Extended Abstract (1.2M)4.3
Recent development of global and regional local ensemble transform Kalman filters (LETKF) at JMA
Takemasa Miyoshi, Japan Meteorological Agency, Tokyo, Japan; and S. Yamane, Y. Sato, K. Aranami, and T. Enomoto
A four-dimensional local ensemble transform Kalman filter (4D-LETKF) has been developed and applied to global and mesoscale models at the Japan Meteorological Agency (JMA). First, in collaboration among the JMA, the Earth Simulator Center (ESC), and the Chiba Institute of Science, a 4D-LETKF has been developed and applied to the AFES (AGCM for the Earth Simulator) at a T159/L48 resolution. Following the successful preliminary investigations with perfect model experiments, real observations are assimilated to perform experimental ensemble reanalysis beginning from May 2005. The products will be publicly available through the Internet at the webpage of the ESC.
Following the successful experience assimilating real observations on the Earth Simulator, the 4D-LETKF has been applied to JMA's operational global model known as the GSM but with a lower TL159/L40 resolution (currently the operational model resolution is TL319/L40). The 4D-LETKF system is integrated in the same manner as the operational system including the observational quality control. Real observations are assimilated in the period from July 21, 2004 to September 9, 2004, so that the results are verifed in entire August. The system performance will be compared with JMA's operational global 4D-Var system.
Finally, the 4D-LETKF has been modified in accordance with the difference in the lateral boundaries and prognostic variables, so that it was applied to JMA's operational nonhydrostatic mesoscale model (MSM). Following the successful perfect model experiments with 5-km grid spacing and a small computing area around Tokyo, real observations in July 2004 are assimilated with 20-km grid spacing and the same area as JMA's operational MSM. The results will be compared with the operational MSM analysis/forecast system.
Session 4, Advanced Methods for Data Assimilation
Tuesday, 16 January 2007, 1:30 PM-5:15 PM, 208
Previous paper Next paper
Recorded presentation