Advancing Earth System Science Using DART: State-of-the-Art Models, Novel Ensemble Algorithms and Extensive Observations

The Data Assimilation Research Testbed (DART) is a community facility for ensemble data assimilation developed and maintained by the National Center for Atmospheric Research (NCAR). DART utilizes ensemble filtering and combines realizations of the model state, such as ocean salinity and soil moisture, with available observations in order to obtain improved estimates of the state and the associated uncertainty. This work will highlight recent efforts and capabilities that have been added to DART. These include:

• New filtering algorithms including a Quantile Conserving Ensemble Filter for non-Gaussian quantities such as ice thickness and a Randomized Dormant Ensemble Kalman Filter that addresses extreme sampling errors.
• Support for new ocean models: the Modular Ocean Model (MOM6) and the Marine Biogeochemistry Library (MARBL).
• Improved model interfaces and functionality for existing models such as TIEGCM, MPAS, CAM, CAM-SE and MITgcm.
• New ocean observation converters that allow the assimilation of ocean biogeochemistry data such as nutrient concentration, dissolved oxygen and satellite surface chlorophyll.
• Incorporation of bias correction tools and flux tower observations for enhanced land data assimilation using the Community Land Model; CLM.
• Ability to hybridize ensemble covariances with time-invariant climatological ones for improved hydrological modeling and flood prediction.
• Enhanced software efficiency and improved scaling for high resolution models.
• Flexible asynchronous assimilation scripts, helpful documentation and setup scripts for porting new models and upgraded webpage.