MANDI: With climate change fuelling a rise in extreme weather events and slope failures, IIT Mandi researchers have developed an operational landslide early warning system for the Indian Himalayan Region, one of the country’s most landslide-prone belts.The Landslide Early Warning System, or LEWS, has been developed by a team led by Prof Dericks Praise Shukla of the School of Civil and Environmental Engineering, IIT Mandi, along with research scholars Ankit Singh and Nitesh Dhiman.The system uses rainfall data, terrain susceptibility and machine-learning models to forecast the probability of rainfall-induced landslides. It issues daily warnings through a web-based platform, giving authorities and disaster management agencies advanced information on high-risk locations.“At the very onset of the monsoon, our Landslide Early Warning System provides daily landslide forecasts through a web-based application. The system is designed to help identify high-risk areas in advance, enabling authorities and communities to undertake timely evacuation and disaster preparedness measures,” said Prof Shukla.He said satellite-based early warning systems are among the most effective investments in disaster risk reduction as they turn scientific data into timely, actionable decisions. A region-wide landslide forecasting platform, he added, can strengthen preparedness, speed up response and improve coordination among disaster management agencies, especially during the monsoon when landslide risk peaks.Unlike several landslide warning systems in India that are limited by geography, IIT Mandi’s LEWS covers the Indian Himalayan Region, making it one of the most extensive systems of its kind in the country.The research team developed the platform through a multi-stage process. First, nearly 26,000 landslides from the Geological Survey of India database were used to prepare a landslide susceptibility map. Multiple landslide-triggering factors were then processed through ensemble machine-learning models.The team next built the P-RIL, or Probability of Rainfall-Induced Landslides, model using data from the NASA Global Landslide Catalogue and seven rainfall parameters drawn from IMERG satellite datasets. Since rainfall conditions change continuously, the model uses rainfall data from the previous 15 days, making it dynamic rather than static.The final daily forecast is generated by combining the static susceptibility map with the dynamic P-RIL model through a probabilistic approach. To make the output easier for users to interpret, the forecast is displayed in percentile-based risk categories.For wider access, the IIT Mandi team has created a Google Earth Engine-based web portal where users can view the landslide forecast for the current day and the previous three days. The portal also allows users to download PDF bulletins and receive WhatsApp alerts for selected locations.Researchers said the system could significantly improve disaster preparedness in the Himalayan region by issuing timely, location-specific warnings and helping reduce loss of life, damage to property and economic disruption.
