Comparison between an Area-based and Individual Tree Detection Method for Low-pulse Density Als-based Forest Inventory

Airborne laser scanning (ALS)-based forest inventories are being increasingly used. The two main approaches in deriving forest information from small-footprint laser scanner data are the area-based and individual tree detection (ITD) methods. In the present study we test the accuracies of an area-based k-nearest neighbour (k-NN)-method and ITD with a practical low-pulse density (1.8/m) ALS dataset at the plot level. The research material consists of 333 treewise measured circular plots from southern Finland. A test dataset of 97 plots was selected for stand characteristic accuracy observation. The root-mean-squared errors (RMSEs) for basal area, total volume, mean height and mean diameter were with ITD 26.7%, 26.9%, 8.2% and 12.2% and with kNN 23.1%, 23.7%, 12.4% and 14.8%, respectively. The results obtained in this study demonstrated that both ALS-based inventory methods are practical, even with low-pulse density data. A Combination method could be developed to utilize the strengths of both methods and should be further investigated. * Corresponding author.