Impacts of SST Warming in tropical Indian Ocean on CMIP5 model-projected summer rainfall changes over Central Asia

Based on the historical and RCP8.5 experiments from 25 Coupled Model Intercomparison Project phase 5 (CMIP5) models, the impacts of sea surface temperature (SST) warming in the tropical Indian Ocean (IO) on the projected change in summer rainfall over Central Asia (CA) are investigated. The analysis is designed to answer three questions: (1) Can CMIP5 models reproduce the observed influence of the IO sea surface temperatures (SSTs) on the CA rainfall variations and the associated dynamical processes? (2) How well do the models agree on their projected rainfall changes over CA under warmed climate? (3) How much of the uncertainty in such rainfall projections is due to different impacts of IO SSTs in these models? The historical experiments show that in most models summer rainfall over CA are positively correlated to the SSTs in the IO. Furthermore, for models with higher rainfall-SSTs correlations, the dynamical processes accountable for such impacts are much closer to what have been revealed in observational data: warmer SSTs tend to favor the development of anti-cyclonic circulation patterns at low troposphere over north and northwest of the Arabian Sea and the Bay of Bengal. These anomalous circulation patterns correspond to significantly enhanced southerly flow which carries warm and moisture air mass from the IO region up to the northeast. At the same time, there is a cyclonic flow over the central and eastern part of the CA which further brings the tropical moisture into the CA and provides essential moist conditions for its rainfall generation. In the second half of twenty-first century, although all the 25 models simulate warmed SSTs, significant uncertainty exists in their projected rainfall changes over CA: half of them suggest summer rainfall increases, but the other half project rainfall decreases. However, when we select seven models out of the 25 based on their skills in capturing the dynamical processes as observed, then the model projected changes are much closer. Five out of the seven models predicted more rainfall over CA. Such a result is helpful for allowing us to attribute part of the observed upward rainfall trend in the CA region in the last several decades to the IO SST warming.