Abstract:

【Objective】Drawing on unmanned aerial vehicle (UAV) remote sensing and photogrammetry techniques from forestry surveys, this study aims to develop an optimal selection process for UAV solutions tailored to the investigation of urban street trees. This approach addresses the problem of uneven data sources caused by the various flight parameters and flight plans.【Method】Six different drone flight plans for urban street tree surveys were designed, and the analytic hierarchy process (AHP) was applied to select a flight plan that balanced accuracy and efficiency. Based on the selected plan, a corresponding model for estimating tree height using tree diameter at breast height (DBH) was established. The best predictive model for DBH was determined by evaluating the coefficient of determination, mean relative error (MRE), and root mean square error (RMSE). By comparing the accuracy of the tree number, tree height, crown width, and DBH, as well as the efficiency of data extraction, it was demonstrated that an optimized flight plan for urban tree surveys could improve efficiency, while maintaining the required level of accuracy.【Result】After conducting an AHP analysis, a Tic-Tac-Toe flight plan at a flight altitude of 70 m was identified as the optimal solution that balances accuracy and efficiency. The optimized flight plan achieved a precision of 94.44%, a recall of 80.95%, and an F₁-score of 87.18% for tree number extraction. The average MRE for tree height extraction was 8.66%, with an RMSE of 1.22 m, and a coefficient of determination of 0.88. The average MRE for crown width extraction was 27.77%, with an RMSE of 1.30 m and a coefficient of determination of 0.69. The established model for predicting DBH achieved a coefficient of determination of 0.81. Compared to other flight plans, the optimized flight plan was conducted in 70% less time, while maintaining a high level of accuracy in tree number, tree height, crown width, and DBH measurements.【Conclusion】The optimized UAV flight plan, after achieving the required accuracy for street tree surveys, was 70% faster and obtained better quality data for tree number, tree height, crown width, and DBH. The UAV flight plan optimization process adopted in this study was proven to be effective and can be applied to the selection for urban street tree surveys.

Key words: street tree survey, oblique imagery, unmanned aerial vehicle (UAV), analytic hierarchy process (AHP), regression model