Machine Learning-Based Assessment of High-Impact Low Likelihood Precipitation Events in North India

Climate change has led to notable discrepancies in the frequency and intensity of precipitation extremes in the Himalayan region of North India, posing significant challenges for water management, agriculture, and disaster preparedness. Historical data indicate a 5–10% increase in extreme precipitation events over recent decades, resulting in severe floods, landslides, and crop failures that have heavily impacted local communities and the regional economy. This study applies a machine learning framework to assess and project precipitation extremes in North India using IMDAA reanalysis data. Supervised machine learning models—Random Forest (RF) and Support Vector Machine (SVM)—were employed for spatial classification across diverse topographies. The RF model achieved higher accuracy (86%) in low-elevation, less complex terrains, while the SVM model performed better (87%) in high-altitude, complex regions. Additionally, the RF model demonstrated superior probabilistic prediction with a lower Brier score of 0.07.  The varying model performance reflects the influence of topography, atmospheric dynamics, and data resolution. RF effectively captures non-linear relationships in simpler terrains, whereas SVM’s ability to define optimal hyperplanes enhances its performance in mountainous areas. Our analysis highlights significant spatial heterogeneity in precipitation extremes, revealing intensifying extreme events and identifying hotspots of substantial change across Northern India. These insights are crucial for informing targeted adaptation strategies in water resource management, flood risk assessment, and climate resilience planning. By pinpointing vulnerable regions and potential future hotspots, this study can support policy-making, infrastructure development, and community preparedness, ultimately reducing economic losses and safeguarding lives. Integrating machine learning with IMDAA reanalysis data proves valuable in understanding extreme precipitation events. Future research will incorporate CMIP6 climate projections to refine predictions and offer a more comprehensive evaluation of future climate scenarios, thereby enhancing preparedness for extreme weather events amid ongoing climate change.