Geoscience Reference
In-Depth Information
and found that the reduction in analysis error in terms of the difference of the
vertically integrated analysis ensemble spread of the geopotential height (OSE-
ALERA) extends eastward in Kelvin waves and westward in Rossby waves. Moteki
et al. also found signals indicating that the tropical cyclogenesis is affected by
the Kelvin wave. In the Arctic Ocean, Inoue et al. ( 2009 ) conducted data denial
experiments to evaluate the surface pressure observations by drifting buoys. The
analysis ensemble spread, not only in the Beaufort Sea, where buoys are densely
deployed, but also throughout the whole Arctic Ocean, increased without the surface
pressure observations north of 70 N. The influence of the buoys reaches 700 hPa.
The relationship between the number of buoys and accuracy is confirmed with
consistency among the different reanalysis data sets.
The analysis is not static but varies in time under the influence of atmospheric
disturbances. Enomoto et al. ( 2010 ) investigated the relationship between the anal-
ysis ensemble mean and the spread of ALERA in various atmospheric phenomena.
These researchers found an increase in the analysis ensemble spread prior to the
westerly bursts in the tropical eastern Indian Ocean, in the onset of the monsoon
westerlies in southern Vietnam, and even in the stratospheric sudden warming.
Because the error growth implies instability in the linear perturbation theory, it is
anticipated that the analysis ensemble spread, which is an estimate of the analysis
error, contains some precursory signals. The actual error growth is, however,
nonlinear owing to the finite amplitude and complex physical processes and is not
fully understood. Further investigations into the precursory signals contained in the
analysis ensemble spread require the variables related to physical processes, which,
unfortunately, are not included in ALERA.
Motivated by the success of the observing-system experiments and predictability
studies, JAMSTEC formed a research team called OREDA (Observing-System
Research and Ensemble Data Assimilation development research team). This article
reports the activities of OREDA related to the development of the ensemble data
assimilation system of the global atmospheric data and OSEs. The updated version
of the ensemble data assimilation system (ALEDAS2) is described in Sect. 21.2 .
OSEs were conducted with ALEDAS2. Preliminary results are shown in Sect. 21.3 .
Finally, the summary and our plan for future research are given in Sect. 21.4 .
21.2
The Ensemble Data Assimilation System
ALEDAS2 is composed of improved versions of AFES and LETKF. In the
forecast step, an ensemble forecast is conducted with AFES to propagate the
mean and covariance to the next time level. In the analysis step, LETKF, one
of the deterministic Kalman filters that assimilate observations to the ensemble
mean, is used. This section summarizes the updates of the forecast model and
analysis scheme and describes the forecast-analysis cycle of ALEDAS2. With this
system, we are producing ALERA2, a successor to ALERA. The configurations for
ALERA2 are described in comparison with those for ALERA.
Search WWH ::




Custom Search