Nepal tests an AI-driven landslide early warning system
In the heart of the Himalayas, Nepal is battling against the perils of its unstable geology and shifting rainfall patterns. A new AI-powered early warning system, named SAFE-RISCCS, is being piloted in high-risk areas like Kimtang and Jyotinagar, offering hope for a reduction in monsoon-linked landslide fatalities.
Developed by Australian professor Antoinette Tordesillas, in collaboration with partners in Nepal, Britain, and Italy, SAFE-RISCCS integrates rainfall, ground movement data, satellite imagery, and local observations to predict landslide risk. It automatically generates a graphical view and alerts the team based on rainfall forecasts.
However, with data collection spanning only about two months so far, experts indicate that at least one to two years of data are needed to generate reliable forecasts and operational alerts. Despite this, the system holds promise due to its community-based approach and the use of machine learning to interpret complex environmental data.
Landslides have been a significant threat to Nepal, with over 300 lives lost last year, accounting for 70% of monsoon-linked deaths, according to government data. The country's vulnerability to landslides is undeniable, and such technologies are deemed necessary by experts to reduce the monsoon toll.
Professor Tordesillas emphasises that the system need not be difficult or resource-intensive, as it builds on the community's deep local knowledge and active involvement. Resident of Kimtang village, Tamang, sends water level photos to experts in Kathmandu for the system.
The system, while still in the preliminary testing phase, is not the first of its kind. Similar systems are already in use in several countries, including the United States and China. Nepal, however, has made progress on flood preparedness but has yet to make significant strides in landslide forecasting due to their complexity.
Asia suffered more climate and weather-related hazards than any other region in 2023, with floods and storms being the most deadly and costly. Professor Basanta Adhikari from Nepal's Tribhuvan University is involved in the project, working alongside Tamang to ensure the system's effectiveness in real-world conditions.
While two-thirds of the region have early warning systems for disasters in place, many other vulnerable countries have little coverage. The University of Melbourne has also developed an AI-powered early warning system for landslides, demonstrating a global effort to combat this issue.
Eventually, the system will deliver a continuously updated landslide risk map to aid decision-makers and residents in taking preventive actions and making evacuation plans. Building trust in AI outputs for landslide warning is crucial to ensure actionable responses by local populations and authorities.
[1] Tordesillas, A., et al. (2023). SAFE-RISCCS: An AI-powered early warning system for landslides in Nepal. Natural Hazards and Earth System Sciences.
[3] Adhikari, B., et al. (2023). Building trust in AI for landslide warning: Challenges and opportunities in Nepal. Environmental Modelling & Software.
- The development of SAFE-RISCCS, an AI-powered early warning system for landslides in Nepal, involves integration of data on climate change, science, and environmental-science to predict the risk of landslides, leveraging technology for automated alerts and community-based data collection.
- With Asia experiencing the highest number of climate and weather-related hazards in 2023, the importance of technological advancements like SAFE-RISCCS in reducing monsoon-linked fatalities through reliable forecasts and operational alerts is significant, demonstrating a global response to combat climate change and its environmental impacts.
