Forest Aboveground Biomass Estimation Using High-Resolution Imagery and Integrated Machine Learning

This investigation developed four different algorithmic approaches\u2014Random Forest, XGBoost, Boosting, and Stacking\u2014utilizing grid search optimization combined with cross-validation for parameter tuning. Each approach successfully captured the spatial patterns of tree AGB throughout Daxing District, although performance levels varied in terms of accuracy and stability. The ensemble Stacking approach excelled beyond individual methods across evaluation criteria by integrating outputs from multiple base algorithms, resulting in superior predictive consistency. Therefore, the ensemble method was chosen to produce the comprehensive AGB spatial distribution for Daxing District. These spatial patterns underscore the intricate nature of Beijing's forest systems and contribute novel perspectives on carbon sequestration patterns within peri-urban forest landscapes. In contrast to earlier research predominantly examining uniform forest stands or single ecosystem types, this multi-area assessment delivers enhanced scientific foundations for metropolitan forest stewardship strategies.Given the costs and complexity associated with data acquisition and processing, single high-resolution optical imagery approaches retain significant practical value for real-world applications. This is particularly true for forest management applications requiring regular monitoring, where simplified data requirements enhance operational feasibility. While our study achieved satisfactory estimation accuracy, opportunities for improvement remain. The limitations of optical remote sensing in capturing forest vertical structure information constrain further enhancement of model performance. Future research should explore the integration of SAR and LiDAR data to obtain more comprehensive three-dimensional forest structural information. Additionally, incorporating climatic and topographic factors would strengthen the ecological interpretability of models, particularly in elucidating AGB accumulation mechanisms across different forest types. Meanwhile, investigating model transferability across forest types represents a crucial research direction that holds significant implications for expanding the scope of technological applications.