IIT Mandi Unveils AI Powered Landslide Warning System For The Indian Himalayan Region

Scientists at IIT Mandi have unveiled a fully operational Landslide Early Warning System, or LEWS , built specifically for the Indian Himalayan Region , offering daily forecasts of slope failure risk through a web based platform right as monsoon season tightens its grip on the mountains. Picture entire valleys, previously flying blind before every downpour, suddenly receiving forecasts as routine as a weather update.

The announcement, made Wednesday, comes at a moment when the stakes could not be higher. The Indian Himalayan Region remains among the country's most landslide prone zones, and shifting climate patterns are only making things worse, triggering sharper, more frequent slope failures that regularly cost lives and destroy property across hill states.

Leading the research is Professor Dericks Praise Shukla from IIT Mandi's School of Civil and Environmental Engineering, working alongside scholars Ankit Singh and Nitesh Dhiman . Their system works by fusing terrain susceptibility mapping with live rainfall data, essentially teaching the mountains to speak before they fall silent under debris.

To build it, the team first combed through nearly 26,000 historical landslides recorded in the Geological Survey of India database, using that data to construct a detailed susceptibility map. Multiple landslide triggering factors were then layered together using ensemble machine learning models , sharpening prediction accuracy across wildly varied terrain. The crown jewel of the system is the Probability of Rainfall Induced Landslides, or P RIL, model , trained on NASA's Global Landslide Catalogue alongside seven distinct rainfall parameters drawn from IMERG satellite datasets .

What truly sets this apart from India's existing patchwork of localized warning tools is scale. While most current systems cover only small, isolated stretches, IIT Mandi's LEWS spans the entire Indian Himalayan Region , instantly making it one of the most extensive operational landslide forecasting networks the country has ever deployed.

This latest breakthrough builds on the institute's earlier work, including a separate low cost, AI powered sensor system already active across more than 60 high risk locations in Himachal Pradesh , capable of flagging ground movement up to three hours before a slope gives way, boasting over 90 percent accuracy in field trials. Together, the two systems paint a picture of a mountain range slowly being wired with intelligence, one designed not to stop the rain, but to make sure nobody is caught beneath it when the earth finally moves.