Automated estimation of localized forest volume from large-scale aerial photographs and ancillary cartographic information in a boreal forest

Abstract

The development of an automated method for obtaining locally reliable estimates of forest volume is demonstrated for a mixed-species boreal forest of the Lac St. Jean region of Quebec. The method relies on the ability of an algorithm based on local maxima to identify individual stems from a scanned aerial photograph under the assumption that the points of maximum light reflectance will be the highest points on individual trees. This information is linked via regression analysis to mean heights of dominant and co-dominant trees and ground-based forest inventory data to provide a statistical relationship with forest volume. It was demonstrated that, by using the method, the local uncertainty of volume estimates could be decreased by 61% relative to standard forest inventory procedures. The method is not applicable to young or disturbed stands. The greatest difficulty with the method is that sample plots used for validation must be locatable with absolute accuracy on the scanned aerial photographs something that is likely to be problematic in many forest conditions.