A dominant role for SST pattern change in both fast and long-term tropical rainfall responses to CO2 forcing

The regional tropical rainfall response to CO2 forcing is very important, due to the potential for large human and environmental impacts, but highly uncertain. Idealised climate experiments, such as abruptly increasing CO2 in a coupled climate model, provide a useful and traceable framework for investigating the mechanisms behind rainfall projections, and understanding both robust and uncertain regional changes. Bony et al. (2013, Nat. Geosci.) demonstrate the power of these idealised experiments: the first year response of dynamical precipitation change over the tropical oceans to an abrupt CO2 increase is similar to the century-scale response. They suggest that a large part of the regional precipitation change is a direct atmospheric response to CO2 that may be explored in very short atmosphere-only forecast simulations. While this response may be partly independent of global mean warming, we show that in fact the pattern of surface warming is crucial to the pattern of tropical rainfall response. Even in year 1 of the coupled experiment, tropical sea surface temperature (SST) patterns have emerged with a similar shape and approximately half the magnitude of the century-scale SST pattern change, and can account for most of the pattern of dynamical rainfall change over the tropical oceans (Fig. 1b-e). In contrast, in an experiment where CO2 is increased but SSTs fixed (Fig. 1a), the pattern of rainfall change does not resemble the century-scale coupled response. This confirms the `warmer-get-wetter' hypothesis, which emphasises the role of SST pattern change in driving regional tropical precipitation change in climate models. Future regional rainfall changes will depend on the SST response to both direct CO2 forcing and mean surface warming, and thus should be studied primarily in coupled ocean-atmosphere models. Fig. 1: CMIP5 multi-model mean precipitation and SST changes for abrupt4xCO2 a) fixed SST and b)-e) coupled experiments.