Comparative Energetics of FGGE Re-Analyses Using Normal Mode Expansion

Abstract

In this study, spectral energetics characteristics of FGGE III-b original analyses and the re-analyses are compared in order to assess the achievement of the FGGE project. We compare both of the GFDL and ECMWF versions of the old and new FGGE analyses. It has been noted that the old version of the FGGE GFDL data is noisy, containing excessive small-scale disturbances. The GFDL dataset also appears to be inconsistent with respect to the wind and mass balance. These characteristics result largely from the technique of the continuous data injection as well as from GFDL's philosophy such that the observed data should avoid model bias as much as possible. In contrast, the old version of the FGGE ECMWF data appears to be too smooth, showing a significant damping of divergent wind, especially in the tropics. The data assimilation and initialization cycle at the ECMWF is based on a philosophy such that a successful analysis is one which leads to a successful weather prediction. With this concept, the ageostrophic component of the atmospheric motion is damped to less than 5 percent in magnitude. According to normal mode energetics diagnosis, which separates atmospheric motions into Rossby and gravity modes, it turns out that the gravity mode energy level ECMWF is only 58 percent of that in GFDL for the old analyses.
The FGGE original analyses have been revised by the FGGE re-analysis based on knowledge and criticism gained after the FGGE. The GFDL re-analysis is now smoother, and the ECMWF re-analysis contains much divergent wind. In this study we have conducted the same normal-mode energetics analysis for the FGGE re-analyses, and the results are compared with the FGGE original analyses. It is found, however, that the gravity-mode energy level in ECMWF is still only 42 percent of that in GFDL. The significant difference remains in the re-analyses as before The possible causes of this discrepancy are discussed.