Exploring evolutionary patterns in the teleconnections between Indian summer monsoon rainfall and Indian Ocean dipole over decades

The study employs a hybrid Employing a hybrid ensemble empirical mode decomposition-temporal convolutional network (EEMD-TCN) machine learning approach to predict the IOD index and sea surface temperatures for various time spans (3, 6, 9, and 12 months ahead), and explores changes in the relationship between IOD and Indian summer monsoon rainfall (ISMR) across India from 1960 to 2020, considering risk and uncertainty. The methodology involves decomposing the IOD index and SST into consistent subcomponents using EEMD, followed by TCN modeling to forecast these subcomponents, resulting in predictions affected by risk and uncertainty. The IOD index’s forecast precision surpasses that of SSTs, attributed to the higher number of high-frequency elements in the SST data, posing a greater challenge in prediction risk and uncertainty. The research also examines anomalies in rainfall, SST, and low-level wind circulation, highlighting the impact of IOD events. ISMR across India shows sensitivity to IOD events, with positive IOD (pIOD) events linked to increased rainfall and negative IOD (nIOD) events to decreased rainfall, except during the initial phase of the Indian summer monsoon, emphasizing the role of risk and uncertainty. Variations in SST, wind circulation, and moisture transport mechanisms in the Indian Ocean lead to notable precipitation changes during different IOD stages, especially from 1991 to 2020. Notably, a recent rise in the frequency of low-level equatorial jets (LEJs) in the Indian Ocean equatorial region and the Arabian Sea during pIOD events is observed, in contrast to the earlier decades of 1960–1990.