| description abstract | Earth radiation budget measurements, important to climate monitoring and to validating climate models, require that radiances measured by satellite instruments be converted to hemispherical flux. This paper examines that problem theoretically, using inhomogeneous cloud models constructed from Landsat scenes of marine boundary layer clouds. The spherical harmonics discrete ordinates method (SHDOM) code is applied to the model scenes to compute full two-dimensional radiation fields, which then simulate measured radiances. Inversion to flux is performed by several different methods, including plane-parallel table lookup and empirical angular distribution models with three different ways of determining scene identification, to examine error sources and relative magnitudes. Using a simple plane-parallel table lookup results in unacceptably large flux bias errors of 11%?60%, depending on the orbital viewing geometry. This bias can be substantially reduced, to no more than 6%, by using empirical angular distribution models. Further improvement, to no more than 2% flux bias error, is obtained if known biases in optical-depth retrievals are taken into account when building the angular models. Last, the bias can be further reduced to a fraction of a percent using scene identification based on multiple views of the same area. There are limits, however, to the reduction in the instantaneous error with this approach. Trends in the flux error are also identified, in particular an equator-to-pole trend in the flux bias. Given the importance of satellite measurements for determining heat transport from equator to pole, this consistent bias should be kept in mind, and efforts should be made to reduce it in the future. | |
