Empowering Geoscience with Improved Data Assimilation Using the Data Assimilation Research Testbed "Manhattan" Release

The capabilities of the Data Assimilation Research Testbed (DART) at NCAR have been significantly expanded with the recent "Manhattan" release. DART is an ensemble Kalman filter based suite of tools, which enables researchers to use data assimilation (DA) without first becoming DA experts.
Highlights:

• significant improvement in efficient ensemble DA for very large models on tens of thousands of processors,
• direct read and write of model state files in parallel,
• more control of the DA output for finer-grained analysis,
• new model interfaces which are useful to a variety of geophysical researchers,
• new observation forward operators and the ability to use precomputed forward operators from the forecast model.

The new model interfaces and example applications include the following:

• MPAS-A; Model for Prediction Across Scales - Atmosphere is a global, nonhydrostatic, variable-resolution mesh atmospheric model, which facilitates multi-scale analysis and forecasting. The absence of distinct subdomains eliminates problems associated with subdomain boundaries. It demonstrates the ability to consistently produce higher-quality analyses than coarse, uniform meshes do.
• WRF-Chem; Weather Research and Forecasting + (MOZART) Chemistry model assimilates observations from FRAPPE' (Front Range Air Pollution and Photochemistry Experiment).
• WACCM-X; Whole Atmosphere Community Climate Model with thermosphere and ionosphere eXtension assimilates observations of electron density to investigate sudden stratospheric warming.
• CESM (weakly) coupled assimilation; NCAR's Community Earth System Model is used for assimilation of atmospheric and oceanic observations into their respective components using coupled atmosphere+land+ocean+sea+ice forecasts.
• CESM2.0; Assimilation in the atmospheric component (CAM, WACCM) of the newly released version is supported. This version contains new and extensively updated components and software environment.
• CICE; Los Alamos sea ice model (in CESM) is used to assimilate multivariate sea ice concentration observations to constrain the model's ice thickness, concentration, and parameters.