Radar Retrieval Algorithm for Snow Water Equivalent Applied to SnowSAR and Scatterometer Data

We have developed a radar retrieval algorithm of snow water equivalent (SWE) based on a parameterized model of bicontinuous dense media radiative transfer (Bic/DMRT). In the algorithm, regression formulas are derived from look-up tables (LUTs) of the physical model of bicontinuous media / DMRT to relate the X- and Ku- band backscatter with two parameters: the scattering albedo and SWE. The background scattering is subtracted from the radar observations to separate out the volume scattering of snow. Based on the volume scattering and the a priori estimation of scattering albedo, a cost function is utilized to determine SWE. The retrieval algorithm was validated against 3 sets of airborne SnowSAR data (2011 and 2012 campaigns in Finland; 2013 campaign in Canada). The performance of the algorithm has a root-mean-square error (RSME) below 30 mm of SWE and a correlation coefficient larger than 0.6. The most challenging aspect of the radar retrieval algorithm is the estimation of the scattering albedo. To obtain a reasonable a priori estimation of the scattering albedo (strongly influenced by snow microstructure, or grain size), we combine the high sensitivity of passive microwave measurements to snow microstructure, with the high spatial resolution of the radar data. The radar retrieval algorithm takes an initial passive microwave derived microstructure guess and retrieves both albedo and SWE. This new scheme is tested using ground-based scatterometer data from Finland (ESA NoSREx; 2009-2013) and Colorado (NASA SnowEx 2017) dataset. The performance of the retrieval is improved compared to estimates derived solely using radar data. We also apply the SnowEx in-situ data to improve the background scattering estimation, which reduces the uncertainty caused by radar interactions with the soil surface.