Assessing Carbon Stocks in A Tropical Forest Ecosystem of High Conservation Value: A Case Study in Bac Kan Province, Vietnam

Carbon stocks (CS) is a crucial factor in understanding the carbon cycle and enhancing forest quality monitoring, supporting sustainable forestry management and policy adjustment. This study shows a non-invasive method for estimating aboveground biomass (AGB) and CS, integrating multi-source remote sensing, topography and field data. The tropical forest ecosystem in Bac Kan province was chosen because this is the province with the largest forest cover in Vietnam and is considering participating in the global carbon credit market. The Cubist model was developed using Sentinel-1A SAR data, indices from Landsat-9 OLI-2 images and topographic data. The accuracy of Cubist models is evaluated through statistical indicators such as: Root mean square deviation (RMSE), Coefficient of determination (R2) and Mean absolute error (MAE). Among 130 Cubist prediction models, the Model 1 demonstrated the highest accuracy for estimating AGB in the study area. Results indicate that the accumulated carbon stock is primarily concentrated in Evergreen broadleaved forest on soil mountain (EBFS) accounting for about 40 or approximately 2.6 million MgC; Mixed wood-bamboo forest on soil mountain (MWBF) accounting for about 23%, or 1.5 million MgC; and Plantation on soil mountain (PFS) accounting for about 22%, or 1.5 million MgC. The study provides a comprehensive view of forest ecosystem conservation and improves monitoring, protection and restoration of high-value forests.