Fractal Method Revolutionizes 3D Tree Modeling with High Accuracy

Researchers at the East China University of Technology have developed a breakthrough method for reconstructing 3D tree models that significantly reduces common inaccuracies found in existing techniques. By integrating skeleton graph optimization with fractal self-similarity, this innovative approach addresses issues like incorrectly connected branches and gaps resulting from incomplete data scans. The findings were published in the journal Plant Phenomics on June 1, 2025.

The study evaluated the new method, referred to as SfQSM, on 29 trees across various tropical forest sites in Peru, Indonesia, and Guyana. These locations represented different ecological types, such as moist terra firme, peat swamp, and lowland forests. The results were impressive, achieving a concordance correlation coefficient of 0.994, indicating near-perfect accuracy, and outperforming popular models like TreeQSM and AdQSM.

Trees are vital to ecosystem regulation, biodiversity maintenance, and climate change mitigation. Accurate 3D models are essential for calculating important parameters like diameter at breast height (DBH), above-ground biomass, and wood volume. These measurements are crucial for estimating carbon stocks and understanding forest structure. While LiDAR (Light Detection and Ranging) technology has provided a means to capture point cloud data for tree modeling, traditional methods often struggle with fragmentations or erroneous branch connections.

In their quest to create a more reliable modeling method, the researchers focused on overcoming the limitations of current techniques that are sensitive to outliers and data quality issues. To rigorously assess the performance of SfQSM, the team employed a Riegl VZ-400 terrestrial laser scanner operating at 1550 nm, which produced high-resolution point clouds with a precision of 1 cm. All trees were destructively harvested, allowing for accurate volume calculations of stems, buttresses, and large branches using standard forestry techniques. This provided crucial benchmark data for evaluating the new modeling method.

The results demonstrated that SfQSM produced highly accurate volume estimates, with deviations from harvested volumes mostly falling between −1 m³ and 1 m³ across all sites. Interestingly, smaller-diameter trees from Indonesia exhibited lower deviations, while larger trees from Peru showed more significant discrepancies, highlighting the influence of tree size on modeling precision. Quantitative evaluations revealed that SfQSM achieved a mean deviation of 0.162 m³, a root mean square error of 1.023 m³, and relative errors as low as 0.01% and 0.09%. In contrast, TreeQSM’s errors were more than twice as high, and AdQSM’s deviations exceeded those of SfQSM by over thirtyfold.

Visual assessments further underscored the robustness of SfQSM. While TreeQSM often resulted in fragmented trunks and AdQSM produced overfitted or non-existent branches, SfQSM consistently generated continuous and realistic models that adhered closely to point cloud data.

These findings not only demonstrate that the SfQSM method exceeds the accuracy of existing approaches but also provide a reliable foundation for ecological applications. By delivering precise individual tree models, this method enhances biodiversity assessments, species classification, and habitat analysis. It also facilitates accurate estimations of tree volume and biomass, which are essential for calculating carbon stocks and understanding forests’ contributions to the global carbon cycle.

Moreover, the high-fidelity reconstructions produced by SfQSM can improve virtual ecological landscapes and inform sustainable forestry practices, reforestation initiatives, and climate change mitigation strategies. This innovative approach to 3D tree modeling represents a significant advancement in the field, paving the way for more effective ecological research and management.

The research was supported by various funding bodies, including the National Key Laboratory of Uranium Resources Exploration-Mining and Nuclear Remote Sensing, Jiangxi Province’s Outstanding Young Talents Funding, and the National Natural Science Foundation of China, among others. The collaborative efforts of these institutions underscore the importance of continued investment in ecological and environmental research.