Isabelle Soulan and Miloud Bessafi (Meteo-France/DIRRE/CRC La Reunion),
Ryad El Khatib (Meteo-France/CRNM/GMAP),
Jean-Marie Willemet (Toulouse and Universite de La Reunion)
Like most of the atmospheric models, ARPEGE does not locate the tropical cyclones with enough accuracy. That is why the CRC (Cyclone Research Cell) of Meteo-France in La Reunion island has been developping two bogussing techniques to improve the initial conditions of ARPEGE and the subsequent tracks forecastings. The first one, after an idea of Jean-Frangois Geleyn, is using ALADIN to position the tropical cyclone at the right place. Then the ALADIN "guess" is re-injected in the global model ARPEGE. The second one is easier : we just force the global model to assimilate the mean sea level pressure of the center of the cyclone (calculated by the Dvorak method).
We use the observed position of the tropical cyclone to make a limited area centered on this point. If the tropical cyclone seen by the model is inside this area, we translate the dynamic fields in a non-uniform manner so that the tropical cyclone reaches the observed position while the points near the borders of the domain remain at the same place. This translation technique, using basically the post-processing tools, enables a soft replacement of the ARPEGE fields by the ALADIN fields over and near the cyclone. For the time being, only one kind of translation has been tested : we multiply the translation vector of the center of the tropical cyclone by a cubic fonction of the latitude and a cubic fonction of the longitude. Other functions will be tested later.
We only have a few experiments available, but the preliminary results are rather good : the new ARPEGE fields are not too distorted and the forecasted tropical cyclone track is better than without this bogus. But the low in the center of the cyclone is not deeper.
The forecasters of the RSMC (Regional Specialized Meteorological Centre) for tropical cyclone of La Reunion use the Dvorak method to calculate the position and the minimum of pressure for each cyclone within their area of responsibility. As ARPEGE does not locate well the tropical cyclones, this data is usually rejected during the assimilation cycle. But when we force the model to accept it, an intense tropical cyclone is taken into account though the forecasted cyclone track is not as good as we could have expected. So we intend to apply the two bogussing techniques simultaneously. We hope to improve significantly the skill of ARPEGE to forecast tropical cyclones tracks and intensities.
We shall also try to take advantage of the high resolution of ALADIN by running a short range forecast before the translation : that way we hope to get in the background a deeper low at the center of the cyclone before the next assimilation cycle.
Agricole Wills et BenAmadi Yahaya
Evaluation d'une nouvelle mithode d'assimilation de cyclones tropicaux faisant appel au couple Arpege etire-Aladin. Rapport de stage de recherche, Mastere de l'ENM, annee 1996/1997.
Goolaup Premchand et Felix Randrianavalona
Assimilation de cyclones tropicaux a l'aide des modeles Arpege etire-Aladin. Rapport de stage de recherche, Mastere de l'ENM, annie 1995/1996.
|
Home |