DART-simpleCanopy-lidar

This dataset provides a high-resolution synthetic 3D point cloud of a tree orchard, generated using the DART radiative transfer model. The simulation includes only leaf elements—branches, trunks, and fruits are not represented. It features spatially distributed LiDAR points, voxelised ground-truth Leaf Area Density (LAD) values, and detailed simulation metadata. The scanning sequence emulates (771 poses) a terrestrial laser scanning (TLS) campaign conducted from a ground vehicle, following a predefined Unmanned Ground Vehicle (UGV) trajectory and configured through DART's sequential mode with realistic sensor geometry. The dataset is designed to support the development and benchmarking of LiDAR-based algorithms for vegetation analysis. It is particularly suited for: This dataset is intended for the validation of algorithms for: Estimating LAD from 3D point clouds Inferring Leaf Area (LA) and Leaf Area Index (LAI) at tree or plot level Semantic segmentation of canopy components Radiative transfer and energy balance modelling in orchards For the simulation of canopy angular structure in 100 (10x10( synthetic trees using DART software, 24 different leaf angle distribution (LAD) profiles were defined. Each profile was designed to represent realistic architectures observed in crops and woody species, ensuring representative coverage of structural variations reported in the literature. 📁 Included Files and Content whole_point_cloud.txt:Contains the full 3D point cloud in ASCII format. Each point includes: Cartesian coordinates (X, Y, Z) in an absolute reference frame Local coordinates (x, y, z) and corresponding spherical coordinates (Rho, Phi, Theta) Laser return metadata (NumberReturns, ReturnIndx, AmplitudeMaxGaussian) besm-based augmented metrics (passing_beams, inside_beams, total_beams, gap_fraction) Sequence number (seq_num) identifying the single LiDAR scanning LAD.txt:Ground-truth LAD values computed per voxel (0.05 m³). Each row contains: Voxel centroid position (vox_X, vox_Y, vox_Z) Total leaf area (LA) within the voxel Leaf Area Density (LAD, in m²/m³) dart_trees_config.txt:Describes the spatial arrangement and biophysical parameters of 100 trees used in the scene. Each tree entry defines: Trunk and crown geometry Crown type and LAI Species ID and spatial position distributions.txt:Defines leaf angle distribution and optical properties per species. Parameters include: Distribution type (e.g., spherical, erectophile) Structural and optical coefficients used in the simulation 📐 Simulation Details LiDAR scanner configuration:A TLS unit mounted on a UGV, with the following characteristics: Central zenith angle: 90° (downward-looking) Zenith FOV: 90°, Azimuth FOV: 360° Angular resolution: Horizontal=0.1759°, Vertical= 0.6976° Scanner position: (X=5, Y=6, Z=1.5) m Laser properties: Beam diameter: 9.5 mm Footprint and field-of-view defined by half angles: 0.0015 and 0.003 radians respectively Range interval stored: 0–50 m relative to the sensor Scene composition: 100 trees, with varying species, geometry and LAI Individual leaf area: 0.003 m2 Configured using DART's Earth Scene:trees module Species properties assigned through Simulation Editor and distributions.tx 🔑 Keywords LiDAR, synthetic point cloud, voxel LAD, canopy structure, radiative transfer, DART model, semantic vegetation analysis, precision agriculture, LAI, tree modelling, UGV, MLS, gap fraction, ground truth.