Day 4

Detailed paper information

Back to list

Paper title Towards avalanche activity monitoring using Sentinel-1 data
Authors
  1. Anna Karas CNRM Centre National de Recherches Météorologiques - CNRS Météo France UMR3589 Speaker
  2. Fatima Karbou Center of the Snow Study, Météo-France
  3. Sophie Giffard-Roisin ISTERRE, University Grenoble Alpes, University Savoie Mont Blanc, CNRS, IRD, IFSTTAR
  4. Philippe Durand Centre National d’Etudes Spatiales (CNES), Toulouse, France
  5. Nicolas Eckert UR ETNA, INRAE, University Grenoble Alpes
Form of presentation Poster
Topics
  • D1. Managing Risks
    • D1.01 Satellite EO for Geohazard Risks
Abstract text This work relies on a novel method developed to automatically detect areas of snow avalanche debris using a color space segmentation technique applied to Synthetic Aperture Radar (SAR) image time series of Sentinel-1. The relevance of the detection was evaluated with the help of an independent database (using high resolution SPOT image). Results of detection will be presented according to the direction of the orbit and the characteristics of the terrain (slope, altitude, orientation). The basic idea behind the detection is to identify high localised radar backscatters due the presence of snow avalanche debris compared to the surrounding snow by comparing winter images with respect to reference images. The relative importance of reference images have been studied by using well selected individual or mean summer images. The method was found successful to detect almost 66 % of the avalanche events of the SPOT database, by combining the ascending and descending orbits. Best detection results are obtained with individual reference dates chosen in autumn with 72 % of verified avalanche events using the ascending orbit. We also tested a false detection filtering using a Random Forest classification model.