Himalayan Geology, Vol. 45 (1), 2024, pp. 39-57, Printed in India
Bivariate statistical models for Landslide susceptibility mapping at local scale in the Aizawl municipal area, Mizoram, India
Z.D. LALTANPUIA1,3, TAPAS R. MARTHA2*, K. SRINIVASA RAO3, KIRTI KHANNA2
1Mizoram Remote Sensing Application Centre (MIRSAC), Aizawl, Mizoram, India
2National Remote Sensing Centre (NRSC), ISRO, Hyderabad, Telangana, India
3Department of Geology, Mizoram University, Aizawl, Mizoram, India
*Email (Corresponding author): tapas_martha@nrsc.gov.in
Abstract: Landslide is the most frequent and prevalent form of natural hazards experienced by the state of Mizoram in India. Due to climatic extremes and anthropogenic disturbances, landslide susceptibility maps need periodic updation to meet present day requirements. In this study, landslide susceptibility mapping of Aizawl Municipal Area has been attempted using three bivariate statistical models, namely, Information Value (IV), Frequency Ratio (FR) and Weight of Evidence (WofE) models. Based on field survey, historical records and from Google Earth image interpretation, a total of 381 landslides have been mapped within the study area which were split into two parts i.e., 70% of the landslides were used for model calibration while the remaining 30% were used for validation. Twelve landslide conditioning parameters including Slope, Aspect, Elevation, Lithology, Drainage Density, Lineament Density, Normalized Difference Vegetation Index (NDVI), Land Use/Land Cover (LULC), Plan Curvature, Profile Curvature, Stream Power Index (SPI) and Topographical Wetness Index (TWI) were prepared from a 10m resolution DEM of Cartosat-1 data, Sentinel-2 data and Google Earth image. Landslide susceptibility maps were prepared showing three susceptible classes ranging from low to high susceptibility. The resulting maps have been validated using the area under the receiver operating characteristic curve which showed success rate for IV, FR and WofE models as 83.8%, 83.7% and 83.9% respectively. Similarly, the prediction rate for the three models are found to be 83.7%, 83.5% and 83.7%. The landslide susceptibility map was compared with the previous map prepared by GSI. Results show many new areas are now under higher hazard category in the Aizawl municipal area.
Keywords: Aizawl Municipal, Bivariate Statistical Model, Landslide Susceptibility, ROC
