Observed and Modeled Cloud Responses to Climate Variability

Abstract

Clouds play a significant role in the Earth’s climate, yet cloud feedbacks remain one of the largest sources of inter-model spread in climate predictions. Studying how clouds respond to internal modes of climate variability can improve our understanding of how cloud varies on monthly to inter-annual time scales, improve our understanding of how clouds and cloud feedbacks might respond in a changing climate, and provide a validation metric for climate models. This study uses several satellite cloud datasets to identify interactions between cloud occurrence and various modes of climate variability in the historical record using a combination of linear regression and cluster analysis, including an in-depth examination of interactions between marine stratus cloud and wintertime sea surface temperature reemergence. These results are then used to evaluate cloud occurrence and monthly to annual cloud variability in historical simulations from several climate models. Implications for climate modeling are discussed. Finally, a new technique is developed to cluster unique meteorological and cloud regimes using a deep convolutional neural network.