GDP 5.0 – Improved Total Ozone Columns from GOME based on the GODFIT Algorithm
Michel Van Roozendael(1), D. Loyola(2), R.J.D. Spurr(3), C. Lerot(1), D. Balis(4), M. Koukouli(4), J-C. Lambert(1), J. van Gent(1), J. Granville(1), W. Zimmer(2), S. Otto(2) and A. Doicu(2)
(1) BIRA-IASB, 3, Avenue Circulaire, B-1180 Brussels, Belgium
(2) German Aerospace Center (DLR-IMF), PO Box 1116, D-82234 Wessling, Germany
(3) RT-Solutions, Inc., Cambridge, MA, United States
(4) Aristotle University of Thessaloniki, Laboratory of Atmospheric Physics, Thessaloniki, Greece
We present the last upgrade of the GOME level 1-to-2 operational algorithm: GDP version 5.0. GDP-5 implements the direct-fitting total ozone retrieval algorithm developed as part of the GODFIT project. The GODFIT algorithm is based on a non-linear least-squares minimization where simulated radiances and all Jacobians from the state vector parameters are calculated “on-the-fly” using, as forward model, the linearized scalar radiative transfer (RT) model LIDORT version 3.3. In contrast to previous versions of the GOME processor which all used a DOAS-type approach implementing single wavelength air mass factor calculations, a direct fit for the vertical column is here performed which involves a comparison between simulated and measured radiances in the full spectral interval from 325 to 335 nm. Ring effect is treated using pre-calculated correction spectra obtained using LIDORT-RRS, an off-line version of the RT code that includes Raman scattering processes. Ozone profiles are iteratively adjusted to the retrieved columns based on the recently developed neural network NNORSY climatology. Temperature profiles are adjusted as part of the retrieval process using an innovative and highly accurate temperature-shifting scheme. Likewise effective surface properties are derived using an internal albedo closure approach. Clouds are treated in the Lambertian approximation using the latest version of the OCRA/ROCCIN cloud retrieval algorithm. In comparison with the current GOME data product version 4.x, the major benefit of the new GDP-5 algorithm lie in the use of a physically sound multi-wavelength direct fitting approach which also enables the better treatment of atmospheric temperatures and surface effects, coupled to a more accurate cloud correction. Resulting improvements are illustrated based on various sensitivity analysis and selected validation results. GDP-5 is being integrated in the UPAS system used for the operational processing of GOME data. The official release of the GDP-5 product for the complete GOME mission is planned for early 2010.