Increasing uncertainty in large flood and drought projections over central India

Abstract Extreme precipitation events over the Indian subcontinent are becoming more frequent, posing increasing risks to society and ecosystems. This study investigates large floods and droughts, defined as the 90th percentile of the percentage deviation of seasonal rainfall anomaly from long-term mean using historical all-forcing runs and future climate projections from the Coupled Model Intercomparison Project Phase 6 (CMIP6) models . Although individual models showing biases, the multi-model ensemble reproduces the historical frequency and return periods reasonably well, as biases in the mean tend to cancel out. Future scenarios, however, show contrasting projections with substantial uncertainty across models. We find that large floods are projected to increase in frequency and exhibit shorter return periods, indicating a shift toward more frequent and intense wet seasons. In contrast, large droughts show a weaker and more uncertain response. Sub-seasonal synoptic activity, particularly low-pressure systems (LPS), plays a key role in shaping these extremes. During flood years, LPS are more frequent and show coherent northwestward propagation, which is largely absent during drought years. This pattern remains largely consistent in future scenarios, although the overall number of LPS tracks shows only a marginal decline. ENSO’s influence on Indian summer monsoon (ISM) extremes also changes under warming: the clear historical relationship La Niña with floods and El Niño with droughts becomes less predictable. Under high-emission scenarios, floods increasingly occur during El Niño years, suggesting a shift in ENSO-ISM interactions. These results underscore the growing importance of understanding synoptic variability and ENSO changes in a warming climate to improve flood and drought preparedness in South Asia.