A JEDI-Based Snow Depth Analysis for NOAA's GFSv17/GDAS Global NWP System

As a part of a major upgrade to our global NWP system, NOAA is developing a new land data assimilation system. In addition to introducing a soil moisture and temperature analysis, the new land data assimilation system will include an update to our snow depth analysis. The current snow depth analysis is a rule-based merging of an externally-produced snow depth analysis (SNODEP) and a satellite snow cover product (IMS). We are replacing this with an Optimal Interpolation (OI) assimilation of station snow depth observations and the IMS satellite snow cover. Since NOAA will be adopting JEDI as our data assimilation platform, the new snow depth analysis has been implemented in JEDI. JEDI does not yet include the OI algorithm, and so we have approximated the OI using the Local Ensemble Transform Kalman Filter (LETKF). The LETKF-OI uses the ensemble localization to mimic the error covariance functions assumed by the OI. Assimilation experiments at full resolution using an offline (land-only) version of our future land surface model (Noah-MP) show that while the JEDI LETKF-OI does not perform quite as well as the classic OI, both are a dramatic improvements over an open-loop with no snow depth data assimilation. Tests are underway in the coupled land/atmosphere modeling system, to confirm that the improved snow depth from the JEDI LETKF-OI results in the expected improvements to the atmospheric forecasts.