ESTIMATING SEASONAL SNOW ACCUMULATION FROM NASA UAVSAR AT CAMERON PASS, CO

In the western US, seasonal snowpacks yield >70% of the water supply, making the monitoring of this resource invaluable. Daily measurements of snow water equivalent (SWE) are observed by >800 SNOTEL stations and modeled using data assimilation techniques, such as SNODAS. However, the SNOTEL station network is too sparse to measure the spatial distribution of SWE accurately, and SNODAS has known limitations and uncertainties. The InSAR SWE-retrieval, currently under evaluation by NASA SnowEx, is a promising method for high resolution (<100 m) SWE-retrievals, particularly at L-band frequencies (1–2 GHz) where the snowpack is nearly transparent. As part of the ongoing SnowEx Campaigns, NASA UAVSAR flew a fully polarimetric L-band SAR sensor over 13 sites from January to March 2020 and seven sites from January to March 2021, creating a weekly to bi-weekly time-series of InSAR acquisitions. To validate UAVSAR measurements, our team conducted repeat surveys of snow conditions for each flight date, consisting of ground-penetrating radar transects (GPR; ~1 km in 2020, ~2.4 km in 2021), probed snow depths, and snow pits. Additionally, terrestrial lidar surveys were conducted during five survey dates to derive spatially distributed snow depths. We chose to compare two different methods for correcting interferograms for atmospheric delays: (1) a linear phase ramp, derived from a statistical fit of snow-free pixel phase regressed against signal path length, and (2) a correction based on the HRRR weather reanalysis model. For both methods, SWE-changes are subsequently calculated from the phase and incidence angles, following Guneriussen et al. (2001), and then summed sequentially to estimate season-long accumulation. We evaluate the changes in SWE from both methods using the ground observations, which are converted to SWE measurements using the pit-measured snow density and then binned to the UAVSAR grid.