Spatial statistical techniques for aggregating point objects extracted from high spatial resolution imagery
Date
2001
Authors
Nelson, Trisalyn
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
At present most forest management is based on stand units; however, detailed information on individual trees is increasingly required for accurate forest management and decision making. Developing methods to collect individual tree information for medium and large areas will improve the quality of forestry information. A further improvement may come from deriving forest stands based on individual tree characteristics. If stands are derived their composition is known. In this thesis, research efforts focus on the development of a technique, which uses individual tree information to generate forests polygons which are homogenous in age. The goal of this research is to extract points from a lm spatial resolution IKONOS image and apply spatial statistics to aggregate points into polygons based on forest structure.
Using an existing local maximum technique, on an IKONOS 1m panchromatic image covering a range of forest ages, points representing individual trees were extracted. Point based spatial statistics were applied to the local maximum points and metrics representing forest structure generated. The effectiveness of nearest neighbour distance statistics and Voronoi point pattern analysis techniques for generating forest structure attributes were examined. A metric generated using nearest neighbour distances was used to aggregate the points into polygons related to forest structure. The forest structure polygons show a meaningful association with a series of age classes used by the British Columbia Ministry of Forest. The metrics show potential for identifying forest areas approximately homogenous in age, which may be used as a basis for polygon development.