Snow detection (snow mask) by VIS/IR radiometry

Contours of snow-covered areas, presented as mask-like maps (binary image). VIS/IR images from both LEO and GEO are used. The product may be processed in different ways and have different quality depending on the surface being flat, forested or mountainous.

  • Coverage: The H-SAF area [25-75°N lat, 25°W-45°E long]
  • Cycle: Daily
  • Resolution: 1 to 5 km, depending on the instrument providing the retained pixel (best for MODIS, worst for SEVIRI)
  • Accuracy: POD 95 %, FAR 10 % - Depending on geographical situation (flat/forested areas, mountainous regions)
  • Timeliness: Fixed time of the day, product updated to account for data available until 1 h before delivery
  • Dissemination: By dedicated lines to centres connected by GTS - By EUMETCast to most other users, especially scientific
  • Formats: Values in fixed grid points of the Meteosat projection (GEO satellites) or fixed latitude-longitude grid (WGS 84) representing the resolution of the used satellite (polar orbiting satellites). Also JPEG or similar for quick-look.

Short description of the basic principles for product generation

The product is an output of image classification processing. The snow signature is recognised as differential brightness in more short-wave channels, intended to discriminate snow from no-snowed land and snow from clouds. Both radiometric signatures are used (specifically, the 1.6 micron channel as compared with others), and time-persistency (for cloud filtering by the "minimum brightness" technique applied over a sequence of images). The Meteosat/SEVIRI contribution is mostly for southern Europe (including mountainous regions) and minimum brightness technique application. For mountainous regions multispectral threshold technique implemented on VIS and IR satellite reflectance values is used in order to get maximum daily snow coverages.