AI mapping flags landslide hotspots across Meghalaya

An artificial intelligence–based study has identified the most landslide-prone areas of Meghalaya, offering authorities a new tool to plan prevention and emergency response in a state hit by landslides every monsoon.

The Department of Information Technology at North-Eastern Hill University has developed a Landslide Susceptibility Map using an ensemble machine-learning framework that combines ten different models. Researchers say the approach improves prediction accuracy and reliability compared to single-model assessments.

Meghalaya’s terrain, marked by fragile geology, seismic activity and heavy rainfall, makes it especially vulnerable to slope failures that cause loss of life and damage to infrastructure each year. Scientists involved in the study stress that early identification and continuous monitoring of high-risk zones are key to reducing these impacts.

The research team, led by Dr K Amitab and funded by the Science and Engineering Research Board under the Department of Science and Technology, trained the model using historical landslide data from the Geological Survey of India and the North Eastern Space Applications Centre. The system recorded an accuracy of over 90 per cent in predicting landslide-susceptible zones.

The map categorises the state into five risk levels — very high, high, moderate, low and very low. About 7 per cent of Meghalaya has been classified as very high risk, with another 6 per cent marked high risk. East Khasi Hills emerged as the most vulnerable district, with nearly 730 square kilometres falling in the highest risk category. Other districts identified as vulnerable include Ri Bhoi, Eastern West Khasi Hills, West Khasi Hills, Southwest Khasi Hills, East Jaintia Hills and West Jaintia Hills.

The study also found that proximity to roads is the single most influential factor triggering landslides, largely due to slope cutting, disrupted drainage and construction activity. Other contributing factors include slope angle, vegetation cover, soil type, elevation, road density and underlying rock formations.

Researchers say the map can help disaster management agencies prioritise high-risk areas, guide infrastructure planning and strengthen mitigation strategies, potentially reducing landslide-related losses across the state.