Assessing UAV imagery and high-resolution LiDAR for tree height estimation: The role of flight speed and image overlap
- 1University of Arkansas Division of Agriculture, Arkansas Forest Resources Center, University of Arkansas System, 110 University Court, Monticello, AR, 71656, USA
- 2College of Forestry, Agriculture, and Natural Resources, University of Arkansas at Monticello, 110 University Court, Monticello, AR, 71656, USA
Abstract
Tree height is a fundamental attribute in ecological research and commercial forestry, serving as a key indicator of site productivity. Unmanned Aerial Vehicles (UAVs) equipped with RGB cameras and Light Detection and Ranging (LiDAR) sensors offer a cost- and time-efficient alternative to traditional field-based, satellite, and manned aircraft methods for tree height measurement. The objectives of this study are: (1) to evaluate the effects of UAV flight speed and image overlap on data quality, mission efficiency, and processing requirements; (2) to compare the performance of Structure-from-Motion (SfM) photogrammetry and LiDAR in generating canopy height models (CHMs); and (3) to quantify the accuracy of UAV-derived tree height estimates relative to field measurements and identify optimal flight configurations. Three flight speeds (10, 15, and 20 mph) and four forward/side overlap levels (50%, 60%, 70%, and 80%) were tested, with UAV-derived height estimates validated against 920 field-measured trees. Both datasets were processed in ArcGIS Pro to produce canopy height models (CHMs), RGB imagery via Structure-from-Motion (SfM) photogrammetry, and LiDAR data from laser-scanned point clouds. Orthomosaic quality was assessed using tie point density, reprojection error, ground resolution, image georeferencing deviation, Global Positioning System (GPS) root mean squared error (RMSE), and block adjustment success, while LiDAR point cloud quality was evaluated using point density (pts/m 2 ) and height percentiles (P25, P50, P75, P95). Height estimation accuracy for both sensors was quantified using the coefficient of determination (R 2 ), RMSE, and Bias. Results indicate that image overlap exerted a stronger and more consistent influence than flight speed across all dimensions of mission efficiency, data volume, and processing time. Flight duration more than doubled and image counts increased sixfold when overlap increased from 50:50 to 80:80. Higher overlaps improved orthomosaic continuity, tie point density, reprojection accuracy, and CHM quality, though at the cost of longer processing times, greater storage demands, and increased computational requirements. LiDAR point density similarly increased with overlap, yielding smoother CHMs at ≥70% overlap, while height percentiles remained stable across configurations. In terms of accuracy, UAV imagery at 10 mph with 80:80 overlap achieved the best photogrammetric performance (R 2 ≈ 0.60, RMSE = 4.5 m, Bias = 4.3 m), though all imagery-derived estimates exhibited systematic height underestimation. LiDAR-derived heights were substantially more robust across all flight configurations, with the best performance at 10 mph and 80:80 overlap (R 2 ≈ 0.89, RMSE < 1.5 m, Bias < 0.5 m). These findings demonstrate that higher overlap, particularly at moderate flight speeds, substantially enhances data quality and tree height estimation accuracy, offering practical guidance for optimizing UAV-based forest inventory workflows.
Related
- Assessing the feasibility of up-scaling grapevine physiological trait estimation using multispectral UAV imagery
- Estimation of rice leaf angle distribution from geometric features in millimetric-resolution UAV RGB imagery acquired at 45° viewing angle
- The capability of very high-resolution satellite imagery for plot-level early wheat stem rust disease detection, monitoring, and phenotyping in Ethiopia
- Detection of Red Crown Rot of Soybean in Illinois Fields Using High-Resolution Satellite Imagery and Machine Learning
- Neural Processes with Normalizing Flows for Wheat Height Estimation
BibTeX
@article{subedi2026assessing,
title = {{Assessing UAV imagery and high-resolution LiDAR for tree height estimation: The role of flight speed and image overlap}},
author = {Prajwol Babu Subedi and Hamdi A. Zurqani},
journal = {Science of Remote Sensing},
year = {2026},
month = jun,
doi = {10.1016/j.srs.2026.100448},
url = {https://doi.org/10.1016/j.srs.2026.100448},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Assessing UAV imagery and high-resolution LiDAR for tree height estimation: The role of flight speed and image overlap
AU - Prajwol Babu Subedi
AU - Hamdi A. Zurqani
PY - 2026
DA - 2026/06/01
JO - Science of Remote Sensing
DO - 10.1016/j.srs.2026.100448
UR - https://doi.org/10.1016/j.srs.2026.100448
AB - Tree height is a fundamental attribute in ecological research and commercial forestry, serving as a key indicator of site productivity. Unmanned Aerial Vehicles (UAVs) equipped with RGB cameras and Light Detection and Ranging (LiDAR) sensors offer a cost- and time-efficient alternative to traditional field-based, satellite, and manned aircraft methods for tree height measurement. The objectives of this study are: (1) to evaluate the effects of UAV flight speed and image overlap on data quality, mission efficiency, and processing requirements; (2) to compare the performance of Structure-from-Motion (SfM) photogrammetry and LiDAR in generating canopy height models (CHMs); and (3) to quantify the accuracy of UAV-derived tree height estimates relative to field measurements and identify optimal flight configurations. Three flight speeds (10, 15, and 20 mph) and four forward/side overlap levels (50%, 60%, 70%, and 80%) were tested, with UAV-derived height estimates validated against 920 field-measured trees. Both datasets were processed in ArcGIS Pro to produce canopy height models (CHMs), RGB imagery via Structure-from-Motion (SfM) photogrammetry, and LiDAR data from laser-scanned point clouds. Orthomosaic quality was assessed using tie point density, reprojection error, ground resolution, image georeferencing deviation, Global Positioning System (GPS) root mean squared error (RMSE), and block adjustment success, while LiDAR point cloud quality was evaluated using point density (pts/m 2 ) and height percentiles (P25, P50, P75, P95). Height estimation accuracy for both sensors was quantified using the coefficient of determination (R 2 ), RMSE, and Bias. Results indicate that image overlap exerted a stronger and more consistent influence than flight speed across all dimensions of mission efficiency, data volume, and processing time. Flight duration more than doubled and image counts increased sixfold when overlap increased from 50:50 to 80:80. Higher overlaps improved orthomosaic continuity, tie point density, reprojection accuracy, and CHM quality, though at the cost of longer processing times, greater storage demands, and increased computational requirements. LiDAR point density similarly increased with overlap, yielding smoother CHMs at ≥70% overlap, while height percentiles remained stable across configurations. In terms of accuracy, UAV imagery at 10 mph with 80:80 overlap achieved the best photogrammetric performance (R 2 ≈ 0.60, RMSE = 4.5 m, Bias = 4.3 m), though all imagery-derived estimates exhibited systematic height underestimation. LiDAR-derived heights were substantially more robust across all flight configurations, with the best performance at 10 mph and 80:80 overlap (R 2 ≈ 0.89, RMSE < 1.5 m, Bias < 0.5 m). These findings demonstrate that higher overlap, particularly at moderate flight speeds, substantially enhances data quality and tree height estimation accuracy, offering practical guidance for optimizing UAV-based forest inventory workflows.
SN - 2666-0172
ER -