Sacada-Logo Contribution of RIU

Current status (Oktober 2002) as presented at the AFO 2000 meeting at Schliersee (HTML).
Current status (March 2004) as presented at the AFO 2000 meeting at Bad Tölz (PDF, 1.1 MB).
Poster presented at the COSPAR meeting 2004, Paris (PDF, 1.6 MB).
Article published in Proceedings of the ENVISAT Symposium 2004, Salzburg (PDF, 3.3 MB).
Poster presented at the EGU General Assembly 2005, Vienna (PDF, 1.7 MB).

The Rhenish Institute for Environmental Research (RIU) is developping the software kernel of the two data assimilation algorithms needed to comply with the SACADA objectives. These algorithms are the four dimensional variational data assimilation methods (4D-VAR) and the four dimensional Physical-space Statistical Analysis System (4D-PSAS) or Pondriagin optimization. The latter technique will be extended to take into account the model errors. These methods are able to combine the following sources of information in a mathematically and chemically consistent way:

  1. Satellite observations in nadir and limb mode scattered in space and time within an analysis interval
  2. The knowledge encoded in the model formulation, as given by the chemical mechanism and the transport and diffusion modules,
  3. A priori knowledge taken from preceding forecasts, climatologies or sufficiently strong correlations, as for examples given by the ozon potential vorticity correlation in the lower extratropical stratosphere.

4D-VAR and 4D-PSAS offer, on the basis of well defined assumptions, an optimal exploitation of the information contents of available of chemical constituents. Both algorithms require the development of modules, which are adjoint and tangent linear to the chemistry-transport model. Coding of the adjoint and tangent linear model components and putting together its modules, the inclusion of minimization routines and the covariance matrices are the principal contributions of RIU, followed by extended testing based on case studies with augmented data sets. To this end, measurements from GOME (ERS-2), CRISTA 1 and 2 and SCIAMACHY, MIPAS, GOMOS, AATSR and MERIS (ENVISAT) sensors as well as other available data from NDSC and measurement campaigns will be used. Systematic model errors and their covariances as identified by DFD will be introduced in a statistically well weighted way. The PSAS-algorithm is selected as the optimization procedure is performed in the observation space, which is lower dimensional than the grid- point space.

A further issue is the objective to quantify the quality standards impacted by simplifications in the operational version, which will be introduced to make daily operational analysis feasible in near real-time. This task will be handled by the development of a significantly more complex model version (reference algorithm) which allows for a critical evaluation of the operative version. In combination with this, special observations retrieved from non operational data from MIPAS and SCIAMACHY will serve as an ideal opportunity for quality control of the analysis products.

Principal Investigator:

Hendrik Elbern

Contact:

Hendrik Elbern
Rheinisches Institut für Umweltforschung
Projekt EURAD
Aachener Straße 201-209
50931 Köln

This project is funded by:

AFO2000 BMBF