Extended Abstract (268K)J2.4
A local ensemble Kalman filter for the NCEP GFS model
Istvan Szunyogh, University of Maryland, College Park, MD; and G. Gyarmati, B. R. Hunt, E. Ott, A. V. Zimin, E. Kalnay, D. Patil, and J. A. Yorke
In this paper, we present the implementation of the Local Ensemble Kalman Filter (LEKF, Ott et al. 2003 ) on the T62, 28-level version of the full operational NCEP GFS model. We will demonstrate that the LEKF scheme is efficient in assimilating a large number of observations, as it is well suited to a massively parallel computing environment. Our experiments, assimilating simulated observations (obtained by perturbing a known true state), show that, with the current version of the code, the assimilation of 1.7 million observations takes about 10 minutes on a 40-processor cluster of 2.8 GHz Xeon processors (a $150,000 computer). Also, assimilating observations at a mere 3% of the model grid points provides analyses that are as accurate as those obtained by observing the atmospheric state at all grid points. A modest size (40-member) ensemble is sufficient to achieve a global analysis rms error that is significantly lower than the uncertainty in the observations for all observed variables at all model levels. Preliminary results, assimilating real observations by the LEKF, will also be presented.
Joint Session 2, Data Assimilation and observational network design: Part II (Joint between the Symposium on Forecasting the Weather and Climate of the Atmosphere and Ocean and the 20th Conference on Weather Analysis and Forecasting/16th Conference on Numerical Weather Prediction) (ROOM 6A)
Monday, 12 January 2004, 1:30 PM-2:30 PM, Room 6A
Previous paper