Assimilation of Combined Snow Cover and Snow Depth Data into NCEP Operational Global Forecast System

NOAA/NCEP Gridpoint Statistical Interpolation (GSI) is a unified variational data assimilation system utilized in operations for both the Global Forecast System (GFS) from May 2007 and the North American Mesoscale (NAM) forecast system from June 2006. However, the GSI lacks a land component and cannot assimilate land surface data such as soil moisture and snow. We are using the existing NASA Land Information System (LIS) to serve as a Global Land Data Assimilation System (GLDAS) for GFS. LIS integrates the operational land surface model (Noah) used in NCEP models, high-resolution satellite and observational data, and land DA tools. The LIS EnKF-based DA tool is used to assimilate snow cover area (SCA) from the Interactive Multisensor Snow and Ice Mapping System (IMS) by the operational National Ice Center (NIC) and AFWA snow depth (SNODEP) products.

The land EnKF DA technique is mature after more than a decade of study. Previous assimilation experiments with snow and soil moisture have demonstrated some improvements in land process simulations, which can improve numerical weather predictions. However, some key issues need to be addressed before LIS DA tool is transitioned into the NCEP operational system. The preliminary experiments with current EnKF in LIS show no updates when model simulates no snow cover although the snow is observed, and show a minor updates when model snow amount is small. Targeting these issues, we will improve the representation of error covariance from both model and observations. We will further conduct a set of offline numerical experiments driven by the GFS forecast forcing to evaluate the impact of assimilating snow.