ABSTRACT
The separation of the individual contributions of aerosol and gases to the total attenuation of radiation through the atmosphere has been the subject of much scientific investigation since remote sensing experiments first began. We describe a new scheme to account for the spectral variation of the aerosol extinction in the inversion of transmission data from occultation measurements. Because the spectral variation of the aerosol extinction is generally unknown,the inversion problem is underdetermined and cannot be solved without a reduction in the number of unknowns in the set of equations used to describe the attenuation at each wavelength. This reduction can be accomplished by a variety of methods, including use of a priori information, the parameterization of the aerosol spectral attenuation, and the specification of the form of the aerosol size distribution. We have developed and implemented a parameterization scheme based on existing empirical and modeled information about the microphysical properties of aerosols. This scheme employs the eigenvectors from an extensive set of simulations to parameterize the aerosol extinction coefficient for incorporation into the inversion algorithm. We examine the accuracy of our method using data sets containing over 24,000 extinction spectra and compare it with that of another scheme that is currently implemented in the Polar Ozone and Aerosol Measurement (POAM) satellite experiment. In simulations using 80 wavelengths in the UV-visible-near-IR spectral range of the Stratospheric Aerosol and Gas Experiment III (SAGE) instrument, we show that, for our optimal parameterization, errors below 1% are observed in 80% of cases, whereas only approximately 20% of all cases are as accurate as this in a quadratic parameterization employing the logarithm of the wavelength.
