Retrieval of optical depth and vertical distribution of atmospheric aerosols from light intensity and polarization in O2 A and B bands

Abstract:

Although UV radiance at the top-of-atmosphere (TOA) is known to be sensitive to the centroid height of absorbing aerosols, several methods were proposed recently to retrieve centroid height of aerosols (in particular, scattering aerosols) from hyperspectral radiation measurements in and around the O₂ A and B bands. However, most such retrievals ignore polarization effects in the reflected radiation and are limited to dark surface scenarios. Here, we evaluate the feasibility of combining intensity and polarization measurements in the O₂ A and B bands for retrieving aerosol optical depth and aerosol height, especially over bright surfaces.

The feasibility study is conducted in two steps. The first step comprises a theoretical analysis of the effects of various aerosols on the behavior of the degree of linear polarization (DOLP) of light in the O₂ A and B bands under different surface conditions. We use the Unified Linearized Vector Radiative Transfer Model (UNL-VRTM) to simulate the full Stokes vector and its Jacobians with respect to the aerosol optical depth and parameters characterizing the aerosol vertical profile. In order to identify the most sensitive spectral wavelengths, the concepts of ‘Shannon information content (SIC)’ and ‘degree of freedom for signal (DFS)’ are used to examine the information contained in each individual wavelength in both O₂ bands. In the second step, case studies are conducted to retrieve the optical depth and vertical distribution of aerosols. Here, we use GOME-2 and/or SCIAMACHY backscatter measurements at selected most sensitive wavelengths in both O₂ bands. The retrieval method is based upon non-linear optimal estimation theory. We compare the retrieved aerosol parameters (optical depth and height of aerosols) directly with CALIPSO observations. We also estimate the error of the retrieved parameters due to various sources of uncertainty.