Integrated Landslide Hazard Assessment Using Machine Learning, Rainfall Threshold, and RAMMS Based Rockfall Simulation in NH-05, Himachal Pradesh, India

In Himalayas, landslides are one of the most common Natural Hazards. They cause loss of livelihood, and destroy infrastructure which brings daily life to a halt. A spike has been observed in Landslide Numbers along the NH-05 in Himachal Pradesh, India. The state depends on these National Highways for its tourists, logistics, etc., and is important in terms of national security given its proximity to the border. We have attempted to develop a Landslide Susceptibility Zonation (LSZ) Map based on twelve geo-environmental factors using the Analytical Hierarchy Process (AHP). Landslide statistics extracted from the map suggests that 33.7% (119 out of 353 landslides) of the training landslides and 35.3% (31 out of 88 landslides) of the verification landslide data points fall under "Very High" landslide likelihood zones. A risk map was produced by geospatial integration of the vulnerable zones (viz roads, settlements, agricultural land) with the susceptible zones. Additionally, a Rainfall threshold equation was also developed to assess the effect of rainfall on the initiation mechanism of landslides. This equation could also be used to build early warning systems which will take the historical landslide occurrences and the associated parameters into account. Numerical modelling of the selected sites was performed in the Rockfall module of RAMMS software to have an understanding of the scale of the affected area for mitigation and remediation. Jeori landslide is found to be more impactful by having the larger values of maximum jump height, maximum kinetic energy, and maximum velocity as 39.85 m, 2879.47 kJ, 39.01 m/s(-1) respectively.