Uncertainties in radiative transfer computations: consequences on the MERIS products over land

Abstract

The main objectives of MERIS (MEdium Resolution Imaging Spectrometer) consist of atmospheric processes related to the water vapour column and aerosol optical properties designed for meteorological applications, and the land surface properties as well as the bio‐optical oceanography. In this context, operational MERIS level‐2 processing uses auxiliary data generated by two radiative transfer tools. These two codes simulate upwelling radiances within a coupled ‘atmosphere–land’ system, using different approaches based on the matrix‐operator method (FUB, Freie Universität Berlin), the discrete ordinate method and the successive orders technique (ULCO, Université du Littoral Côte d'Opale). Intervalidation of these two radiative transfer tools was performed in order to implement them in the MERIS level‐2 processing. For cases without gaseous absorption, the scattering processes both by the molecules and the aerosols were retrieved within a few tenths of a percentage point. Nevertheless, some substantial discrepancies occur if the polarization is not accounted for, mainly in the Rayleigh scattering computations. Errors on the aerosol optical thickness reach up to 25% in some geometries as observed in the MERIS images. The parametrization of gaseous absorption (H₂O and O₂) defined for each of these two codes leads to a good agreement for the MERIS bands with residual absorption. In the strong absorption bands (761.75 nm and 900 nm), the FUB computations well match the results derived from a line‐by‐line (LBL) code with a very high spectral resolution. Note that the oxygen absorption at 761.75 nm is very sensitive to the characteristics of the sensor spectral response and requires accurate calculations with the LBL code. Consequently, the ULCO code has been implemented in the MERIS level‐2 processing to include polarization in the scattering processes and to correct for slightly gaseous absorption, the FUB code to derive the water vapour abundance, and the LBL code to determine the barometric pressure. Impacts of the differences in the look‐up table generation on the level‐2 products (aerosol model, surface reflectance and barometric pressure) are also analysed and illustrated.

Acknowledgments

This work has been financially supported by the European Space Agency (ESA) under the contract 14558/00/NL/DC. We first thank Jean Paul Huot and Steven Delwart from ESA for their useful comments concerning this work. The authors would like specially to thank Eric Dilligeard from ULCO, Peter Albert, Frank Fell and René Preusker from FUB, for their assistance in the use of the RTC/ULCO (GAME and SO) and RTC/FUB (MOMO), respectively. We are also grateful to Ginette Aubertin from ABB‐BOMEM for her technical help.