Objective As an important component of urban ecosystems, urban forests possess significant carbon sequestration potential. This study aims to conduct a large-scale field survey of urban forests in the central area of Beijing, which is most severely affected by urbanization, and to integrate multi-source data to investigate the long-term, multi-scale dynamics of carbon stocks and their influencing factors, intending to provide a crucial scientific reference for the formulation of urban forestry management policies in China.

Methods Taking Beijing as the study area, we conducted a field survey of typical urban parks in the central area during the summers of 2023 and 2024. We combined multiple remote sensing data to comparatively analyze the carbon stocks within the Fifth Ring Road, the old urban area, and typical urban parks within the old urban area. Based on the NDVI-carbon density model, we assessed the carbon sequestration potential of urban forests at different regional scales in Beijing over the past decade. Using land use classification maps of multiple urban parks, we investigated the influence of urban ecosystem landscape structure on carbon density.

Results (1) A total of 118 sample plots were investigated in the field, with the highest carbon density found in Ditan Park (106.4 Mg/ha) and the lowest in Jingshan Park (22.4 Mg/ha). (2) A comparison of carbon density estimation results from imagery with different spatial resolutions reveals that high-resolution remote sensing imagery (Sentinel-2A) offers significant advantages for carbon density assessment. Medium- and low-resolution imagery (Landsat-8, Spot-6) tend to underestimate urban forest carbon density to varying degrees, and the degree of this underestimation increases as the spatial scale decreases, with the effect being particularly pronounced at the park scale. (3) Based on estimation products derived from Sentinel-2A imagery, the carbon sequestration capacity of Beijing's urban forests has continuously enhanced over the past decade (2015–2024). The increase in carbon storage is significant across all spatial scales. Among these, the urban forests within the 5th Ring Road exhibit the greatest carbon sequestration potential, with their carbon stock in 2024 being approximately 1.9 times that of 2015. (4) At the urban park scale, landscape composition and configuration significantly affected urban forest carbon stocks. Urban park carbon density, based on Sentinel-2A image estimation products, was significantly negatively correlated with PD, ENN_MN, DIVISION, and SHDI.

Conclusion Beijing's urban forest carbon sequestration capacity has shown significant variations across different spatiotemporal scales over the past decade. The greatest carbon sink potential was observed at the regional scale within the Fifth Ring Road,, indicating a steady increase in Beijing's forest carbon stock and a potential for sink enhancement that should not be overlooked. Assessing the carbon sink potential of urban forests in metropolitan areas requires higher spatial resolution remote sensing imagery to reveal the influence of landscape structure on carbon stocks. Research on the carbon sink potential of forests in the core areas of megacities, achieved by integrating multi-source data for spatially explicit analysis, can provide a crucial reference for the quality improvement, efficiency enhancement, and management of urban forests in other rapidly urbanizing regions of China.