87th AMS Annual Meeting

Tuesday, 16 January 2007: 4:00 PM
Comparing Local Ensemble Transform Kalman Filter with 4D-Var in a quasi-geostrophic model
208 (Henry B. Gonzalez Convention Center)
Shu-Chih Yang, Univ. of Maryland, College Park, MD; and E. Kalnay
Two data assimilation schemes, Local Ensemble Transform Kalman Filter (LETKF) and 4D-Var, are implemented in the quasi-geostrophic model. Data assimilation experiments are performed in order to understand the advantages/disadvantages of these two schemes for operational purpose.

Our results show that the LETKF is comparable to 4D-Var with a longer (24 hr) assimilation window but it performs much better than 4D-Var with 12-hour assimilation window. In our experiments, parameters like the localization of observations and additive random perturbations are important for LETKF to derive the best performance. Results of different strategies for variance inflation (multiplicative, additive, associated with observation errors) will be discussed.

Supplementary URL: