Robust EnKF using L1 and Huber norms

Adrian Sandu*
Virginia Tech

Presence of large errors in some observational data, e.g., data collected from a faulty instrument, negatively affect the quality of the data assimilation results.
We discuss a systematic framework for robust data assimilation using EnKF. The approach is based on replacing the traditional L2 norm formulation of data assimilation problems with formulations based on L1 and Huber norms. Numerical experiments using the Lorenz-96 and the shallow water on the sphere models illustrate how the new algorithms outperform traditional EnKF approaches in the presence of data outliers.

*Preference: Oral