguish some cloud regimes, leading to overestimation in the frequency of a few cloud regimes. With warming, low-level clouds shift toward regimes that are optically thinner, and this finding is robust when using cloud regimes defined from three different satellite products.
Cloud regimes are defined using three satellite products (ISCCP, MODIS, and MISR). For each satellite data set, the cloud-top-pressure-vs-optical-depth histograms are clustered into a few regimes. This follows previous approaches, but this study introduces the Wasserstein distance as an improved measure of distance between histograms. The improvement reduces the tendency to have one “catch-all” category that in previous studies contains a very large fraction of histograms with small cloud cover. Using the improved observational clusters, cloud simulator output from CESM2 is classified into the cloud regimes. The model has difficulty capturing the differences that distinguish the observed regimes, and tends to overestimate the occurrence of one or two regimes (usually a low-cloud regime, and sometimes a high-cloud regime). Similar classification is applied to climate change experiments to show that the model simulates shifts from low, optically thick cloud regimes towards low, optically thin ones. The cloud radiative effect of each cloud regime is relatively unaffected by the warming, emphasizing the shift in frequency of regimes. This shift happens within large-scale conditions, and is part of what has previously been described as the “thermodynamic” response of clouds.