Multi-scale Analysis of the Multi Source DEMs and Derived Topographic Parameters for Suitability of Avalanche Hazard Mapping

Abstract Topography is the most essential input in process of avalanche hazard modelling, simulation and mapping. Formation of the avalanches and their movement through mountainous terrain is governed by the topographic parameters and their scales. Digital Elevation Model (DEM) is the only source for deriving various topographic parameters spatially. Spatial resolution and elevation accuracy of the DEMs depend on the DEM sources. In the present work a multi-scale and multi resolution analysis is carried out using six different DEMs (SRTM -90m, ASTER- 30m, TOPO-30m, ALOS-12.5m, CartoSat 1-10m, and Airborne Survey DEM -1m) of the Manali-Dhundi area, India. The objective of present work is to understand the variation and sensitivity of various topographic parameters (point and local) at different cell sizes and neighbourhood sizes and also to identify the most suitable scales of different contributing topographic parameters for identification of the avalanche prone areas. High resolution airborne derived DEM elevation is compared with DGPS elevation value and high coefficient of correlation (R2 =0.97), and low RMSE (~0.97 m) and MAE (~0.04 m) has been observed. Further, various topographic variables (i.e. elevation, slope, aspect, curvature, terrain ruggedness) and avalanche simulation parameters (i.e. avalanche release area and debris extents in xyz direction) derived from DEMs were evaluated with the reference DEM at various spatial scales for avalanche sites of study area. Analysis results suggests that, ALOS and ASTER DEM produces fairly matching representation of actual terrain and debris extent with respect to reference DEM. However, TOPO DEM provide good accuracy for calculation of release area as compared to other DEMs. Slope and ruggedness derived from ADS DEM at 5m to 10m resolution provide fairly good representation of avalanche terrain. Extraction of topographic curvature was found most suitable at ≥ 30m resolution for avalanche terrain. Suitable scales for identification of avalanche release area and estimation of avalanche flow extents were found between 5 to 10m. In the present work validation of the estimated release area and flow extent was carried out with the limited point observations from the field. Detailed field investigation using geospatial approach of data collection will be taken in future for detailed and multiple profile-based validation of these results.