Interannual relationships between Indian Summer Monsoon and Indo-Pacific coupled modes of variability during recent decades

International audienceVarious SST indices in the Indo-Pacific region have been proposed in the literature in light of a long-range seasonal forecasting of the Indian Summer Monsoon (ISM). However, the dynamics associated with these different indices have never been compared in detail. To this end, the present work re-examines the variabilities of ISM rainfall, onset and withdrawal dates at interannual timescales and explores their relationships with El Niño-Southern Oscillation (ENSO) and various modes of coupled variability in the Indian Ocean. Based on recent findings in the literature, five SST indices are considered here: Niño3.4 SST index in December-January both preceding [Nino(-1)] and following the ISM [Nino(0)], South East Indian Ocean (SEIO) SST in February-March, the Indian Ocean Basin (IOB) mode in April-May and, finally, the Indian Ocean Dipole (IOD) averaged from September to November, also, both preceding [IOD(-1)] and following the ISM [IOD(0)]. The respective merits and associated dynamics of the selected indices are compared through various correlation and regression analyses. Our first result is a deceptive one: the statistical relationships with the ISM rainfall at the continental and seasonal scales are modest and only barely significant, particularly for the IOD, IOB and Nino(-1) indices. However, a detailed analysis shows that statistical relationships with the ISM rainfall time series are statistically biased as the ISM rainfall seems to be shaped by much intraseasonal variability, linked in particular to the timing of the onset and withdrawal of the ISM. Surprisingly, analysis within the ISM season shows that Nino(-1), IOB and SEIO indices give rise to prospects of comparatively higher ISM previsibility for both the ISM onset and the amount of rainfall during the second half of the ISM season. The IOD seems to play only a secondary role. Moreover, our work shows that these indices are associated with distinct processes occurring within the Indian Ocean from late boreal winter or early spring onwards. The regression analyses also illustrate that these (local) mechanisms are dynamically and remotely linked to different phases of ENSO in the equatorial Pacific, a result which may have useful implications in terms of forecasting strategies since the choice of the better indices then hinges on the concurrent phasing of the ENSO cycle