Mangrove ecosystems are highly efficient natural carbon sinks, yet quantifying soil organic carbon density (SOCD) across stand ages is critical for evaluating restoration benefits. This study integrated WorldView-2, LiDAR, and Sentinel-1 SAR data with machine learning to map SOCD in Qinglan Harbor, China. eXtreme Gradient Boosting achieved the highest accuracy (R 2 = 0.72, RMSE = 2.87 kg m-2) among four regression methods. Lasso regression identified LiDAR-derived structural metrics and optical vegetation indices as key predictors, highlighting the importance of 3D canopy structure and spectral data. Results showed age-dependent SOCD accumulation: mangroves older than 15 years stored significantly more carbon (11.47 kg m-2) than 0-5 year stands (10.49 kg m-2). These findings underscore the value of long-term restoration for blue carbon sequestration and provide a scalable monitoring framework that integrates multi-sensor remote sensing and machine learning to support coastal management and climate mitigation.
🌊
