Snow Depth and Snow Water Equivalent Estimation in the Northwestern Himalayan Watershed using Spaceborne Polarimetric SAR Interferometry

Abstract

Snow depth (SD) and Snow Water Equivalent (SWE) constitute essential physical properties of snow and find extensive usage in the hydrological modelling domain. However, the prominent impact of the hydrometeorological conditions and difficult terrain conditions inhibit accurate measurement of the SD and SWE— an ongoing research problem in the cryosphere paradigm. In this context, spaceborne synthetic aperture radar (SAR) systems benefit from global coverage at sufficiently high spatial and temporal resolutions. Still, existing polarimetric and interferometric SAR techniques are susceptible to high volume scattering resulting from the increased snow grain sizes due to the standing (or old) snow formation driven by the temperature induced snow metamorphosis process. Hence, to model this volume decorrelation, the polarimetric SAR interferometry (Pol-InSAR) technique can be effectively applied. In this work, the standing snow depth (SSD) and its corresponding standing snow water equivalent (SSWE) are estimated using the single-baseline Pol-InSAR based hybrid Digital Elevation Model (DEM) differencing and coherence amplitude inversion model. To achieve this, six TerraSAR-X, TanDEM-X Coregistered Single look Slant range Complex (CoSSC) bistatic quad-pol acquisitions between December 2015 and January 2016 over Dhundi (situated in the Beas watershed, northwestern Himalayas, India) are used. Due to the associated problems of model parameter tuning, complex topographical conditions, and limited ground-truth measurements, appropriate sensitivity analyses have been carried out for the parameter optimisation. Furthermore, the uncertainty sources are identified by performing a summer (June 8, 2017) and wintertime (January 8, 2016) comparative analysis of the study area which quantitatively highlights the changes in the percentages of the surface and volume scatterings. Evidently, the improved model displays sufficiently high overall SSD accuracy with coefficient of determination (R^2) ≈ 0.96, Mean Absolute Error (MAE) ≈ 1.61 cm, and Root Mean Square Error (RMSE) ≈ 2.16 cm. Additionally, the respective SSWEs have been calculated by assuming a fixed snow density for each epoch wherein the overall error metrics are R^2 ≈ 0.71, MAE ≈ 5.19 mm, and RMSE ≈ 6.84 mm. Therefore, this research successfully demonstrates the practicability of the improved Pol-InSAR model for SD estimation over rugged terrains.

A Novel Method for Snow Depth Retrieval Using Improved Dense Medium RVoG Model

“…R represents the correlation between the remote sensing retrieval value and the measurement value, as shown in (25). RMSE and MAE can measure the specific difference between the remote sensing retrieval value and the in situ measurement value, as shown in (26) and (27). Compared with RMSE, MAE is less susceptible to outliers and generally regarded as more reliable than RMSE…”

“…However, compared with forests and glaciers, snow has a smaller thickness and is a dense medium, thus traditional retrieval methods for PolInSAR cannot be directly applied. In recent years, some studies have tried to use PolInSAR to retrieve SD and proved that PolInSAR has great potential in retrieving SD, for example, semi-empirical model [26], CAI model [4], hybrid DEM differencing and coherence amplitude algorithm [27], and dense medium random-volume-over-ground (DM-RVoG) model [18]. Among them, the recently proposed DM-RVoG model involving the dense medium properties of snow is very attractive for retrieving SD.…”