 |
| Titel |
A dynamical analysis of the impact of targeted observations |
| VerfasserIn |
E. Irvine, S. Gray, J. Methven, I. Renfrew, K. Bovis, R. Swinbank |
| Konferenz |
EGU General Assembly 2009
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 11 (2009) |
| Datensatznummer |
250025366
|
|
|
|
|
|
| Zusammenfassung |
| The impact of targeted observations on forecasts for Northern Europe
is presented. The targeted data consists of dropsondes from four
flights conducted between Iceland and Greenland during the Greenland
Flow Distortion Experiment in February and March 2007. The sonde data
was sent to the global telecommunications system, enabling it to be
assimilated into operational 12Z forecasts. Later, hindcast studies
were performed to assess the impact of the targeted observations on
the forecast. These used the Met Office 4D-VAR data assimilation
scheme with a 48km grid and the Unified Model over the North Atlantic
European domain with a 24km grid.
Here we focus on the case with greatest impact on downstream
forecasts. Targeted sondes were released over the Denmark Strait and
Irminger Sea into a region that total-energy singular vectors (from
ECMWF) predicted to be sensitive to initial condition errors. The
impact from the targeted sondes moved from the target region (where
the sondes were released) with a developing polar low, into the
verification region over Scandinavia. This resulted in a small
forecast improvement at short lead times of 7%, measured in terms of
total energy. It is shown that the improvement was caused by the
modification of the position of a tropopause fold and its associated
circulation, which was forcing the development of the polar low. It
is shown that it is possible to increase the impact from targeted
observations using two methods. Firstly, reducing the dropsonde
observation error variance gives the targeted data increased weighting
during the assimilation process, and results in an increase in average
forecast improvement of a few percent. Secondly, assimilating all
dropsondes, even if released close together, doubles the maximum
forecast improvement. Together these methods resulted in a forecast
improvement due to the inclusion of targeted sondes of 17%. |
|
|
|
|
|
|