Inversion of snow geophysical parameters using the VHR PAZ X-band dual polarimetric SAR data: first known experiments in the Himalayan region

• First known experiment of VHR PAZ –X band SAR data for snow geophysical parameter retrieval. • Utilized the dynamic range of snow density instead of constant value to retrieve the CPD thereby improving estimates. • Improved algorithm using average CPD shows high accuracy with R 2 = 0.85 for snow depth and R 2 = 0.77 for SWE. Results indicate the applicability of the approach in old/settled alpine snowpack with predominantly dry conditions. Snow plays a vital role in mountain hydrology, water resources, and Earth’s planetary budgets. Therefore, monitoring snow geophysical parameters (SGPs), such as snow density, Snow depth (SD), and snow water equivalent (SWE), is imperative hydrological dynamics and forecasting water availability. Moreover, radar remote sensing offers significant capabilities for estimating these parameters. In this study, we utilized PAZ X-band dual-polarimetric data to estimate SGPs. To the best of our knowledge, this is the first known experiment using PAZ data for SGP estimation. In the present work, we utilized the copolar phase difference (CPD) for SD and Integral Equation model (IEM) for snow density. In this study, we proposed an improved algorithm for SD inversion, instead of relying on a single in-situ snow density value, we incorporated a range of snow densities (0.15 to 0.27 g/cm3), optimizing the axial ratio between 1.13 and 1.17. This density range and optimized axial ratio were used to minimise the error between C P D O b s and the average C P D M o d . The algorithm yielded high-resolution modelled SD and density at a 2.5 m spatial resolution, which were later used to estimate SWE. Algorithm validation was performed using in-situ data of Gulmarg region of Kashmir, India, with statistical metrics such as mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R2), and percentage error (PE). SD estimates showed high correlation, with R2 = 0.85, RMSE = 3.18 cm, PE = 1 %, and MAE = 2.85 cm. Similarly, SWE estimates had an R2 of 0.77, RMSE = 1.032 cm, PE = 5 %, and MAE = 0.814 cm, demonstrating the model’s accuracy and reliability.