Validation of Absorption Based Dual-Frequency Radar Retrieval Algorithm of Snow Water Equivalent using SnowScat and SnowSAR Data

Abstract - In this paper we validate a new radar retrieval algorithm of snow water equivalent (SWE) using X- and Ku- band backscatter. The algorithm is absorption based. The algorithm is validated against available ground based SnowScat and airborne SnowSAR data and proved to be effective. The study is based on experiences and data collected in support of the proposed ESA Cold Regions Hydrology High Resolution Observatory (CoReH2O) and the NASA Snow and Cold Land process (SCLP) Satellite Missions. Both missions proposed to deploy a dual frequency (X- and Ku-band) and dual-polarized (VV and VH) Synthetic Aperture Radar (SAR). The adopted retrieval algorithm addresses two main difficulties (i) the snow microstructure and stratigraphy (ii) and background backscattering from the underlying ground and vegetation. In the algorithm, the background scatterings are subtracted from the total scattering which then gives the volume scattering of snow. From the volume scattering of snow, the radar retrieval algorithm is based on the absorption loss of the snowpack which is directly proportional to the SWE. Our approach adopts a physical based bicontinuous media / DMRT scattering model to establish regression formulas between multiple and single scattering and correlations between the scattering albedos and optical thicknesses at dual-bands. With the two co-polarization backscatters and the background scatterings, the snow volume scatterings are extracted and used to invert the scattering albedos and optical thicknesses at two bands, giving a retrieval of the absorption loss and SWE. We have applied the algorithm to the Finland SnowScat / SnowSAR, the Canada SnowSAR and the Austria AlpSAR data. The retrieval algorithm is shown to be effective, achieving a mean error of 15 mm for both the tower mounted SnowScat time series and airborne SnowSAR data, which is smaller than the 20mm RMSE requirement of SCLP. Previously the algorithm has only been validated for location fixed and time-varying SnowScat data. We also report on snow microstructure characterization and proposed background measurements of vegetation and ground.