Carbon Stock Mapping in Urban Areas Based on Vegetation Index Comparison from Sentinel-2A Imagery

Abstract

This study investigates the spatial distribution of urban vegetation carbon stocks by comparing multiple vegetation index algorithms derived from Sentinel-2A satellite imagery. Urban carbon stock serves as a critical indicator of environmental sustainability, reflecting land cover characteristics and vegetation quality. The research workflow included radiometric correction, calculation of four vegetation indices: NDVI, EVI, ARVI, and SAVI conversion of index values into biomass estimates, and subsequent carbon stock computation using established allometric equations. Field-based measurements were used for validation. The results showed strong positive correlations between all vegetation indices and field-observed biomass (R² > 0.90). Among them, NDVI demonstrated the highest predictive accuracy (R² = 0.91), while EVI, ARVI, and SAVI tended to slightly underestimate biomass in areas with dense canopy cover or sparse vegetation. Spatial mapping of NDVI-derived carbon stocks revealed significant heterogeneity across sub-districts, with values ranging from below 600 tons/ha in densely built-up zones to over 900 tons/ha in peri-urban green areas. The total estimated carbon stock in Malang City was 3,861.35 tons/ha, with the highest concentrations identified in urban parks, forested areas, and historical green corridors. Overall, NDVI proved to be the most reliable index for urban carbon stock estimation using Sentinel-2A data. These findings highlight the necessity of preserving high-carbon zones, enhancing ecological buffer regions, and implementing targeted urban greening initiatives to support climate change mitigation and sustainable city planning.