An integrated and visualized forest carbon storage estimation platform: Functional design and prototype implementation

Forest carbon storage estimation demands expertise in processing point cloud data, extracting structural features, and applying machine learning to platforms like GEE. Existing tools mainly serve researchers, limiting nonexperts. In this study, an automated cloud platform that integrates individual tree segmentation, carbon estimation, and field data with GEE was developed. The system incorporated open-source tools including Laspy, PDAL, GDAL, and PyCrown for point cloud reading, resampling, denoising, normalization, and CHM-based individual tree segmentation. Machine learning models were implemented via Python for carbon storage estimation by remotely accessing the GEE API. Compared with traditional methods, application resulted in significantly improved efficiency, with an Rˆ2 value of 0.66, an RMSE value of 19.08 Mg/ha, and an rRMSE value of 11.04%. While the platform enhances accessibility for nonexperts, some overestimation at low biomass and underestimation at high biomass remain. Future work will focus on model optimization to mitigate these biases and improve scalability. • This cloud platform integrates data processing, tree segmentation, and carbon estimation. • In this study, open-source tools were used to implement biomass estimation modeling, reducing costs. • In this study, GEE and Python were merged to increase the efficiency of forest carbon estimation.