https://www.sciencedirect.com/science/article/abs/pii/S2210670725007735
Authors: Kailong Cui, Yaoping Cui, Xiangzheng Deng, Chaosheng Zhang, Yufei Jia, Tianwei Zhao, Nan Li, Zhifang Shi, Xiang Zhao, Hua Qin
10 October 2025
Highlights
•Extracting urban tree patches based on RGB satellite images
•Reconstructing four 3D parameter by quantitative structure model
•Calculating spatial carbon stock of urban tree in Dublin, Ireland
•Dublin has 401,866 trees with a total carbon stock of 12.11 Mt
•Proposing a method with fewer input parameters for urban tree carbon stock accounting
Abstract
Urban trees play a crucial role in regulating the urban environment. Their carbon stock capacity and the importance of 3D information are increasingly recognized by urban managers. However, accurately characterizing urban trees and estimating their carbon stock is hindered by the complexity of urban landscapes and the structural and spatial tree diversity. Here we used RGB satellite imagery and locally sampled data to extract 3D information on urban trees in Dublin, Ireland, and to calculate their carbon stock. Our method consisted of: 1) extracting complete urban tree patches (UTP) using a newly defined “urban tree canopy index” and morphological operations; 2) reconstructing 3D UTP parameters with a quantitative structure model and a stacked random forest regression (S-RFR) algorithm; 3) calculating UTP carbon stocks based on tree volume and basic wood density. Validation of the extracted UTP information (tree count, volume, height, and diameter at breast height) yielded accuracy and F1-score values of 0.90 and 0.89, respectively, with highly significant (p < 0.005) R² values consistently exceeding 0.86. Overall, we identified 190,000 UTP in Dublin containing 401,866 trees. Corresponding canopy coverage was 16.43% for a total tree volume of 381,105 m3 and a total carbon stock of 12.11 Mt. Moreover, our method was highly versatile and applicable to diverse tree species, maintaining high accuracy (R² = 0.93, p < 0.005) with fewer tree-species parameters than in the complete set of this study. In summary, the implemented method will facilitate the extraction of accurate 3D parameters and enables scientists to calculate the spatial distribution of carbon stocks in urban trees with minimal influence from geographic location or species, overcoming the constraints of species data and supports extensive and refined assessments of urban tree resources. By providing reliable carbon stock data, this method enables urban managers to optimize green space planning and enhance carbon stock capacity, thereby contributing significantly to urban sustainability and climate resilience.
Source: ScienceDirect