The Data Assimilation Research Testbed (DART): New Capabilities for Analysis and Forecasting of Earth System Components

The Data Assimilation Research Testbed (DART), is an open source community software facility for ensemble data assimilation developed at the National Center for Atmospheric Research (NCAR). DART continues to expand in breadth of models and observation sets available, and applications of its products to geophysical research. This poster highlights a few recent DART developments.

DART has been used with the Weather Research and Forecasting (WRF) model, to provide real-time, mesoscale analyses that were used as initial conditions for convective-scale forecasts. Retrospective testing with DART cycling experiments identified systematic WRF model errors and guided physics suite tuning that led to significant improvements in forecast skill.

Ocean analyses generated by DART with the Parallel Ocean Program (POP) assimilating World Ocean Database observations exhibit improved analyses compared to products generated from purely surface-forced model states. These DART analyses generate improved ocean forecasts for up to several years lead time. The POP/DART assimilations also highlight parts of POP and the observation network which need improvement.

Recent availability of global observations of ionospheric parameters, especially from GPS receivers on low Earth orbiting platforms, has motivated attempts to assimilate ionospheric data. However, assimilation of sparse, irregularly distributed thermosphere observations into global models remains a daunting task. Progress has been made by interfacing NCAR's Thermoshere-Ionosphere Electrodynamics General Circulation Model (TIEGCM) with DART. In this context the thermosphere-ionosphere coupling is self-consistently treated both in the forecast model and in the assimilation scheme, enabling the inference of thermospheric states from the better observed ionospheric states.

DART has been used with both the Community Atmosphere Model (CAM) and WRF to guide the tuning of a Stochastic Kinetic Energy Backscatter Scheme (SKEBS). Using SKEBS leads to increased ensemble variance in both models and reduced RMS errors in analyses for many quantities.

DART can now assimilate a number of new climate system observations including aerosol optical depth (AOD), a variety of variables from flux towers, total precipitable water, neutron intensity from COSMOS (a proxy for soil moisture), and oxygen mixing ratios for upper atmosphere models.

DART can now be used in the 'multi-instance' ensemble environment of the Community Earth System Model (CESM; NCAR's global climate model), which facilitates assimilation with any CESM component. This also opens the door to cross-component data assimilation in a fully coupled model, having an active atmosphere (CAM) ocean (POP), and land surface (Community Land Model, CLM). Observations of one component of the earth system will be able to influence all components of the coupled earth system model. For example, observations of cloud cover could directly influence not only the modeled cloud cover, but the modeled ocean and land temperatures as well. Using DART within CESM also enables assimilation using new CESM components as they become available, e.g. the Spectral-Element (previously 'HOMME') dynamical core version of CAM.