Abstract
The shortwave radiation field, i.e., in the solar spectral range, emerging at the top of the atmosphere is anisotropic due to the optical properties of the atmosphere and the reflectance characteristics of the underlying surface. Thus, anisotropy conversion factors are used to account for anisotropy in the derivation of flux densities from satellite measured broadband radiances. Uncertainties in the conversion factors due to uncertainties of the actual parameters of the atmosphere and the surface lead to errors in the derived net fluxes at the top and the bottom of the atmosphere and in the derived surface albedo. The errors for cloud-free situations over land, which this paper is only concerned with, are analyzed by simulation, using measured surface bidirectional reflectance functions and realistic values of the optical parameters of the atmosphere, including gas and aerosol absorption and multiple Battering. To derive surface albedo from planetary albedo, a linear relationship is commonly used. Improved values of its slope and intercept are presented, depending on solar zenith angle, turbidity and absorption.
The anisotropy conversion factors over land surfaces range from about 0.3 to 2 for low and 0.8 to 1.3 for high sun. To achieve an accuracy of 20 W m−2 in the net fluxes at the top of the atmosphere the uncertainty of the anisotropy conversion factors must be less than about 0.07 for vegetated and about 0.02 for surfaces with high albedo, decreasing with solar zenith angle. For the surface albedo derived from satellite measured radiances, the masking effect of the atmosphere gives additional errors. Thus, to achieve an accuracy of 20 W m−2 in the net fluxes at the surface the error due to anisotropy must be even lower. Global averages of anisotropy conversion factors have been derived from Nimbus-7 ERB data which are sufficiently accurate to be applied for global investigations or zonal averages. However, application of these factors to smaller areas, especially to surfaces with specific vegetation, will lead to larger uncertainties according to the deviation of the particular area from the global or zonal average. In such cases, regional anisotropy conversion factors, e.g., derived as described in Part II (Kriebel and Koepke) should be used.
