Causes of large climate model spread in equatorial Pacific cloud feedback

Climate models remain the best tools for predicting the impact of climate change on quantities relevant to human activity, such as precipitation, surface temperature and occurrence of severe weather events. Since many of these changes scale with the models equilibrium climate sensitivity, it is crucial to understand the differences in climate sensitivity between the models, which are primarily driven by inter-model differences in cloud feedbacks. Inter-model differences in cloud feedbacks are largest in the equatorial Pacific. Focussing on the area from 10°S - 10°N, and 160°E – 270°E, we find an inter-model standard deviation in cloud feedback of ~1.36 W m-2 K-1. Using appropriate weighting to account for the area of this region, this equates to a contribution to the global mean cloud feedback uncertainty of ~ 0.07 W m-2 K-1, which represents approximately 20% of the inter-model spread in global mean cloud feedback. Local differences in cloud feedback between models in this region are even larger and may have implications for regional circulation and precipitation changes. This region is also notable as an exception to the high correlation in cloud feedbacks between coupled and atmosphere-only models. In this presentation we will describe analysis of the causes of the inter-model spread in cloud feedbacks in this region. We shall demonstrate that the spread in domain-mean feedback in this region is due to inter-model differences in both dynamic and thermodynamic cloud feedbacks and show how this relates to changes in the properties of different cloud types amongst different models. We will also describe the use of empirical orthogonal function analysis to identify consistent cloud feedback patterns in this region across the ensemble of models and explain the causes of these patterns.