A Community Data Assimilation Facility for Confronting Climate GCMs with Observations

The Data Assimilation Research Testbed (DART) is a community ensemble data assimilation facility developed by the Data Assimilation Research Section at the National Center for Atmospheric Research. DART can assimilate all the in situ observations routinely used for operational numerical weather prediction by centers like the National Centers for Environmental Prediction (NCEP). DART can also assimilate a variety of remote sensing observations like GPS radio occultation, satellite cloud motion winds and scatterometer surface winds. DART is routinely used with both regional and global atmospheric models including NCAR's Community Atmospheric Model (CAM) and the AM2 model from the Geophysical Fluid Dynamics Laboratory. Confronting climate models like CAM and AM2 with observations in NWP mode can assess the relative quality of model dynamics and parameterizations and identify model errors. DART provides a variety of diagnostic tools that compare analyses and forecasts to observations. The ensemble analyses produced by DART facilitate sensitivity analysis that can give insight into the model dynamics and can help to tune physical parameterizations. DART also allows climate modelers to assimilate special observations from field programs or novel instruments that are not used in operational NWP analyses. It is straightforward to add new tracers and observations to DART models. For instance, DART/CAM was modified to support the recent ARCTAS field experiment by including carbon monoxide as a tracer and assimilating observations from the MOPITT instrument on NASA's EOS Terra satellite.