Forest Structure, Health and Regeneration Assessment Using Airborne Digital Camera Imagery

Integrated spectral and spatial analysis of high-resolution multispectral imagery provides a means to assess forest structural condition in a variety of applications. This paper describes a research program to develop methods for characterizing forest structural condition in assessment of damage due to anthropogenic and natural factors and in regeneration assessment. Three applications are presented: 1. development of a multivariate structural condition index representing windfall, and mortality due to acid mine contamination, 2. modelling forest canopy structural damage from a severe ice storm, and 3. automated measurement of conifer crop trees and competing vegetation in regeneration assessment. These applications integrate airborne digital camera image spectral information with spatial measures of texture, semivariance, image fractions, and spatial analysis of image fractions extracted from imagery of 0.25m – 1.0m pixels. Image spatial information is generally complementary to spectral information in derived models, improving their quality significantly over models using spectral information alone. Ice storm damage mapping at regional scales using Landsat data is also described. Regeneration assessment includes automated tree crown detection, delineation and measurement algorithms, as well as modelling of competition cover using digital camera imagery of 2.5cm – 10cm pixels. Initial tests in a research experiment where tree spacing and competition species were controlled resulted in tree detection omission errors of less than 6% and commission errors of less than 10%. Automated crown diameter measurements were within 17% of field-measured diameters taken at the base of the crown. When averaged over the study, this error dropped to 3%. Tests in an operational cutover under a range of competition levels are currently underway. Methods and results to-date of projects in each of these applications are presented.