Mapping forest structure from above and below the canopy with UAV LiDAR and a Cygnus backpack scanner

In dense, complex forest a single LiDAR platform gives low accuracy and incomplete data. The local network signal was weak, and tree cover kept the RTK from fixing at times.

• The UAV LiDAR flew for about 11 minutes while the backpack scanner worked the same area in two scans totalling 30 minutes, one of them a repeat. An external radio shared a base station between the two systems.
• The backpack followed a snake-shaped path around the trees with no more than 10 m between scan lines, and scans were repeated while RTK was fixed to strengthen the trajectory.
• The two point clouds were merged, cut into blocks, resampled and denoised. Ground was classified, a DEM was made and the cloud was normalised.
• Single trees were segmented, with some errors corrected by hand, and species, position, height, trunk diameter and tree structure were extracted.

• The UAV LiDAR alone gave 508.79 points per m² and missed data below the canopy. The backpack scanner gave about 2,539.59 points per m². The merged cloud reached 4,701.48 points per m² and shows the complete structure of the trees.
• SatLab says the approach gives high-precision vertical forest structure and more accurate, cheaper extraction of single-tree measurements, but publishes no accuracy figures.

SatLab S1 UAV LiDAR, a Cygnus backpack SLAM system with RTK, SatLiDAR software and an external radio. The Cygnus models we supply today are the Cygnus 3 and the Cygnus Lite.

• Filling the gaps in airborne data around an expressway with the Cygnus SLAM scanner
• Handheld 3D mapping with a Livox LiDAR at the University of Hong Kong