Data Assimilation Meetings at Reading
Date | Meeting type | Speakers |
4 July 2012 | DARC seminar |
Javier Amezcua (UoR, previously at University of Maryland) Ensemble clustering in deterministic ensemble Kalman filters (pdf) Ensemble clustering (EC) can arise in data assimilation with Ensemble Square Root Filters (EnSRFs) using nonlinear models: an M-member ensemble splits into a single outlier and a cluster of M-1 members. Although the covariance is not directly affected, the higher order moments of the ensemble are. Starting with a very simple mode, we perform a comprehensive study of ensemble clustering. A metric designed to identify it is introduced. We examine whether or not clustering is an irreversible phenomenon of EnSRFs, and what happens as the size of the model increases and further implementation aspects (such as localization and inflation) are needed. |