A new snow depth retrieval method by improved hybrid DEM differencing and coherence amplitude algorithm for PolInSAR

Abstract

Snow depth is a critical parameter for measuring the variations of the snowpack, and its measurement also influences the accuracy of hydrological and climate models. Polarimetric SAR interferometry (PolInSAR) technology combines the capabilities of Polarimetric SAR (PolSAR) and Interferometric SAR (InSAR), and has significant potential for retrieving snow depth. The hybrid DEM differencing and coherence amplitude algorithm (HDCA) is a robust algorithm for PolInSAR data, but encounters such problems as uncertainty in finding the ground scattering phase center, impure volume decorrelation, and the lack of consideration of the dense medium characteristic of snow. In this study, the HDCA is improved by optimizing the phase and decorrelation parameters and by considering the dense medium characteristic of snow. First, the Freeman-Duren decomposition method is put forward to extract the ground and volume scattering phase centers from the snow. Second, the coherence region boundary estimation method is proposed to optimize the volume decorrelation. Following this, the parameter assumed by the original method in free space is adjusted by considering the dense medium characteristic of snow. Finally, the Ku band UAV SAR data were applied to validate the new method, and in-situ data were used to assess its accuracy. The correlation coefficient (R), root mean square error (RMSE), and mean absolute error (MAE) of the proposed method are 0.93, 3.59 cm, and 2.98 cm, respectively, and were significantly superior to those of the original HDCA method of 0.74, 7.44 cm, and 6.04 cm, respectively.