Compression of Meteosat data for image dissemination

Figure 1. Prototype dissemination system The availability of regular transmissions from the Meteosat weather satellite opens up the possibility of disseminating weather images to the general public on the public switched telephone network (PSTN), providing that the image data can be suitably compressed. Images presented in easily interpreted colour graphics format and updated every 30 minutes could be of interest to the tourist industry, for example, and would provide a 'nowcasting' service. Unfortunately, the use of satellite data alone limits the usefulness of any such service, since at present the (half-hourly) information is gathered only from the visible and thermal infrared bands, and it is at low spatial resolution (typically 5.5 x 10 km/pixel at UK latitudes). Even so, classification techniques can still extract useful information, and a classification algorithm suitable for a nowcasting system based solely upon Meteosat data is presented in this paper. Figure 1 shows the three nodes of an experimental image dissemination network. Every 30 minutes a Meteosat Primary Data User Station provides the receive node with raw visible and thermal infrared images of Europe, each quantized to 8 bit/pixel. The raw images are then formatted to 640 x 200 pixels so every third line can be repeated when generating a conventional 640 x 350 pixel EGA display.This process provides some compensation of the geometrical distortion arising from the low elevation angle of the geostationary satellite at UK latitudes. The formatted images are then compressed to generate a file 'weather.cmp' comprising two cloud classes