Landslide Susceptibility and Risk Mapping in the Tectonic Ensemble Comprising of Eastern Himalayan Zone, Northeast India and Bhutan using Logistic Regression and Random Forest Techniques
- 1 February 2024
- journal article
- research article
- Published by Geological Society of India in Journal of the Geological Society of India
- Vol. 100 (2) , 168-180
Abstract
The mountainous terrain of the East-Northeast tectonic ensemble covers an approximate area of 3,11,420 km2 comprising the Eastern Himalayan zone, Northeast India and Bhutan is prone to mass movements. The increasing trend of landslide occurrences in the last few decades sets loud alarm bells for mapping the landslide hotspot zones in terms of Landslide Susceptibility and Risk. Initially, a landslide inventory of around 9751 landslides has been prepared, of which 6826 (70%) has been randomly picked as training dataset and the remaining 2925 (30%) as the testing dataset. Thereafter, Random Forest (RF) and binary Logistic Regression (LR) based Landslide Susceptibility Zonation (LSZ) have been prepared through twelve predictor layers viz. Slope Angle, Slope Aspect, Slope Curvature, Distance to Drainage, Distance to Lineament, Landform, Surface Geology, Distance to Road, Normalized Difference Vegetation Index (NDVI), Landuse/ Landcover (LULC), Rainfall and Epicentre Proximity. Both the Landslide Susceptibility Index (LSI) maps are divided into five classes, viz. ‘low’, ‘moderate’, ‘high’, ‘very high’ and ’severe’, which are then validated statistically by drawing a comparison with the 30% testing chronological landslide inventory database. The statistical index based accuracy assessment in terms of Area Under the Curve (AUC) exhibits an LR model AUC of 0.776 and RF model AUC of 0.820. However, it has been observed that the RF model strongly correlates with the testing inventory dataset exhibiting that around 46% of the landslide-prone terrain is classified in the ‘high to severe’ zones with 41% inventory landslides occurring in these zones. Integrating the Random Forest (RF) based LSI thematic layer with the socio-economic vulnerability layers like the Number of Households and Population Density have demarcated around 20.78% of the region under ‘very high to severe’ socio-economic risk as convoluted by landslide susceptibility in the terrain. The present findings are expected to be useful in urban development and town planning with appropriate slope management and land-use planning.Keywords
This publication has 27 references indexed in Scilit:
- Landslide susceptibility mapping of Noklak Town, Nagaland, North‐east India using bivariate statistical methodGeological Journal, 2022
- Landslide detection in the Himalayas using machine learning algorithms and U-NetLandslides, 2022
- GIS-Based Landslide Susceptibility Mapping in Eastern Boundary Zone of Northeast India in Compliance with Indo-Burmese Subduction Tectonics2021
- Remote sensing GIS-based landslide susceptibility & risk modeling in Darjeeling–Sikkim Himalaya together with FEM-based slope stability analysis of the terrainNatural Hazards, 2021
- Landslide susceptibility zonation mapping using frequency ratio and fuzzy gamma operator models in part of NH-39, Manipur, IndiaNatural Hazards, 2016
- Landslide susceptibility mapping using GIS-based multi-criteria decision analysis, support vector machines, and logistic regressionLandslides, 2013
- Triggering Factors of Landslides and Determination of Rainfall Threshold: A Case Study from North East India2013
- Landslide susceptibility mapping using certainty factor, index of entropy and logistic regression models in GIS and their comparison at Mugling–Narayanghat road section in Nepal HimalayaNatural Hazards, 2012
- Landslide Hazard Zonation using Remote Sensing and GIS: a case study of Dikrong river basin, Arunachal Pradesh, IndiaEnvironmental Geology, 2007
- Assessment of shallow landslide susceptibility by means of multivariate statistical techniquesEarth Surface Processes and Landforms, 2001