School of Meteorology Seminar Series presents...
Ensemble Kalman Filter: Status and New Ideas
Annual Tzvi Gal-Chen Memorial Lecture Series Presents:
Eugenia Kalnay
Distinguished University Professor
Department of Atmospheric and Oceanic Science
University of Maryland
31 March 2009, 3:00 PM
National Weather Center, Room 1313
120 David L. Boren Blvd.
University of Oklahoma
Norman, OK
Directions to the NWC (.pdf, 60 kb)
I will go over the most recent comparisons between the two most advanced data assimilation systems, 4D-Var and Ensemble Kalman Filter (EnKF), showing that in Canada the EnKF is giving results equal or better than 4D-Var, as well as comparisons made at JMA (Miyoshi) and with the GSI (Whitaker). Then I will describe some new ideas developed at the University of Maryland that greatly enhance the potential applications of the EnKF. These ideas include simple methods to estimate and correct model errors, estimate online the inflation and observations errors, speed up the spin-up and improve the accuracy with an "outer loop" and repeated use of the observations, and speeding up the code by performing a coarse analysis and interpolating "weights". One of the most important new ideas allows to estimate the forecast sensitivity to observations like the adjoint method of Langland and Baker (2004). However, since the ensemble sensitivity does not require an adjoint model it can be used for longer forecasts (5 days or more) which could help identify the origin of the NCEP 5-day forecast skill "dropouts".