 |
| Titel |
Comparison of three different methods of perturbing the potential vorticity field in mesoscale forecasts of Mediterranean heavy precipitation events: 10 June 2000 and 9 October 2002 |
| VerfasserIn |
Maria-del-Mar Vich, Romualdo Romero, Evelyne Richard, Philippe Arbogast, Karine Maynard |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250046479
|
|
|
|
|
|
| Zusammenfassung |
| Heavy precipitation events occur regularly in the western Mediterranean region. These events
often have a high impact on the society due to economic and personal losses. The
improvement of the mesoscale numerical forecasts of these events can be used to prevent or
minimize their impact on the society. In previous studies, two ensemble prediction systems
(EPSs) based on perturbing the model initial and boundary conditions were developed and
tested for a collection of high-impact MEDEX cyclonic episodes. These EPSs perturb the
initial and boundary potential vorticity (PV) field through a PV inversion algorithm. This
technique ensures modifications of all the meteorological fields without compromising the
mass-wind balance. One EPS introduces the perturbations along the zones of the
three-dimensional PV structure presenting the local most intense values and gradients of the
field (a semi-objective choice, PV-gradient), while the other perturbs the PV field over the
MM5 adjoint model calculated sensitivity zones (an objective method, PV-adjoint). The
PV perturbations are set from a PV error climatology (PVEC) that characterizes
typical PV errors in the ECMWF forecasts, both in intensity and displacement. This
intensity and displacement perturbation of the PV field is chosen randomly, while its
location is given by the perturbation zones defined in each ensemble generation
method.
A new approach based on a manual perturbation of the PV field has been tested and
compared with the previous results. This technique uses the satellite water vapor (WV)
observations to guide the correction of initial PV structures. The correction of the
PV field attempts to improve the match between the PV distribution and the WV
image, taking advantage of the relation between dark and bright features of WV
images and PV anomalies, under some assumptions. Afterwards, the PV inversion
algorithm is applied to run a forecast with the corresponding perturbed initial state
(PV-satellite).
The non hydrostatic MM5 mesoscale model has been used to run all forecasts. The
simulations are performed for a two-day period with a 22.5 km resolution domain (Domain 1
in http://mm5forecasts.uib.es) nested in the ECMWF large-scale forecast fields. The MEDEX
cyclones of 10 June 2000 and 9 October 2002, that produced heavy precipitations over
Catalonia, are a suitable testbed to compare the performance of each ensemble and the
PV-satellite method.
The performace of both ensemble forecasting systems and PV-satellite technique is
evaluated through the verification of the rainfall field. The results show that the PV-satellite
technique performance lies within the performance range obtained by both PV-gradient and
PV-adjoint ensembles; it is even better than the non-perturbed ensemble member. Thus,
perturbing randomly using the PV error climatology and introducing the perturbations in the
zones given by each EPS captures the mismatch between PV and WV fields better than
manual perturbations made by an expert forecaster, at least for the studied events. |
|
|
|
|
|
|