A large dataset of labelled single tree point clouds, QSMs and tree graphs.

High-resolution data of individual trees are critical for advancing forest monitoring, inventory development, and ecological research. This dataset, BioDiv-3DTrees, comprises 4,952 individual tree point clouds of 19 species, captured using Terrestrial Laser Scanning (TLS) and Unmanned Aerial Vehicle Laser Scanning (ULS), along with 3,386 Quantitative Structure Models (QSMs) and graph representations of the 14 broadleafed species in the dataset. The trees were sampled across the three research areas of the Biodiversity Exploratories in Germany. Each tree is linked to an existing open-access forest inventory dataset, which includes species identity, diameter at breast height (DBH), and tree height. The dataset is suitable for various research applications, including biomass estimation, algorithm development, tree structure analysis, and data fusion with traditional inventory methods. All QSMs were generated using TreeQSM 2.4.1 and have been validated for tree height, diameter at breast height and crown projection area against their underlying point clouds to ensure consistency. The dataset provides a reliable and scalable resource for forest science and remote sensing communities.