Detecting manmade forest activities and natural disasters using landsat TM satellite data - A method presented for controlling continuously updated forest information in Finland

Methodology applying multitemporal Landsat TM satellite images was presented for detecting rapid changes in forest canopy. The accuracy of the method varied from 80 to 87 percent expressed in overall percentage of correct classification. With general change classes, more accurate results were obtained compared to accurate labeling of the reason for change. The method was proposed to be used for controlling the quality of continuously updated operational forest data base. It was estimated that the costs of maintaining the accuracy of forest management data would be about one third compared to the old updating method based on repetitive base line inventories.