Verification of Stratospheric Ozone: News

You are here

26 Feb 2014: Major upgrade of the BASCOE NRT assimilation system

The BASCOE assimilation system used in NRT has been upgraded from version 03 to version 04 on February 01, 2014.

  • All previous operational versions used a diagonal Background Error Covariance Matrix (BECM). Version 4 uses a non-diagonal BECM where spatial correlations are based on an (isotropic) spectral representation as described in Errera and Ménard (ACP, 2012).
  • In the current configuration, these BECM spatial correlations are set up using a gaussian model, both in the horizontal (correlation length scale=600km) and in the vertical (correlation length scale=1 model level).
  • The processing of the following species has changed:
    • Ozone: the Aura-MLS observations above 3 hPa are now rejected. This allows a much better assimilation of the other species (especially H2O and HCl) in the upper stratosphere. The only drawback w.r.t. ozone analyses is seen between 3hPa and 1hPa above the winter pole, where the bias (underestimation by 24-h forecast against assimilated obs) increases from ~5% to ~15%.
    • CO: this species was not usable in any previous operational version as its only source in the middle atmosphere, i.e. the photolysis of CO2, had been forgotten from the chemical scheme. This has been corrected and since CO is retrieved from Aura-MLS, it has been added to the list of assimilated species. Caution: this product is very preliminary and in evaluation phase; it is based on the assimilation of Aura-MLS retrievals of CO which require caution themselves (according to the MLS Version 3 data quality document).
  • The results for all other species are similar or slightly improved. Slightly better results were noted for ozone in the tropical UTLS and in the polar lower stratosphere (thanks to the non-diagonal BECM) and for H2O, HCl in the upper stratosphere (thanks to the blacklisting of ozone observations above 3 hPa).