Simulation of the Indian Ocean Dipole: A relevant criterion for selecting models for climate projections

[1] A multi-model average shows that 21st century warming over the eastern Indian Ocean (IO) is slower than that to the west, but with strong inter-model variations. Is the simulation of the Indian Ocean Dipole (IOD) relevant to these variations? We demonstrate that inter-model variations of this future warming are consistent with how well models simulate historical IOD properties; models with a stronger IOD amplitude systematically produce a slower eastern IO warming rate with greater future rainfall changes in IOD-affected regions. These models also produce a stronger Bjerknes-like positive feedback, involving sea surface temperatures (SSTs), winds and a shoaling thermocline in the eastern IO. As warming proceeds, models with a stronger positive feedback induce a greater response to warming-induced changes such as easterly trends associated with the Walker circulation, generating a smaller warming in the eastern IO. Simulating the present-day IOD properties is, therefore, a relevant criterion for selecting models for climate projections.