High and medium resolution multispectral satellite imagery is used to classify live post-fire forest residuals within two large forest fire affected areas (>100,000 ha) in the northern boreal forest of Alberta, Canada. Residual forest patches are converted to polygons to calculate patch and shape metrics at nine minimum mapping unit classes. Results are analyzed and compared within and between the two fires. Results indicate how sensor spatial resolution, choice of MMU and anthropogenic features (i.e. highways, transmission lines) affect post fire residual patch and shape level metrics.

The spectral angle mapper (SAM) classifier is used to classify forest cover on a highly industrialized mountain region in the central part of Honshu, Japan using Landsat TM 5 satellite imagery. Research findings relate land use and cover change (LUCC) processes on the mountain to image classification challenges. Findings illustrate the importance of calculating the spectral separability between forest cover types when developing a forest cover classification scheme in mountain regions where forest cover has been anthropogenically modified.