Mobile laser scanning point clouds, quantitative structure models and destructive-reference attributes for 98 hardwood trees in Québec, Canada

Description This dataset contains tree-level mobile laser scanning (MLS) point clouds, quantitative structure models (QSMs), and field and destructive-reference attributes for 98 hardwood trees sampled from eight 0.25 ha forest plots in Duchesnay (DUC) and Sainte-Véronique (SVE), Québec, Canada. The sample includes 61 sugar maple (Acer saccharum Marshall) and 37 yellow birch (Betula alleghaniensis Britton) trees. Of these, 58 trees were sampled at DUC and 40 at SVE. The dataset was assembled to validate an automated MLS-QSM workflow implemented in the R package arbor for estimating DBH, total height, and compartment-level tree volume. It includes the individual-tree point clouds used in the validation, the full reconstructed woody QSM (Vtot) for each tree, a table linking field measurements, destructive-reference volumes, allometric estimates, and MLS-QSM-derived attributes, an R script for reproducing the MLS-QSM results, plot-level segmented point clouds and digital terrain models (DTMs) for all eight plots. Scope: this release contains only the 98 validation trees and their derived products, together with automatically segmented plot-level point clouds, and corresponding DTMs for all eight plots. Original unsegmented plot-scale MLS point clouds, trajectories, plot boundaries, and raw FARO acquisitions are not included in this repository but may be made available upon reasonable request to the corresponding author. Spatial and temporal coverage · Study areas: Duchesnay (46.901° N, 71.656° W) and Sainte-Véronique (46.594° N, 74.976° W), Québec, Canada · Sainte-Véronique MLS acquisition and destructive sampling: October 2024 · Duchesnay MLS acquisition: May 2025; destructive sampling: October 2025 · Forest type: uneven-aged temperate hardwood forest dominated by sugar maple and yellow birch Dataset contents Data archives · 01_Sample_trees.zip: 98 .laz files containing one segmented MLS point cloud per validation tree. · 02_QSM.zip: 98 .qsm files containing the full reconstructed woody QSM (Vtot) for each validation tree. · 03_Tree_attributes.zip: Tree_attributes.csv and Data_dictionary.csv, containing tree identifiers, reference measurements, allometric estimates, MLS-QSM attributes, units, sources, and definitions. · 04_Plot-level_segmented_point_clouds.zip: two parts (part 01 and part 02) with eight .las files named <Plot>_segmented_arbortrees.las, one per plot. Each file contains all trees automatically segmented by arbor that were flagged as within the plot and were taller than 4 m. Individual trees retain their segmentation attributes, and all tree point clouds were truncated 0.25 m above the digital terrain model (DTM). · 05_Plot-level_DTM.zip: eight .tif files named <Plot>_dtm.tif, one per plot. Each raster is the DTM generated by arbor during ground segmentation and used to compute height above ground (HAG). Code and reproducibility · QSMs_arbor.R: R script used to reconstruct QSMs and reproduce the MLS-QSM attributes. Documentation and illustrations · Readme.docx: dataset description and usage instructions. · Dataset_98_trees.tif: high-resolution visualization of the 98 validation trees. · Dataset_preview.png: preview image for the Zenodo record. · Plot-level_tree segmentation_transect_WE_7m.tif: visualization of 7 m-wide west–east transects through the eight study plots. Trees taller than 4 m were automatically segmented by arbor, with each individual tree displayed in a distinct colour. File naming and linkage All tree-level files use the Tree identifier from Tree_attributes.csv; corresponding .laz and Vtot .qsm files share the same basename. Plot-level files follow the naming convention <Plot>_segmented_arbortrees.las (e.g. DUC0001-01_segmented_arbortrees.las). Plot-level DTMs follow the naming convention <Plot>_dtm.tif (e.g. DUC0001-01_dtm.tif). Data acquisition and processing MLS data were acquired with a FARO Orbis handheld scanner (FARO Technologies, Inc., Lake Mary, Florida, USA) equipped with a HESAI XT32 lidar sensor. Each 50 m × 50 m plot was scanned using systematic parallel transects spaced 10 m apart and diagonal passes with loop closure, and the resulting point clouds were georeferenced before processing. Plot-scale data were processed in FARO Connect and subsequently analyzed using the default arbor workflow (v.0.14.0, commit 067e97d) and parameters without site-specific fine-tuning, including initial decimation, point-cloud homogenization, height-above-ground computation, semantic leaf-wood classification, individual-tree segmentation, QSM reconstruction, and tree-attribute extraction. The plot-level segmented point clouds included in this release correspond to the outputs of this automated segmentation stage. The corresponding plot-level DTMs were generated by arbor during ground segmentation and used for height-above-ground computation. The 98 validation trees were then identified and exported as the processed individual-tree point clouds provided in this dataset. The archived MLS-QSM attributes and Vtot QSMs correspond to this software version and commit, thereby preserving the exact outputs associated with the accompanying manuscript. Field and destructive reference data Field DBH was measured with a diameter tape at 1.30 m above ground or above basal deformities when required. Standing tree height was measured with a Vertex IV hypsometer. After felling, total height was remeasured from the base to the apex and stem and merchantable branch diameters were measured for destructive volume calculation. Reference volumes were calculated from destructive measurements using Smalian’s formula. Merchantable volumes were defined above the tree-specific stump height to a 9 cm small-end diameter outside bark (DOB). Branches with less than 1 m of merchantable length at DOB ≥ 9 cm were excluded, following Perron (2003)[1]. The exact definitions of Vop, Vstem, Vbranch, and Vbutt are provided in Data_dictionary.csv. Vop_allom_m3 provides the operational merchantable volume predicted from the species-specific model of Fortin et al. (2007)[2] using field-measured DBH and felled-tree height. Original under-bark predictions were converted to over-bark volume using a bark expansion factor of 1.12. QSM files and .qsm format The 02_QSM folder contains the full reconstructed woody QSM (Vtot) for each of the 98 validation trees. Vop and Vstem QSMs are not included because they can be derived from Vtot using the rules implemented in the provided R script.QSMs were reconstructed with arbor and are provided in the .qsm binary format, an open format for exchanging cylinder-based QSMs. The format specification and read/write tools are available in the arbor documentation and repository: https://github.com/r-lidar/arbor. The .qsm files can be read and visualized directly in R with arbor. For example: # Load arbor and define the QSM file path library(arbor) qsm_file <- file.path("02_QSM", "DUC0001-01_53.qsm") # Read and visualize the QSM qsm <- qsm_read(qsm_file) plot_qsm(qsm, color = "branch_order", skeleton = FALSE, cylinder = TRUE) # Optionally load the corresponding tree point cloud and overlay the QSM library(lidR) tree <- readTLS(file.path("01_Sample_trees", "DUC0001-01_53.laz")) x <- plot_semantic(tree, bg = "white") plot_qsm(qsm, add = x, color = "branch_order", skeleton = FALSE, cylinder = TRUE) # Optionally export to OBJ and CSV format qsm_write(qsm, sub("\\.qsm$", ".obj", qsm_file)) qsm_write(qsm, sub("\\.qsm$", ".csv", qsm_file)) Reproducibility Set the R working directory to the root folder of the downloaded dataset and run: source("QSMs_arbor.R"). The script reconstructs the QSMs from the provided .laz files and creates a 04_Output folder containing the reproduced MLS-QSM attributes and, optionally, the reconstructed QSM files. The analysis was performed with arbor v.0.14.0 (commit 067e97d). The archived MLS-QSM attributes and Vtot QSMs therefore provide a fixed record of the outputs associated with the accompanying manuscript. Tree attribute table Tree_attributes.csv contains 22 fields grouped into identifiers, field measurements, destructive-reference measurements, allometric predictions, and MLS-QSM estimates. Units and complete definitions are provided in Data_dictionary.csv. · Identifiers: Site, Plot, Tree, Species · Field measurements: DBH_ref_cm, FieldH_DBH_m, H_standing_m · Destructive references: StumpH_m, H_felled_ref_m, Vop_ref_m3, Vstem_ref_m3, Vbranch_ref_m3, Vbutt_ref_m3 · Allometric benchmark: Vop_allom_m3 · MLS-QSM estimates: DBH_mls_130_cm, DBH_mls_FieldH_cm, H_mls_m, Vtot_mls_m3, Vop_mls_m3, Vstem_mls_m3, Vbranch_mls_m3, Vbutt_mls_m3 Recommended use The dataset is intended for validation and development of tree-level lidar and QSM methods, comparison of field and remotely sensed tree attributes, testing of merchantable-volume algorithms, evaluation of QSM interoperability, and visual inspection of plot-scale segmentation and terrain outputs. Users should retain the Tree identifier when joining point clouds, QSMs, and tabular attributes. License and citation Dataset license: Creative Commons Attribution 4.0 International (CC BY 4.0) When using this dataset, please cite: · The dataset record: 10.5281/zenodo.23348128 · The accompanying manuscrit: Vandendaele et al., “Evaluation of an automated mobile laser scanning workflow for merchantable tree volume estimation in northern hardwood stands” (see Related works in the Zenodo record for publication details and DOI). · The related software: arbor R package and documentation – https://github.com/r-lidar/arbor [1] Perron JY. Tarif de cubage général - Volume marchand brut. Ministère Ressour Nat Faune Parcs 2003:54. https://mrnf.gouv.qc.ca/documents/forets/inventaire/tarif-de-cubage-64.pdf [2] Fortin M, DeBlois J, Bernier S et al. Mise au point d’un tarif de cubage général pour les forêts québécoises : une approche pour mieux évaluer l’incertitude associée aux prévisions. For Chron 2007:754–65. https://doi.org/10.5558/tfc83754-5