A linearized discrete ordinate radiative transfer model for atmospheric remote-sensing retrieval

Abstract

The radiative transfer forward model simulation of intensities and associated parameter derivatives (weighting functions) is a vital part of the retrieval of earth atmospheric constituent information from measurements of backscattered light. The discrete ordinate method is the most commonly used approach for the determination of solutions to the radiative transfer equation. In this paper, we carry out an internal perturbation analysis of the complete discrete ordinate solution in a plane parallel multi-layered multiple-scattering atmosphere. Perturbations in layer atmospheric quantities will translate into small changes in the single-scatter albedos and optical depth values. In addition, we consider perturbations in layer thermal emission source terms and in the surface albedo. It is shown that the solution of the boundary value problem for the perturbed intensity field leads in a natural way to the weighting function associated with the parameter causing the perturbation. We have developed a numerical model LIDORT (linearized discrete ordinate radiative transfer) for the simultaneous generation of backscatter intensities and weighting function output at arbitrary elevation angles, for a user-defined set of atmospheric variations. Results for a 5-layer test atmosphere with two scatterers and thermal emission terms are shown. Intensities are validated against benchmark discrete ordinate results, while weighting functions are checked for consistency against finite difference results based on external perturbations. A second example is presented for a 60-layer terrestrial atmosphere with molecular and aerosol scattering and ozone trace gas absorption in the UV spectral range; weighting functions are shown to correspond closely with results derived from another radiative transfer model.