Canadian Forest Service Publications
Le progiciel expert en foresterie - étude du lac Traverse. 1993. Gougeon, F.A. Forêts Canada, Institut forestier national de Petawawa, Chalk River, Ontario. Rapport d’information PI-X-108F. 17 p.
Available from: Pacific Forestry Centre
Catalog ID: 4369
Series: Information report (Petawawa)
The detection of ongoing changes in forested areas monitored with satellite images is traditionally accomplished with multi-date band rationings, enhancements, and classifications. The Forestry Expert Package (FEP) uses a completely different method and is an example of a new philosophy in image analysis: an artificial intelligence approach. It uses single-date Landsat classifications as input data, compares them, and decides on the significance and cause of detected changes based on rules derived from forestry knowledge.
For the present experiment, classifications of geometrically corrected images of different years and different seasons were obtained with an unsupervised classifier for a test area near Lake Traverse in Algonquin Park, Ontario. Several tests were run to determine if the FEP would detect logging activities and the subsequent regeneration known to have occurred in this area. A secondary objective was to see if any improvement in the classification of forest types of the non-changing areas could be achieved over conventional single data classifications.
The FEP was found to be good at detecting changes in the forest due to logging, and moderately good at monitoring the regrowth of conifers. However, the FEP is not yet at a stage where it could be of use as a forest management decision-making tool.
Also available under the title:
A forestry expert package - The Lake Traverse study (English)