Shillong, February 04 : In a major research initiative, the Department of Information Technology, North-Eastern Hill University, has developed an AI based Landslide Susceptibility Map (LSM) of Meghalaya using an ensemble Machine Learning (ML) framework. The framework combines ten different machine learning models to enhance accuracy, robustness, and reliability of the LSM.
Meghalaya’s complex geological structure, frequent seismic activity, and intense rainfall during monsoon, makes the state highly susceptible to landslides, resulting in loss of life and properties every year. According to experts, the impact of landslides can be reduced by identifying and regularly monitoring the vulnerable areas.
The research was carried out by Dr. K. Amitab and his team with financial support from the Science and Engineering Research Board (SERB), Department of Science and Technology (DST), Government of India. Historical landslide inventory data obtained from the Geological Survey of India (GSI) and the North Eastern Space Applications Centre (NESAC) were used to train and evaluate the model. The framework achieved an accuracy exceeding 90 percent, demonstrating its effectiveness in predicting landslide-prone zones.
The generated LSM classifies landslide susceptibility of Meghalaya into five risk categories: very high, high, moderate, low, and very low. According to the map, approximately 7% of the state falls under very high-risk category, while 6%, 8%, 19%, and 60% fall under the high, moderate, low, and very low categories, respectively. The East Khasi Hills district is the most vulnerable region, with approximately 730 km2 falling under the very high risk category. Other vulnerable districts include Ri Bhoi, Eastern West Khasi Hills, West Khasi Hills, Southwest Khasi Hills, and East Jaintia Hills and West Jaintia Hills.
An analysis of landslide causative factors, revealed that proximity to roads is the most influential factor in landslide occurrence. This is attributed to slope destabilization during road construction, alteration of natural drainage patterns, and disturbance caused by vehicle movements. Other influential causative factors include Slope degree, NDVI, soil type, elevation, road density, and lithology.
The LSM can serve as a valuable tool for disaster management agencies in prioritizing resource allocation to high-risk regions and guiding proactive planning to mitigation the impact of landslide. The research marks a significant advancement in improving public safety and reducing landslide-related hazards in Meghalaya.
