information from observations with predictions from a numerical

model. Assimilation can be used to produce improved analyses and

forecasts of a geophysical system, to improve numerical models,

and to design and evaluate observing systems.

The Data Assimilation Research Testbed (DART) is a mature community

software facility providing researchers access to state-of-the-art

ensemble assimilation tools. The freely-available DART distribution

includes fully functional low-order and high-order models, support

for commonly available observations, hooks to easily add both new

models and observation types, diagnostic programs to interpret the

results, and a full tutorial suitable for self-study or teaching

data assimilation concepts, including exercises using the models

distributed with DART.

Recent applications of DART with both global and regional atmospheric

models are presented to provide an overview of capabilities. DART has

been used with the WRF regional atmospheric model to produce

ensemble analyses and predictions of Pacific and Atlantic tropical

storms. The impact of using radio occultation observations from the

COSMIC satellites has been evaluated in these cases. DART has also

been applied to global prediction with the CAM general circulation

model. Analysis of the assimilated fields has led to the discovery

of a significant coding error in the polar filtering in the CAM finite

volume core. Ensemble assimilation has also identified noise due to a

previously undocumented numerical instability that is found in the

finite volume cores that have been used for several major climate

models. DART was also used with the CAM/CHEM model to provide

operational support during the ARCTAS field campaign.

New tools to improve and simplify the use of ensemble assimilation

continue to be added to the DART facility. In the past year, improved

tools for automatically adapting to model error (adaptive damped

inflation) and a new nonlinear filtering variant have been introduced.

Examples of improvements to global analyses and forecasts are presented.

New models, both small and large continue to be added to the set

compatible with DART. For instance, During the past year, a version of

the MIT/OCEAN model was added by Ibrahim Hoteit and the Navy's COAMPS

model was added by Tim Whitcomb. An overview of requirements for adding

a new large model are presented.