Adapting the Total Ozone Algorithm GODFIT to high Spatial Resolution Push-Broom Instruments
Danckaert, Thomas1; Lerot, Christophe1; Van Roozendael, Michel1; van Gent, Jeroen1; Spurr, Robert2; Loyola, Diego3
1BIRA-IASB, BELGIUM; 2RT Solutions, Inc., UNITED STATES; 3DLR, GERMANY
Within the ESA Ozone Climate Change Initiative project (Ozone_cci), an improved multi-sensor total ozone climate record has recently been generated applying the GODFIT (GOME-type Direct FITting) algorithm to the radiances measured by the European sensors GOME, SCIAMACHY and GOME-2. In order to further extend this record, while keeping a high level of consistency, it is desirable to apply the same algorithm to the data obtained with future instruments such as S5-P/Tropomi. Due to the computationally intensive nature of the GODFIT algorithm, solutions need to be developed to significantly accelerate the retrievals in order to allow for the treatment of very large data volumes.
GODFIT uses a direct-fitting approach where backscattered spectral radiances are simulated on-the-fly at about 100 wavelengths from 325 to 335 nm using radiative transfer tools from the suite LIDORT. The simulation of these spectral radiances represents the most significant part of the processing time. As an alternative approach, we have developed a version of the algorithm which interpolates these radiances, and the associated Jacobians required for the inversion process, from precomputed multi-dimensional lookup tables. Applying this algorithm to OMI L1 data using lookup tables of varying granularity, we study the impact of this approach on accuracy and performance. We also present a few technical details on the implementation of such lookup tables, for which the size may be in the range of several 100 GB for the most fine-grained versions.