Retrieving Snow Wetness Based on Surface and Volume Scattering Simulation
Wei Ma,Pengfeng Xiao,Xueliang Zhang,Yina Song,Tengyao Ma,Lizao Ye
DOI: https://doi.org/10.1016/j.isprsjprs.2020.08.021
IF: 12.7
2020-01-01
ISPRS Journal of Photogrammetry and Remote Sensing
Abstract:Wetness is one of the important physical parameters of snowpack. Its spatial and temporal changes play a key role in snowmelt runoff forecast, regional climate change, and agricultural irrigation. In this study, we proposed a new method to retrieve snow wetness from full-polarimetric synthetic aperture radar (SAR) data. First, the dominant scattering components, i.e. surface and volume scattering, in wet-snow conditions were obtained by polarimetric decomposition. The random rough surface scattering models and the dense media radiative transfer (DMRT) model were then used to establish the surface and volume scattering models, through which the surface and volume scattering lookup tables were created, respectively. Based on the lookup tables and the polarimetric decomposition results, the snow surface and volume wetness were retrieved, respectively. Finally, the effective snow wetness was derived from the weighted summation of surface and volume wetness. The advantage of this method mainly comes from the full consideration the snow surface roughness, the local incidence angle on complex mountain terrain, and the polarimetric information. In experiments, the GaoFen-3 data obtained on January 17, 2018 in the Kelan River Basin and the Radarsat-2 data obtained on March 19, 2014 in the Manasi River Basin were selected to verify the applicability of the proposed method at different conditions. From the analysis of experiment results, the correlation coefficient between the estimated and the ground measured snow wetness in the Kelan River Basin is 0.72. The mean absolute error (MAE) and the root mean square error (RMSE) are 3.35% and 3.89%, respectively. The correlation coefficient between the snow wetness estimated by Radarsat-2 and the measured values in the Manasi River Basin is 0.62. MAE and RMSE are 1.32% and 1.62%, respectively. These results proved that the proposed method can effectively retrieve snow wetness under different SAR data, different areas, and different snow periods.