Projected frequency of low to high-intensity rainfall events over India using bias-corrected CORDEX models

Abstract

Heavy rainfall events and associated floods have emerged as one of the great threats to society that mainly manifested due the climate change. The Indian summer monsoon (ISM) contributes 80 % of annual rainfall and is characterized mainly by high-intensity rainfall events (HiREs) in the recent era. We investigated the spatiotemporal variability of HiREs from a climate change perspective by accessing the India Meteorological Department's (IMD) observed daily gridded rainfall dataset (0.25° × 0.25°) from 1961 to 2020 during the ISM season. Our observational analysis shows that the ISM total and the frequency of low- to high-intensity rainfall events have significantly decreased mostly over the central northeastern, Jammu and Kashmir, and some places in the northeastern and central parts of India. However, they have significantly increased over Gujarat, the northwestern, the Western Ghats, and the southern parts of India during the present climate period (1991–2020) compared to the past climate period (1961–1990). Furthermore, we explored the fidelity of five Coordinated Regional Climate Downscaling Experiments (CORDEX) Regional Climate Models (RCMs) in simulating the spatiotemporal variability of ISM total rainfall and the frequency of low- to high-intensity rainfall events over India during the historical (1976–2005) and future periods (2006–2100). All CORDEX RCMs overestimate the ISM total rainfall over India's heavy rainfall zones during the historical period by ∼10–30 % compared to IMD observations. To improve CORDEX RCM's skills in simulating the frequency of low- to high-intensity rainfall events, we employed a percentile-based bias correction technique. Compared to non-bias-corrected outputs from the RCMs, the quantile-bias-corrected method significantly enhanced the probability of detection rate (hit rate) in all studied models for extreme, heavy, and moderate rainfall events, excluding light rainfall events. Interestingly, the improvement is greater for extreme events, followed by heavy and moderate rainfall events. The composite hit rate of all the models shows 381 %, 146 %, and 44 % improvement for extreme, heavy, and moderate events, respectively. It is noticed that the CCCMA model performed better than the other four CORDEX models in capturing the spatial patterns of ISM total rainfall and the frequency of total extreme and heavy rainfall events over higher rainfall zones in India. Additionally, this study suggests that there will likely be no significant changes in ISM total rainfall over India in the future, but the frequency of total extreme and heavy rainfall events will most likely increase, while the frequency of moderate rainfall events will likely decrease mostly over southern parts of India in future projections.