Stochastically Based Wet Snow Detection with Multitemporal SAR Data

This paper proposes a new method for wet snow detection using multitemporal SAR data. The proposed change detection method is primarily based on the comparison between two X band SAR images acquired during the accumulation (winter) and the melting (spring) seasons, in the French Alps. The new membership decision method is relying on the local intensity statistics of the SAR images by considering the backscattering ratio as a stochastic process: the probability that " the intensity ratio is smaller than the corresponding locally varying dry/wet snow threshold " is larger than a predefined confidence level. The proposed snow backscattering simulations indicate a far more complex relation between the backscattering properties of the two snow types, with respect to the conventional assumption of the augmented electromagnetic absorption associated to the wet snow. The most conclusive results are confirmed at X band by in situ measurements. We show that this complexity is mostly caused by the dominance of a different backscattering component for each type of snow, leading to a significant angular sensitivity. Therefore, we introduce a variable threshold matrix instead of an unique threshold: the threshold is expressed as a function of the local incidence angle (LIA) for each pixel. The proposed snow backscattering simulator (SBS) allows deriving the locally varying threshold as the ratio between dry and wet snow backscattering coefficients computed for the SAR system parameters (polarization, frequency), the local topography parameters (LIA) and for the a priori derived and spatially averaged target parameters (underlying ground and snow cover surface properties, snow grain size, underlying ground dielectric permittivity, snow wetness). The threshold varying with respect to the frequency and the choice of the winter season reference image, allow the proposed algorithm applicability in wider frequency range