<![CDATA[Mangroves are vital ecosystems in combating climate change, primarily through their exceptional capacity for carbon sequestration and long-term storage. To effectively manage and conserve these valuable resources, accurate carbon stock mapping is crucial. Given the inherent variability of mangrove biophysical characteristics, selecting appropriate mapping methodologies is essential. This study aimed to evaluate two distinct approaches: regression modeling using satellite-derived vegetation indices and kriging interpolation, within the Angke Kapuk mangrove area of Jakarta. Regression models were constructed utilizing forest canopy density (FCD) and its constituent indices (derived from Landsat 8), alongside normalized difference vegetation index (NDVI), advanced vegetation index (AVI), and soil adjusted vegetation index (SAVI) from Sentinel 2A, as predictor variables. Field-based carbon stock data, obtained from 50 square plots (10 m × 10 m) using established allometric models, served as the response variable. The study revealed substantial heterogeneity in carbon storage, ranging from 34.76 to 236.87 tons/ha, with a mean of 135.31 tons/ha and a standard deviation of 50.09 tons/ha. Regression modelling, however, demonstrated limited predictive power, achieving a maximum R² value of only 0.03, indicating a poor fit between the predictor variables and observed carbon stocks. Kriging interpolation yielded moderate accuracy, as evidenced by a coefficient of variation of root mean square error (CV RMSE) of 0.39. This disparity in performance can be attributed to several factors, including the homogeneity of the rehabilitated mangrove canopy, which limited the ability of vegetation indices to accurately represent carbon stock variations. Furthermore, kriging's capacity to model spatial autocorrelation proved advantageous in this context. Based on these findings, this paper discusses the influence of mangrove characteristics on modelling performance and provides practical recommendations for area managers regarding future carbon stock mapping initiatives.]]>
Copyrights © 2026
