9617555 McGraw This research tests the feasibility of censusing forest tree populations using remote sensing for the broad purposes of inventory, projection of future population levels, and screening the forest for rare, endangered or otherwise unique individuals ('bioprospecting'). High resolution color and color infrared photographic images of mapped, forested plots will be collected with 35mm cameras in fifteen overflights for a period of one year. Methods will be developed to automate the process of circumscribing individual tree canopies from scanned images. New image processing approaches to identify individuals to species will be developed which rely on differential phenological development of species and differing spectral reflectance properties. The ultimate goal of these techniques will be the censusing of populations of trees from remote imagery in a manner suitable for input to size-dependent matrix population models of forest tree population dynamics. In the future, satellite-borne sensors will likely be capable of both high spatial and spectral resolution image capturing. Population-level applications of remote-sensing are in their infancy. If data from such sources can be used for forest inventory, population projection and 'bioprospecting', the studies made possible will have regional and global applicability to numerous serious environmental problems.
