The DART tutorial

The Data Assimilation Research Testbed tutorial “DART_LAB part 6” guides participants through running ensemble data-assimilation experiments using the real DART software environment.

Building on assimilation concepts covered in DART_LAB part 1-5 you will:

• Install and set up the DART system, including the directory structure and build tools.

• Use the low-order Lorenz 96 model as a testbed to reproduce the DART_LAB Lorenz 96 experiments.

• Explore configurable parameters such as ensemble size, localization, inflation method, and observation networks — and learn how they influence assimilation performance.

• Generate synthetic observations via OSSEs (Observing System Simulation Experiments), assimilate them, and diagnose results using MATLAB tools (time-series error/spread plots, rank histograms, bias/variance diagnostics, etc.).

• Examine observation-space diagnostics, including rank histograms, RMSE/bias evolution, and outlier-observation rejection (via a configurable threshold).

• Extend your familiarity to other low-order models (such as the Lorenz 63 system) and prepare for full-scale geoscience applications using DART’s model interfaces.

Instructions for installing DART and the MATLAB® diagnostics can be found in the DART documentation.