Monitoring fires in areas characterized by grazing pressure

Fire is widely recognized as a factor of deforestation and land degradation in the Mediterranean basin. Mediterranean vegetation is provided with different mechanisms for resisting fire; the limiting factor is constituted by fire frequency which can transform fire from ecological factor into destructive factor. The present paper deals with monitoring of affected surfaces in order inventory areas threatened by desertification because of the high fire frequency. Remote sensed data (Landsat TM) were used to monitor land cover changes due to forest fires in two test areas in Sardinia, an Italian region where fire has historically played an important role and is closely related to agropastoral activities. The application of standardized digital change detection techniques (multitemporal vegetation index difference, supervised classification on multitemporal bands or principal components extracted from a set of multitemporal TM bands) has shown a high classification accuracy in burnt areas inventorying (>93%). The application of simple digital change detection techniques to satellite data on different test areas provided high accuracy in burnt areas inventorying thus assuring an operational use of such techniques.