11.3 Improving the Accuracy, Long-Term Consistency, and Speed of the SAO OMI Ozone Profile Product

Wednesday, 15 January 2020: 3:30 PM
206B (Boston Convention and Exhibition Center)
X. Liu, Harvard-Smithsonian Center for Astrophysics, Cambridge, MA; and J. Bak, C. R. Nowlan, G. Gonzalez Abad, C. Chan Miller, K. Yang, R. J. D. Spurr, G. Huang, K. Sun, and K. Chance

The Smithsonian Astrophysical Observatory (SAO) ozone profile and tropospheric ozone retrieval algorithm has been implemented to produce the OMI ozone profile product (OMPROFOZ) data from October 2004 to present. This presentation will introduce the updates to this algorithm for the next version. We improve a priori O3 profile using a tropopause-based climatology, which better constrains ozone profile retrievals in the upper troposphere and lower stratosphere. Instead of a normal Gaussian, a super Gaussian is employed to better represent slit functions from OMI irradiances. Moreover, slit function differences between radiance and irradiance are taken into account as pseudo absorbers in the spectral fitting. Meteorological data of temperature profiles, surface and tropopause pressure are taken from GEOS-5 FP-IT assimilation products mapped onto OMI footprints. Moving average (31 day) of OMI irradiance spectra instead of multi-year average spectra are used to cancel out the degradation between radiances and irradiances. The empirical soft calibration is re-derived yearly to be consistent with the updates using the mean differences between measured and simulated radiances derived from one week of zonal mean MLS data in the tropics. Use of time-dependent empirical soft spectra and irradiance spectra helps improve the long-term consistency of the ozone profile retrievals. A final set of common mode correction spectra are derived from remaining fitting residual spectra after soft calibration as a function of solar zenith angle in every month for 2006, and are included as a pseudo absorber during fitting. This significantly reduces the discrepancy of ozone retrieval accuracy between lower and higher solar zenith angles. To speed up the on-line radiative transfer calculations, the Principle Component Analysis (PCA) based VLIDORT is implemented with a fast, low accuracy configuration, and the PCA-VLIDORT is combined with look-up table corrections for errors in radiances and Jacobians arising from performing scalar calculation, using lower streams, and coarser vertical grid. Validation with ozonesonde measurements shows significant improvement in retrieval accuracy and long-term consistency.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner