Fire detection and growth monitoring using a multitemporal technique on AVHRR mid-infrared and thermal channels

A technique has been developed to detect and map fire growth by using a multitemporal multispectral analysis of AVHRR images. The application of this technique allows fire growth monitoring. The technique is mainly based on comparison of brightness temperature differences and AVHRR Channels 3 and 4 for successive images. It provides a rapid evaluation of the area affected by the fire but suppresses the ambiguous thermal signal produced by nonvegetated terrain. It has been applied to a set of diurnal images acquired by the Granada University AVHRR reception system during a severe fire outbreak in Spain (July 1994). The Andalusian Regional Government Environmental Directorate provided the information and boundaries of the fires which are used as “ground truth” to validate the AVHRR-derived fires boundaries. Both the derived boundaries for the burned area and the fire evolution agree closely with the information collected by field observers, with differences ranging from 9% to 20% in the total burnt area. The availability of 24-h field reports for one of the fires provided an appropriate test for the continuous monitoring by means of remote sensing technique. The study area, located in southeastern Spain, is characterized by a dry mediterranean climate, and the fires affected different ecosystems, from coniferous forest to shrubland. This variety of land coverage makes difficult the application of other proposed techniques.