We present image analysis and processing tools that can be useful in the extraction of stable and reproducible input data from satellite images for meso-scale meteorological systems. The tools used consist of modern geometric machines capable in quantifying Volterra processes to which the modelled and observed physical fields are subjected, and filter operations with respect to and coupled to those Volterra processes. In particular we apply these tools to satellite images in the visual, infrared and water vapour spectrum in order to retain segmentations and quantifications of correlated physical fields which in turn can be used in air-pollution forecast and simulation systems. These physical fields concern the heat currents due to reflected solar irradiance in the short- and long-wave band. In this context the analysed and processed radiances serve the estimation of the short- and long-wave band transmission function.