| Summary: | The highly dynamic nature of snow requires frequent observations to study its various properties. Keeping this in mind, the present investigation presents results from the analysis of fully polarimetric synthetic aperture radar (POLSAR) parameters for the development of a snow depth (SD) inversion model for SD retrieval. Snow depth retrieved using ground penetrating radar (GPR) at 500 MHz over Austre Grønfjordbreen in the Svalbard region was used to understand the behaviour of certain polarimetric parameters. A significant correlation was found between field-measured SD and POLSAR parameters, namely coherence and normalized volume scattering power (R2 = 0.84 and R2 = 0.73, respectively.) Using the POLSAR scattering powers obtained from the six-component model-based decomposition (6SD), the heterogeneity and anisotropic behaviour in the firn areas are also explained. Further, based on the analyses shown in this work, a polarimetric parameter-based SD inversion algorithm have been proposed and validated. The univariate model with co-polarization coherence has the highest correlation (R2 = 0.84, Root Mean Square Error (RMSE) = 0.18). We have even tested several multivariate models for the same, to conclude that a combination of coherence, normalized volume and double-bounce scattering have a high correlation with SD (R2 = 0.84, RMSE = 0.18). Additionally, temporal and spatial variability in SD was also observed from three polarimetric SAR images acquired between 4 April 2015 and 15 May 2015 over the Western Nordenskiöld Land region. Increase in snow depth corresponding to snow precipitation events were also detected using the POLSAR data.
|
|---|