algorithm devoted to early fire detection by using SEVIRI/MSG images

Forest fires represent the main cause of forest degradation in Italy and in the Mediterranean area countries. This phenomenon, progressively increasing, reached an average of 11000 fires per year in the period 1990-2000 with the destruction of 553,000 ha of vegetated areas in the Italian peninsula. In previous papers the capability to detect fires much smaller than the pixel size (4.0 km at middle latitude) using the SEVIRI (Spinning Enhanced Visible and Infrared Imager) sensor on board of the geostationary satellite MSG (Meteosat Second Generation) has been demonstrated. Due to the reduced spatial resolution of this sensor its capability to provide a prompt fire detection suitable to support operational firefighting activities is still under study. The Centro di Ricerca Progetto San Marco (CRPSM) is acquiring SEVIRI data since 2003 and has developed an innovative technique (SFIDE, System for Fires Detection), based on the exploitation of both the classical MIR and TIR spectral bands and the high observation frequency (15 min), to counteract the low spatial resolution characterizing such a sensor. The present paper aims to present the results obtained applying this technique and, at the same time, to confirm the applicability of the SEVIRI sensor as an instrument suitable to be employed in an operational system of early fire detection. For this purpose, an automatic system of early fire detection, based on the SEVIRI/MSG data, has been developed and tested in the Sardinian region during the month of August 2006. The validation of the results comprises the promptness of the detections (compared with the common ground based warnings), the errors of the geo-location and the accuracy of the sizes estimate of the hot-spots. The assessment of the performances of the system has been obtained mainly comparing its results with those obtainable from higher resolution sun-synchronous sensors dat