Validation Of SAR Derived Wet Snow Maps At Glaciers On Jan Mayen, Norway
Malnes, Eirik1; Rollstad Denby, Cecilie2
1Northern Research Institute (Norut), NORWAY; 2Dept. Math Sciences and Technology, Univ. of Life Sciences, NORWAY

C-band synthetic aperture radar is sensitive to wet snow due to absorption from liquid water in the surface layer. In this study we have investigated a dense time series consisting of 1200 SAR images from Envisat ASAR for the period 2005-2012 and Radarsat-2 from 2012-2013 from the volcanic island Jan Mayen in the North Atlantic sea (71†N, 8°30'W) . All images are geocoded on a common grid in 75 m spatial resolution using a digital elevation model. On the island twenty glacier flow down in all direction from the central Berenberg volcanic crater in the elevation range of 120-2200 m a.s.l. Wet snow has been classified in all images using the common multi-temporal method suggested by Nagler and Rott (2000) where we classify wet snow when the difference in backscatter between the current SAR image and a reference image is less than -3dB. Averaged reference images are calculated for each imaging geometry based on selecting images from periods with dry snow or bare ground. The SAR time series will be compared with simultaneous air temperature measurements from the meteorological station at the island and several automatic weather stations (AWS) situated at several altitudes. The air temperature data are used to validate the accuracy of the wet snow detections with SAR. The sensitivity for wet snow will be characterized by comparing the backscatter change with and the threshold for wet snow detection (usually -3dB) with thresholds for snow melt based on air temperatures.