The algorithm, starting from techniques typical of image processing and information theory, is an instrument for the detection of anomalous variations in any physical parameter measured by remote sensing. Within a software system capable of checking the validity of such a measurement, comparing it with extant, historical data and reporting it, the algorithm acts as a filter able to amplify the anomaly. The aim of this work is to see if it possible to detect anomalous temperature variations in some way connected with fire at ground level by means of the sensors in the infrared heat band of the METEOSAT-2 satellite. The choice of this geostationary satellite was made because, despite low spatial resolution (at the latitude of Sardinia pixels are about 7 km square), it has good time resolution (one slot every 30 minutes).
