Towards the Estimation of Tree Structural Class in Northwest Coastal Forests Using Lidar Remote Sensing

The amount and variability of dead wood in a forest stand is an important indicator of forest biodiversity, and relates to both the structural heterogeneity and the amount of habitat available for biota. In this study, we investigate the capacity of light detection and ranging (lidar) technology to estimate the percentage of dead trees in coastal forests on Vancouver Island, British Columbia, Canada. Twenty-two field plots were established from which the tree structural classes, or wildlife tree (WT) classes, of all stems (DBH > 10 cm) were estimated. For each plot, the frequency distributions of the WT classes were highly skewed, so lognormal distributions were fitted, and the means (μ) and standard deviations (σ) of the log-transformed data were extracted. The relationship between μ and the percentage of dead trees within the plots was highly significant (r = 0.77, p < 0.001). A variety of metrics were extracted from the lidar vegetation returns and compared against μ, and results indicated that the natural logarithm of the coefficient of variation was the best predictor (r = 0.75, p < 0.001), followed by the heights of the 20 percentile (r = 0.69, p < 0.001). In general, results indicated that the lowest lidar height percentiles were more significant predictors of μ, which is likely based on the direct linkage between the number of dead trees in a stand and its canopy architecture. * Corresponding author