Snow Cover Area Monitoring Using SAR Imagery in the Czech Republic
Souckova, Jana; Slacikova, Jana; Potuckova, Marketa
Charles University in Prague, CZECH REPUBLIC

This study discusses the feasibility of using various SAR data sources for mapping snow cover extent in the conditions of non-alpine areas with respect to different land cover types in the Czech Republic. The test-site is an area of small mountainous catchment Bystrice River and Zlatý Brook situated in North Bohemia in Ore Mountains near by the Czech-German frontier.

Melting snowpack significantly contributes to the runoff which frequently causes spring floods in the Czech Republic. Therefore, spatial information on seasonal snow cover and its characteristics are important parameters in hydrological modelling. Due to extensive cloud cover during winter period the use of optical data is limited. The advantage of SAR imagery is its independence on weather conditions. On the other hand, the presence of dense vegetation or surface roughness makes the estimation of snow cover more difficult.

For winter season 2010/2011 the use of fully polarimetric L-band ALOS PALSAR data and single polarised C-band ENVISAT ASAR data, acquired at approximately similar date as ALOS data, is examined. Because of the unavailability of L- and C-band data in the winter season 2011/12, the use of X-band SAR data from TerraSAR-X is explored. In both cases, in-situ measured data on snow cover extent, depth, snow water equivalent and temperature are available for validation.

Two full polarimetric ALOS PALSAR scenes are processed by using polarimetric decomposition (e.g. H/A/alpha) and the complex Wishart classification. Additionally, relations between snowpack characteristics and polarimetric SAR parameters (entropy H, anisotropy A, scattering mechanism angle alpha, eigenvalues and parameters derived from them) are examined. For ENVISAT ASAR images, methodology based on the change in absorption of radar signal in the snow pack due to the liquid water content is used. In case of TSX data, firstly, the interferometric approach was examined. However, even the 11-days long time baseline caused severe loss of coherence. Hence, this method did not provide satisfactory results. Currently this data is exploring by similar methodology as the ENVISAT ASAR data.