Impact of atmospheric response of ENSO and IOD on boreal summer rainfall variability over the Indian Ocean in coupled models

Most dynamical models under-predict Indian summer monsoon rainfall (ISMR) variability despite showing greater dependency on external forcings in comparison with observations. In this study, we first highlight the observed patterns of externally forced rainfall variability over the Indian Ocean during boreal summer and then investigate the mechanisms involved in reduced variability in May initialized reforecasts from a multi-model dataset. The principal component analysis reveals a quad-polar structure in the first two modes of rainfall variability. It is shown that the first mode is driven by ENSO and IOD, while the second mode is independent of ENSO and is mainly driven by IOD. The differential response of ISMR to the ENSO and IOD is related to the difference in the phase relationship of the pole nearest to the Indian landmass with the rest of the poles of the quad-polar EOFs. This difference is found to be associated with the distinct response of winds over India to the ENSO and IOD forcings despite having similar zonal and meridional structures of modes of rainfall variability over the oceanic regions. The models capture the mode-1 spatial structure and temporal variations reasonably well due to high ENSO skill and realistic simulation of ENSO-rainfall teleconnections patterns, except for the Indian monsoon. The reduced skill in capturing the mode-2 structure is found to be related to low skill in simulating atmospheric and oceanic indices of IOD variability. The analysis shows that the reduced rainfall variability in models is due to the subdued response of the atmosphere to the climate modes. In comparison with observations, ENSO-mediated changes on upper tropospheric velocity potential and the meridional thermal gradients are weaker in models. This results in reduced ENSO-forced inter-annual variations of monsoon circulation and Indian rainfall. Southeast equatorial Indian Ocean (SEEIO) emerges as a key area of model deficiency where southeasterlies associated with both ENSO and IOD are underrepresented in the models. By highlighting the pathways associated with reduced rainfall variability, such studies could help in identifying the teleconnections which need to be improved in climate models.