Forest variables from LiDAR: Drone flight parameters impact the detection of tree stems and diameter estimates

Abstract

Ecosystem services provided by central European forests, often dominated by Norway spruce or Scots pine, are increasingly threatened by climate change. Monitoring, while labour intensive, is key to ensure continuing forest health. Consequently, UAV-based LiDAR remote sensing has become a valuable tool. However, the impact of drone flight parameters on LiDAR data quality has not yet been extensively studied. To address this, we first present a methodology for delineating tree stems, estimating their diameter at breast height (DBH), and separating understory vegetation from stems and old-grown trees to subsequently compare the approach to other existing methods. Second, we analyse how drone flight parameters influence the accuracy of forest parameter detection. Our methodology outperformed existing approaches in stem detection and DBH estimation. Understory detection enabled the identification of forest paths, roads, and areas without understory vegetation. Differences in flight parameters had a large effect on the accuracy of the approach. Optimal data usability was achieved by flying the drone at low flight height above the trees, at relatively high speeds, and with high LiDAR stripe overlap, balancing detailed data collection with efficient area coverage. We conclude that the new approach can provide foresters with detailed insights into forest structure and dynamics, reducing the need for extensive fieldwork.