Deep Learn Hub

Disaster Management

Development of a novel real-time warning system for landslide prediction in Sri Lanka using Machine learning techniques.

Due to several reasons, Sri Lanka experiences frequent landslides which have become a major threat that accounts for a significant loss of life and property. Therefore, landslide identification has become critical. There are other problems in the field of landslide detection such as, lack of qualified systems made with multiple landslide detection sources of data (combination of techniques) to predict landslide potential, lack of use of highly predictive machine learning based models to accurately delineate landslide hazard areas, heavy rainfall triggers mass movement in Sri Lanka and some landslide detection methods which are applicable to one region may not be applicable to another.

This project basically addresses landslide issues in Sri Lanka through surveillance and early detection through multiple paths approaches to mitigate human and environmental hazards. The aim of the research we propose is to look at the correlation between different pictures and comes up with a system which uses the existing data and information and couple that with rainfall and other information to give a real-time warning system of landslide prediction, triggering in Sri Lanka using Machine learning techniques with the following objectives.

  • To define the parameters influencing the characters of landslides and the modelling of rainfall run-off in Sri Lanka.
  • To analyze various types of landslide related data in a Geographic Information System (GIS), which will be obtained using Remote Sensing data including LiDAR and Drone/Satellite images.
  • To establish AI based model for landslide susceptibility zonation mapping.
  • To establish rainfall intensity-duration models with take in of Remote Sensing and field data from GIS with machine learning and AI models of landslides to identify the rainfall threshold.
  • To install IoT based sensors to monitor the soil movement at the critical risk areas.
  • To provide landslide hazard map and long-term risk maps.