Comparing multisensor optical-radar approaches for snow water equivalent retrievals

For several decades, satellite remote sensing has been a valuable tool for mapping snow cover globally, regionally, and locally. However, no remote sensing technique can accurately measure snow water equivalent (SWE) from space for mountain hydrologic applications. Optical sensors like MODIS, VIIRS, and Landsat are robust for measuring the fractional snow-covered area (fSCA) at various spatial and temporal resolutions. Yet, these optical methods are limited by cloud cover and do not provide information on SWE. Synthetic aperture radar (SAR) can penetrate clouds, has a fine spatial resolution, and various algorithms allow us to quantify both SWE magnitude and changes. While SAR-based techniques show promise for SWE monitoring, they cannot discriminate between snow-free and snow-covered areas when the snow is dry. To address this SWE monitoring challenge, we evaluate a multisensor approach that leverages the strengths of both optical and radar sensors. Our study aims to better understand the variability between common snow cover data products and how that uncertainty propagates into InSAR-based SWE retrieval techniques. We analyzed three UAVSAR InSAR flight lines over the Sierra Nevada, CA, from the SnowEx 2020 campaign and compared six satellite-based snow cover products. First, we computed InSAR-based SWE change estimates using in situ snowpack data. We then compared the summed SWE change values with a moving window analysis to quantify product variability. Results show that moderate-resolution (375–500 m) NDSI-based products provide broadly similar volumetric SWE change results to those using spectral unmixing retrieval methods. This suggests that the readily available moderate-resolution snow cover products from MODIS and VIIRS are adequate for an optical-radar SWE monitoring approach. Future work should focus on understanding how sub-canopy snow in forested regions affects snow cover product accuracy and variability. Furthermore, near-real-time, high-resolution cloud- and gap-filled optically-derived snow cover data will be important for supporting water resources decision-making.