Document Type : Research article
Authors
1
Corresponding author, Assistant Prof., Faculty of Natural Resources, Razi University, Kermanshah, Iran
2
Aerial Monitoring Group, Razi University, Kermanshah. Iran
3
PhD. Student of Forestry, Faculty of Forest Science, Gorgan University of Agricultural Sciences and Natural Resource, Gorgan, Iran
4
Assistant Prof., Faculty of Natural Resources, Razi University, Kermanshah, Iran
Abstract
Abstract
Background and objectives: Tree structural attributes are crucial in both ecological and economic contexts, with tree height being a fundamental variable and a primary indicator for quantifying forest stand volume. Measuring tree height is among the most difficult and costly tasks, requiring specialized expertise. Unmanned aerial vehicles (UAVs) have gained attention in forestry for their advanced capabilities. Recent UAV advancements enable remote assessment of tree structural characteristics and forest stands at a relatively low cost compared to traditional methods. This study assessed the accuracy and precision of tree height measurements using the Phantom 4 RTK UAV, without ground control points, combined with the local maxima algorithm to measure Arizona cypress (Esperocyparis arizonica (Greene) Bartel) tree heights.
Methodology: The study was conducted in an Arizona cypress plantation in Kermanshah province, Iran. The Phantom 4 UAV, equipped with an RTK system and a 1-inch 20-megapixel CMOS sensor camera, provided high spatial resolution. Key features included the RTK module for real-time horizontal and vertical positioning, a three-axis gimbal to reduce flight vibration, and a mechanical shutter to minimize imaging errors. The GNSS positioning system further enhanced data accuracy, even in enclosed environments. Ninety-eight trees were sampled, and heights were measured on-site using a Leica S910 laser device. The Crown Height Model (CHM) with the local maxima algorithm was used for tree identification. A paired t-test compared measured and estimated tree heights, followed by regression analysis applying linear regression to evaluate the relationship between these values.
Results: The study produced the Digital Elevation Model (DEM), Digital Surface Model (DSM), Orthomosaic, and Crown Height Model (CHM). The local maxima algorithm successfully identified all 98 trees and their locations. Field measurements showed minimum, maximum, and mean tree heights of 1.32 m, 5.84 m, and 3.78 m, respectively, while estimates from the local maxima algorithm were 1 m, 5.65 m, and 3.7 m. The Kolmogorov-Smirnov test confirmed normal distribution of both measured and estimated datasets. The paired t-test revealed the measured heights were significantly greater than the estimated values, indicating the local maxima algorithm tended to underestimate tree height. Regression analysis demonstrated a strong positive linear correlation between measured and estimated heights (R² = 0.96) with a low error margin (RMSE = 0.26).
Conclusion: The study concludes that RTK-UAV combined with the local maxima algorithm can precisely identify tree spatial positions while reducing time and cost. This method offers forest managers reliable tree height estimates, supporting efficient forest management decisions.
Keywords
Main Subjects
)