Files
Abstract
LiDAR data are used to detect hemlock (Tsuga canadensis) defoliation due to hemlock woolly adelgid (Adelges tsugae) (HWA) in the Chattahoochee National Forest (CNF) located in Georgia. LiDAR data are used to quantify leaf area index (LAI) and fractional cover (fCover). Traditional field methods are used to validate LiDAR results with hemispherical photography of stands in varying stages of decline. Single linear regression results suggest that LAI (R2 = 0.4355) and fCover (R2 = 0.4597) are not well predicted from LiDAR. Multivariate principal component analysis using LiDAR variables and multivariate stepwise regression with ground data improve results for LAI (R2 = 0.7307) and fCover (R2 = 0.4666). Multivariate cluster analysis finds a significant relationship between ground and LiDAR derived clusters of high, medium, and low defoliation with 75% cluster agreement in classification of defoliation. This suggests that 3 health status ranks can be created from remote data.