Snow Cover Mapping over the mountainous areas in Europe with MSG-Seviri

Snow is one of the main water resources, therefore monitoring and estimating the snow water equivalent play important role in predicting discharges during melting seasons. Distributed snow models may require the following spatially distributed parameters: snow-covered area, grain size, albedo, snow water equivalent, snow temperature profile and meteorological parameters, including solar radiation. In mid-latitudes, snow can be seen continuously in the mountainous terrains. The difficulty in accessibility to perform the measurements at the remote sites makes the use of satellite images and/or aerial photographs in monitoring and estimating the snow parameters more valuable. The use of snow products retrieved from satellite images in hydrological applications and to observe the impact of the products are the key issues in Hydrology Satellite Application Facilities (H-SAF) project, which is financially supported by EUMETSAT. Turkey is a part of the H-SAF project, both in developing satellite-derived snow products (snow recognition, effective snow cover and snow water equivalent) for mountainous areas, cal/val of satellite-derived snow products with ground observations and impact studies with hydrological modeling in the mountainous terrain of Europe. In this paper, the early findings of the H-SAF project regarding the snow recognition product for mountainous regions are evaluated.

A pixel value based model has been developed for snow recognition over mountainous areas of Europe. The method is mainly using the satellite images acquired every 15 minutes from geostationary satellite; Meteosat Second Generations (MSG) instrument Spinning Enhanced Visible and Infra-Red Imager (SEVIRI). The algorithm for snow recognition uses the channel-1 (0.6mm), channel-3 (1.6mm), channel-4 (3.9mm), and channel-9 (10.8mm). Discrimination of snow and clouds are implemented using channel difference algorithms for each scene and accumulated statistics of cloud and snow classification are used in determining the daily snow product. In this way, it is possible to distinguish clouds from snow by considering their different spectral characteristics. Current algorithm has been used for generating daily snow maps since September 2006. The routine generation includes a certain amount of online re-calibration/validation to monitor product quality stability and continuously improve error structure characterization. Preliminary validation was performed on several images and the test will continue for the following months where more snow will be possible to classify. From the preliminary results the proposed algorithm showed a quite good performance in classifying the snow, however some type of the clouds were classified as snow. In order to make a better discrimination between snow and cloud, the tuning of the algorithm will continue by determining the statistics of the reflectance values of each snow free pixels for MSG SEVIRI bands 1 and 3.