A retrieval-assimilation method has been developed as a quantitative means to exploit the information in satellite imagery regarding shading of the ground by clouds, as applied to mesoscale weather analysis. Cloud radiative parameters are retrieved from satellite visible image data and used, along with parameters computed by a numerical model, to control the model's computation of downward tadiative fluxes at the ground. These fluxes, in turn, influence the analysis of ground surface temperatures under clouds. The method is part of a satellite-model coupled four-dimensional analysis system that merges information from visible image data in cloudy areas with infrared sounder data in clear areas, where retrievals of surface temperatures and water vapor concentrations are assimilated.
The substantial impact of shading on boundary-layer development and mesoscale circulations was demonstrated in simulations, and the value of assimilating shading retrievals was demonstrated with a case study and with a simulated analysis that included the effects of several potential sources of error. The simulation results imply that assimilation is preferable to ignoring shading, even if the errors in the retrieval-assimilation process happen to compound each other. The case study was performed in the northwestern Texas area, where convective cloud development was influenced by the shading effects of a persistent region of stratiform cloud cover. Analyses that included shading retrieval assimilation had consistently smaller shelter-height temperature errors than analyses without shading retrievals. When clear-area surface temperature retrievals from sounder data were analyzed along with cloudy-area shading retrievals, the contrast in heating between the shaded and clear parts of the domain led to large variations in anallyzed boundary-layer depths and had a modest impact on analyzed wind flow. The analyzed locations of upward vertical motion corresponded roughly to areas of convective cloud development observed in satellite imagery, whereas analyses without shading assimilation lacked substantial vertical motions. Assimilation of water vapor information retrieved from sounder data was beneficial to the representation of water vapor in the analysis.