Assessing CESM2 clouds and their response to climate change using cloud regimes

To evaluate the simulated clouds in the present climate and characterize their response with climate warming, a clustering approach is applied to three independent satellite cloud products and a set of coupled climate simulations. Using k -means clustering with a Wasserstein distance cost function, a set of typical cloud con fi gura- tions is derived for the satellite cloud products.

Using satellite simulator output, the model clouds are classi fi ed into the observed cloud regimes in both current and future climates. The model qualitatively reproduces the observed cloud con fi gurations in the historical simulation using the same time period as the satellite observations, but it struggles to capture the observed heterogeneity of clouds which leads to an overestimation of the frequency of a few preferred cloud regimes.