ROSE Vertically Resolved Global Ozone Distribution


ROSE/DLR is a global 3D-chemistry-transport model used to derive vertically resolved ozone distributions from ERS2-GOME total column ozone and ENVISAT ozone profile measurements. ROSE/DLR is an adapted version of the NCAR-ROSE model. A ROSE model description (including available vertical, horizontal and temporal resolution and differences between ROSE/DLR and the original NCAR-ROSE) can be found here.

Assimilation into ROSE

Observations are sequentially assimilated using an optimal interpolation scheme (OI). For GOME data anew profile is estimated by applying a statistical regression model. It uses potential vorticity (PV) and ozone column information to derive a reference ozone profile. Model first-guess and stochastic results are then combined by a Bayesian "Maximum A Posteriori" estimate (MAP). For ENVISAT observations (2003 and later) only OI will be used to assimilate ozone profiles.

Level 2 data of total column ozone are produced at the German Remote Sensing Data Center of the German Aerospace Center (DLR-DFD) operationally from GOME measurements. Due to the orbital parameters and scanning geometry of GOME all measurements are distributed inhomogeneously in time and space. Global coverage is achieved within three days. GOME total ozone columns are assimilated sequentially into the ROSE model using optimal interpolation with prescribed errors for GOME (4%) and model column values (18%). Thus the models horizontal and vertical ozone distribution changes characteristically according to column observations. Up to March, 2003 an additional bias correction was applied based on a SAGEII climatology. Extensive validation of results for 1996 using ground based and satellite instruments shows a very small model bias in general below 6%.


Results of several comparisons of ROSE derived ozone profiles with measurements from the CRISTA instruments, DWD - Hohenpeissenberg ozone soundings and satellite instruments can be found here.


HDF formated data files are available with 6h resolution. Download here.

Known data problems:

  • Current and historical GOME instrument status information can be found here .
  • Due to an instrument failure data from 17. January 2001 to 8. February 2001 is missing.

Please note: All results are preliminary and result of a development phase.
They are therefore generated with different DLR versions of the ROSE model.

Last up-date March 2003: optimal profile estimate using PV information

15. February 2002
11. January 2002
05. December 2000
23. November 2000
16. November 2000
11. August 2000
off-line SAGE-II bias correction
OI data assimilation
ROSE/DLR version 2.4
ROSE/DLR version 2.3
ROSE/DLR version 2.2 .
ROSE/DLR version 2.1.

Quality control and re-processing (if needed) is done every few days. If any unexpected features occur in the data please mail to F. Baier.

GOME Product Survey

The GOME Product Survey gives an overview on data products derived from the Global Ozone Monitoring Experiment, GOME, at DLR-DFD. Data products are briefly described and examples for application are given. Different ways to get access to the data products are presented. However, for a much more detailed description of data products and processing of higher level data products please refer to the GOME product guide.

Related Links

At DLR a neural network is also used to derive near realtime vertical ozone profiles directly from GOME radiation measurements.

The GOME-ROSE NRT service was set-up by collaboration with the following partners: NCAR and MPI (AFO2000 project INVERT) and EURAD (AFO2000 project SACADA). We are grateful to the BADC data center providing daily UK Met-Office meteorological analysis.


The GOME-ROSE service is amply supported by the Federal Ministry of Education and Research (BMBF) and the European Comission.


Dr. Frank Baier