Sea Surface Temperature Biases in CMIP5 Coupled Models and Their Implications for Regional Climate Simulations and Projections

In this paper we examine some of the larger (in amplitude, spatial extent), systematic biases in CMIP5 model simulations of sea surface temperatures (SST) and their influence on model simulations of regional climates around the globe. Emphasis is on common biases in the climatological annual cycle of SSTs in the global tropics, which tend to occur in regions having substantial coastal and equatorial upwelling. After the spatial patterns of climatological SST bias in the CMIP5 simulations have been identified (and quantified), they are used in atmospheric model experiments to examine their influence on regional climate. Two atmospheric models are used (CAM5 and ECHAM5) to examine common responses in different experiments where the models are forced with observed monthly SSTs for a 30-yr period where different climatological biases have been added. The output is compared to control runs where the observed SSTs are used. Changes in seasonal precipitation, including changes in the annual cycle, are emphasized for different regions inside and outside of the tropics. Some implications for climate change projections will be discussed, with examples given which suggest some regional climate change projections may be substantially influenced by an amplification of SST biases in simulations under increasing greenhouse gas forcing.