Development of a forest health index using high-resolution remotely-sensed multispectral digital imagery.

摘要

This research was conducted at an abandoned acid mine site near Timmins, Ontario, to determine if forest health is affected by adjacent tailings and to determine if the variation in forest health can be modeled with high-resolution multi-spectral digital imagery. Eighty-four plots were sampled to obtain sets of soil, forest structure and image variables. The objectives of this thesis were (1) to examine the spatial distribution of soil, forest structure and image variables to determine the integrated effect of mine site stress conditions on forest health as a function of distance from the contaminant source, (2) to conduct a multivariate analysis between the sets of variables to determine (a) if changes in forest structure are dependent upon soil properties, and (b) if image variables are related to the observed variation in forest structure and (3) to develop an image-based forest health index from the relations between forest structure and image variables. Significant trends with increasing distance from the tailings boundary existed for soil calcium, magnesium and iron content, all forest structure variables, and some spectral and textural image variables, thus demonstrating the effects of the mine site on the adjacent study area. Relations between the sets revealed changes in forest structure associated with certain soil properties, most notably increased blowdown was associated with coarse-textured soils. From relations between forest structure and image variables, an image-based forest health index was developed that explained 34% of the variance in three health-related forest structure components.