General Approach for Forest Woody Debris Detection in Multi-Platform LiDAR Data

Abstract

Woody debris (WD) is an important element in forest ecosystems. It provides critical habitats for plants, animals, and insects. It is also a source of fuel contributing to fire propagation and sometimes leads to catastrophic wildfires. WD inventory is usually conducted through field surveys using transects and sample plots. Light Detection and Ranging (LiDAR) point clouds are emerging as a valuable source for the development of comprehensive WD detection strategies. Results from previous LiDAR-based WD detection approaches are promising. However, there is no general strategy for handling point clouds acquired by different platforms with varying characteristics such as the pulse repetition rate and sensor-to-object distance in natural forests. This research proposes a general and adaptive morphological WD detection strategy that requires only a few intuitive thresholds, making it suitable for multi-platform LiDAR datasets in both plantation and natural forests. The conceptual basis of the strategy is that WD LiDAR points exhibit non-planar characteristics and a distinct intensity and comprise clusters that exceed a minimum size. The developed strategy was tested using leaf-off point clouds acquired by Geiger-mode airborne, uncrewed aerial vehicle (UAV), and backpack LiDAR systems. The results show that using the intensity data did not provide a noticeable improvement in the WD detection results. Quantitatively, the approach achieved an average recall of 0.83, indicating a low rate of omission errors. Datasets with a higher point density (i.e., from UAV and backpack LiDAR) showed better performance. As for the precision evaluation metric, it ranged from 0.40 to 0.85. The precision depends on commission errors introduced by bushes and undergrowth.