|Title||A robust but spurious pattern of climate change in model projections over the tropical Indian Ocean|
|Publication Type||Journal Article|
|Year of Publication||2016|
|Authors||Li G., Xie SP, Du Y.|
|Journal||Journal of Climate|
|Type of Article||Article|
|Keywords||CMIP5 models; coupled; dipole mode; el-nino; equatorial; gcm experiments; model; multimodel ensemble; pacific; regional patterns; sea-surface temperature; seasonal cycle|
Climate models consistently project reduced surface warming over the eastern equatorial Indian Ocean (IO) under increased greenhouse gas (GHG) forcing. This IO dipole (IOD)-like warming pattern, regarded as robust based on consistency among models by the new Intergovernmental Panel on Climate Change (IPCC) report, results in a large increase in the frequency of extreme positive IOD (pIOD) events, elevating the risk of climate and weather disasters in the future over IO rim countries. These projections, however, do not consider large model biases in both the mean state and interannual IOD variance. In particular, a "present-future relationship" is identified between the historical simulations and representative concentration pathway (RCP) 8.5 experiments from phase 5 of the Coupled Model Intercomparison Project (CMIP5) multimodel ensemble: models with an excessive IOD amplitude bias tend to project a strong IOD-like warming pattern in the mean and a large increase in extreme pIOD occurrences under increased GHG forcing. This relationship links the present simulation errors to future climate projections, and is also consistent with our understanding of Bjerknes ocean-atmosphere feedback. This study calibrates regional climate projections by using this present-future relationship and observed IOD amplitude. The results show that the projected IOD-like pattern of mean changes and frequency increase of extreme pIOD events are largely artifacts of model errors and unlikely to emerge in the future. These results illustrate that a robust projection may still be biased and it is important to consider the model bias effect.