Estimation of forest parameters by using remote sensing single-tree detection and field plots with tree positions

Template matching, originally developed in Canada by Richard Pollock, is a method for single tree detection in high resolution (pixel size 0.3 m – 1 m) aerial digital imagery. The method was tested for Scandinavian conditions and further developed. It is a potential method to be included in a concept for estimating forest parameters from aerial imagery, such as stem number, stem volume, and tree biomass. Template matching gives estimates of position, crown diameter and crown shape for most trees in a stand. The concept proposed in this article includes a quick field method for measuring tree positions on field plots, which currently is under development at SLU. Another vital part is a method for rendering templates of whole field plots (with radius about 10 m), whereby the field plots can be exactly located in the aerial image. Then the vector of field-surveyed forest parameters can be related to the vector of image derived features for the same plot. If the image is divided into a grid, with approximately the same area as the field plots, the relations between image derived tree features and field plot data could be used for estimating forest variables for all grid cells, or as correlated information in a sampling scheme. In addition, maps of all trees visible from above can be created