JPSS VIIRS Snow Products at NOAA: Review and Performance Assessment

Information on the global snow cover properties constitutes one of the primary deliverables of the Visible Infrared Imaging Radiometer Suite (VIIRS) instrument onboard Joint Polar Satellite System (JPSS) satellites. High, 350m, spatial resolution along with daily global coverage make VIIRS snow products an important source of information for numerical weather prediction, climate and hydrological models along with other practical applications. VIIRS snow products have been generated continuously since the launch of the first VIIRS instrument onboard SNPP satellite in 2012. A principal enhancement of VIIRS snow cover products occurred in 2018 when a new set of NOAA Enterprise retrieval algorithms was implemented into operations. Using these algorithms, information on the snow cover properties is currently inferred from observations of all three operational JPSS satellites, SNPP, NOAA-20 and NOAA-21. Two primary snow products inferred from the VIIRS data include the binary snow cover map and the snow cover fraction. Daily global gridded snow products are available on a latitude-longitude grid with 0.01⁰ grid cell size or at about 1 km spatial resolution.

In the presentation we give a brief description of VIIRS Enterprise snow retrieval algorithms and present examples of the derived products. Comprehensive quality evaluation and accuracy assessment of the VIIRS snow products is provided routinely. The accuracy of snow identification is estimated by comparing VIIRS snow products with snow cover charts generated interactively by NOAA analysts in the National Ice Center (IMS products) as well as with snow depth reports from ground-based weather stations. Validation studies demonstrate that the agreement rate of VIIRS snow products both IMS and in situ reports consistently exceeds 90% and averages to 94-95% in the course of the year. Besides finite accuracy of snow identification in the VIIRS imagery, part of the disagreement should be attributed to subjectivity of snow cover analysis within IMS and potential errors in automated snow depth reports from ground stations.

We further discuss opportunities to enhance characterization of snowpack properties using combined observations from different satellite sensors. We show that aggregating VIIRS cloud-clear observations from all three JPSS satellites, SNPP, NOAA-20 and NOAA-21 provides on the average 10-15% increase in available clear-sky views at mid- and high-latitudes and a corresponding improvement in the effective daily area coverage compared to data from only one satellite. Even better potential to for improving snow mapping consists in blending VIIRS-derived snow cover with satellite observations in the microwave. We present first results of application of an experimental technique which complements VIIRS snow data with snow retrievals from DMSP SSMIS. The blended product provides continuous (gap-free coverage) on a daily basis, however its mean accuracy drops to about 90-92% in the course of the year. Both products are planned for quasi-operational implementation in the end of 2024.