Ozone analysis and forecast from a high resolution satellite data assimilation

To improve the representation of the atmospheric ozone field given by some observations or a numerical model, one can combine these two sources of information by data assimilation techniques. The pixel size of the data coming from the instruments on board the latest generation of satellites is howev...

Full description

Bibliographic Details
Main Author: Pajot, Benjamin
Other Authors: Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique (CERFACS), Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique - CERFACS (CERFACS), Université Paul Sabatier - Toulouse III, Daniel Cariolle(daniel.cariolle@cerfacs.fr)
Format: Doctoral or Postdoctoral Thesis
Language:French
Published: HAL CCSD 2011
Subjects:
Online Access:https://theses.hal.science/tel-00713190
https://theses.hal.science/tel-00713190/document
https://theses.hal.science/tel-00713190/file/these.pdf
Description
Summary:To improve the representation of the atmospheric ozone field given by some observations or a numerical model, one can combine these two sources of information by data assimilation techniques. The pixel size of the data coming from the instruments on board the latest generation of satellites is however much lower than the horizontal resolution usually used within global Chemical-Transport Models (CTMs). These data bring information on small-scale structures of the ozone field that can only be represented in the model by increasing its resolution. In order to assimilate small-scale information within the Météo-France Mocage CTM with a high resolution horizontal grid while keeping reasonable execution cost, we have developed a spectral version of the Cerfacs Valentina data assimilation system.We use the external loop algorithm in conjunction with the 3D-Var with FGAT variational method. We show that this combination deteriorates the analyses for atmospheric situations governed by rapid transport. As a consequence we choose to use the variational 4D-Var method in terms of increment to perform high resolution ozone simulations. We use the Valentina system with the Mocage CTM both at low and high horizontal resolutions for the month of September 2008 during which small-scale structures are present in the ozone field at the vicinity of the Antarctic polar vortex. The total ozone columns from the IASI instrument have been assimilated because of their high spatial repartition. They have been averaged on the low and the high model grids giving two super-observation sets. We combine them with the profiles from the MLS instrument in order to constrain the vertical repartition of the information in the stratosphere and the upper troposphere during the data assimilation process. The model resolution increase improves the CTMs direct simulations and, to a lesser extent, the analyses obtained from the assimilation of the low resolution data set. However, the increase of the super-observation resolution deteriorates the analyses ...