Enhancing LiDAR Data Positioning Accuracy in National Forest Surveys through Multi-Source Point Cloud Matching in Terrasolid software

Abstract. National Land Surveys (NLS) worldwide extensively utilize LiDAR (Light Detection and Ranging) technology for forest inventory, integrating airborne (ALS) and terrestrial/mobile (TLS/MLS) LiDAR to obtain detailed 3D forest structure data. Efficient multi-modal data co-registration is essential for applications such as biomass estimation, forest volume assessment, growth monitoring, and tree mapping. Given the vast scale of NLS projects, often covering thousands of kilometres, efficient data processing is crucial. TerraScan provides two fully automated methods for co-registering TLS/MLS and ALS datasets: (1) signal marker-based registration and (2) tree stem-based registration. These methods achieve an average planimetric RMSE of 1.3–4.8 cm, offering state-of-the-art registration accuracy. The methods have been tested for robustness against ALS resolution deterioration, maintaining statistically similar performance even when point density is reduced to 26 pts/m². Also, the ALS data from National Land Survey (NLS) of Finland with 5-8 pts/m² were tested and demonstrated the average co-registration RMSE comprising 7.5 cm. Optimized multi-threaded CPU processing enables rapid co-registration of massive datasets, making these methods highly suitable for large-scale national and global land surveys. Specifically, TerraScan tools enable the rapid co-registration of hundreds of millions of points within seconds.