Metsamaki, S; Mattila, OP; Pulliainen, J; Niemi, K; Luojus, K; Bottcher, K (2012). An optical reflectance model-based method for fractional snow cover mapping applicable to continental scale. REMOTE SENSING OF ENVIRONMENT, 123, 508-521.
An advanced approach for estimating the sub-pixel fraction of snow covered area for boreal forest and tundra belt from optical data is presented. The method named SCAmod by the Finnish Environment Institute (SYKE) is based on a forward semi-empirical model where the at-satellite observed reflectance is expressed as a function of the fractional snow cover (FSC). The effective forest transmissivity determined for each target unit-area and the generally applicable reflectances of three major contributors (wet snow, forest canopy and snow-free ground) serve as model parameters. The forest transmissivity describes the visibility of the ground through the forest canopy from above, and can be determined from visible reflectance data acquired at full snow cover conditions. SCAmod can be applied to data from various sensors operating at optical and near-infrared region. We apply the method to Envisat/AATSR and Terra/MODIS data and validate the resulting FSC against ground truth data over Finland, with considerations of feasibility of these ground truth data in a scale of the MODIS/AATSR pixel. In comparison, also NASA MOD10_L2 fractional snow product is validated. The results indicate that SCAmod performs better than MOD10_L2 particularly in forested areas; an RMSE of 0.11 for the fraction of snow covered area (range is 0-1) is achieved. For large-scale snow mapping, using reflectance data for transmissivity determination would evidently be very time-consuming. Here we present also a new method for transmissivity generation using global land cover data and demonstrate its use for snow cover mapping in continental scale. The resulting FSC data are compared against FSC from high-resolution Landsat TM/ETM + data for selected areas in Northern and Eastern Europe. The comparison indicates that SCAmod is feasible also in continental-scale snow mapping and is superior in identifying snow in dense forests. (C) 2012 Elsevier Inc. All rights reserved.