Retrieval of vertical concentration profiles from OSIRIS UV-visible limb spectra

The OSIRIS instrument, launched on the Odin satellite in February 2001, includes an optical spectrograph that will record UV-visible spectra of sunlight scattered from the limb over a range of tangent heights. These spectra will be used to retrieve vertical profiles of ozone, NO2, OClO, BrO, NO3, O-2, and aerosols, for the investigation of both stratospheric and mesospheric processes, particularly those related to ozone chemistry. In this work, the retrieval of vertical profiles of trace-gas concentrations from OSIRIS limb-radiance spectra is described. A forward model has been developed to simulate these spectra, and it consists of a single-scattering radiative-transfer model with partial spherical geometry, trace-gas absorption, Mic scattering by stratospheric aerosols, a Lambertian surface contribution, and OSIRIS instrument response and noise. Number-density profiles have been retrieved by using optimal estimation (OE) to combine an a priori profile with the information from sets of synthetic "measurements". For ozone, OE has been applied both to limb radiances at one or more discrete wavelengths and to effective-column abundances retrieved over a broad spectral range using differential optical absorption spectroscopy (DOAS). The results suggest that, between 15 and 35 km, ozone number densities can be retrieved to 10% accuracy or better on 1 and 2 km grids and to 5% on a 5 km grid. The combined DOAS-OE approach has also been used to retrieve NO2 number densities, yielding 13% accuracy or better for altitudes from 18 to 36 km (in a 2 km grid. Differential optical absorption spectroscopy - optimal estimation retrievals of BrO and OClO reproduce the true profiles above 15 km in the noise-free case, but the quality of the retrievals is highly sensitive to noise on the simulated OSIRIS spectra because of the weak absorption of these two gases. The development of inversion methods for the retrieval of trace-gas concentrations from OSIRIS spectra is continuing, and a number of future improvements to the forward model and refinements of the retrieval algorithms are identified.