Near-real-time derivation of snow cover maps for hydrological modeling using operational remote sensing data

A fully automatic operational algorithm to derive the snow covered area from optical data will be presented. Snow-cover maps are processed and provided to the user within a few minutes after online reception of NOAA-AVHRR data. Processing consists of calibration, geometric correction and classification, including automatic cloud detection. To get maximum information and in order to reduce the cloud cover, the daily pathes from all available NOAA-AVHRR satellites were analyzed synergistically. For operational applications high geometric accuracy is necessary. This is achieved by integrating an automatic iterative procedure using satellite orbit information and parametric geocoding including a DEM based terrain correction. Further innovations of the algorithm are the threshold dependency on landuse and the classification of areas with expected melting/temperated snow. For operational application, the processing and distribution chain includes a post-classification, snow line detection and quality assessment of the product before delivery. Since late 2001 the automatic generation of snow maps for Southwest Germany is provided as an operational service to flood forecast centers based on NOAA-AVHRR data. Results for the winter seasons 1998/1999 and 2001/2002 show the applicability of the selected methods in the investigated highland area (accuracy: 95%). During the last two winters, the operational service of snow cover maps provided more than 250 products, containing spatial information for the integration in operational water balance modeling.