NASA'S EXPERIENCE IN DERIVING TOTAL OZONE FROM SATELLITES

摘要

Our experience in deriving total ozone from 3 decades of satellite observations indicate that with some exceptions (noted below) total ozone can be estimated accurately using just two wavelengths. This simple method works primarily because logI_0, where I_0 is the radiance received at the satellite in absence of molecular absorption, has very nearly a linear relationship with wavelength. The method runs into problems when significant deviations from the linear relationship occur. This happens primarily in the presence of UV-absorbing aerosols (smoke, mineral dust, and volcanic ash), or when highly non-Lambertian surfaces such as water and ice are observed in a particular geometry. A 3-wavelength modification to the basic SBUV/TOMS algorithm is currently in development and shows promise of reducing much of the remaining errors. It should be noted, however, that at least one additional wavelength is needed to correct the satellite-derived total ozone for volcanic SO_2 (anthropogenic SO, is typically not a problem), and a wavelength around 310 nm is needed at high solar zenith angles to account for the rapidly increasing ozone profile sensitivity of the airmass factor. The algorithm we describe here is based on an approach that uses few wavelengths carefully selected from a broad spectral window to derive as many geophysical parameters as there are measurements and uses the residues at any additional wavelengths for diagnostics. This should be contrasted with the Differential Optical Absorption Spectroscopy (DOAS) method that is currently used to process GOME data, in which a large number of measurements in a narrow spectral window are used to retrieve just a few geophysical parameters. Understanding the pros and cons of these two fundamentally different measurement techniques for different applications remains an active area of research.