The Data Assimilation Research Testbed (DART) is being extended to support the Community Land Model (CLM 4.0) to assimilate observations of land-based quantities such as snow cover, soil moisture, soil temperature, and carbon and water fluxes. This is the first effort linking DART and a land surface model. The major obstacle is to communicate a heterogeneous set of state information and metadata between CLM4.0 and DART, and to accommodate the use of multiple instances of CLM within the Community Earth System Model (CESM 1.0). The DART facility allows extensible support for additional observation types with a small amount of effort and provides a rich suite of assimilation algorithms and diagnostic tools.

Experience with other models has shown that the spread of the ensemble of CLM states is more likely to be maintained if each ensemble member has a distinct meteorological forcing. A freely available ensemble of reanalysis data created by DART and the Community Atmospheric Model (CAM V4.0) is used and compared to a similar experiment using only one meteorological forcing field. The advantages and disadvantages of the forcing fields will be discussed. The work is focused on snow data assimilation with the purpose of improving the performance of CLM on snow simulation. As the first step, Moderate Resolution Imaging Spectroradiometer (MODIS) snow cover fraction observations are assimilated daily. The effectiveness of the assimilation is assessed by evaluating the 24-hour forecasts against the observations before they are assimilated.