Assessment of Hydrological Impacts from Assimilating Different Satellite-based Snow Cover Fraction Products Using the Updated Unified Forecast System Land-only Data Assimilation Workflow

Seasonal snow is an important part of Earth's climate system. NOAA has recently developed a snow data assimilation (DA) system that uses the Joint Effort for Data assimilation Integration (JEDI) platform to assimilate in-situ snow depth and satellite-based snow cover observations. Due to the flexibility of the JEDI software design, the same snow DA system can also be applied to offline (land-only) model simulations, which will be beneficial to further facilitate research-to-operations transitions (or vice versa).

This study analyzes the impacts of assimilating different satellite-based snow cover fraction products, namely products obtained from NOAA Interactive Multisensor Snow and Ice Mapping System (IMS) at 4km and NASA Visible Infrared Imaging Radiometer Suite (VIIRS) at 5km. The IMS snow cover is an aggregated daily product, using a large variety of sources of imagery and products which, together, provide data for all regions of the Northern Hemisphere. IMS has very good accuracy with human input as an integral part of the detection process. However, it has been reported to suffer from delayed snowmelt timing over North America. By comparison, the daily VIIRS snow cover product is derived from visible and infrared imageries with flags to denote uncertain snow cover detections. This work seeks to determine the relative performance of assimilating these two products at similar spatial resolution into the land component model, and assess specifically whether the additional accuracy derived from the multiple information sources used in IMS can compensate for the delays in detecting changes in snow.

Assimilations are conducted using the updated Unified Forecast System (UFS) Land DA system forced by the fifth generation ECMWF (ERA5) atmospheric reanalysis of the global climate at a spatial resolution of ~25km. Specifically, in this study, we evaluate the Local Ensemble Transform Kalman Filter-Optimal Interpolation (LETKF-OI) assimilation of the satellite snow cover, in terms of its impact on the simulated snow covered area, snow depth, surface albedo, and runoff. These evaluations are conducted through comparing against independent gridded reference products as well as the newly-developed, remote sensing-based extension of the monthly Global Runoff Data Centre discharge estimates. Furthermore, different VIIRS snow cover observation aggregation strategies are assessed. Hydrological impacts from assimilating IMS and VIIRS snow cover products are highlighted.