GEDI-Sentinel-NFI 를 이용한 산림 지상부 바이오매스 산정* -경기도 6개 지역을 대상으로- = Forest Aboveground Biomass Estimation Using GEDI, Sentinel, and NFI Data* -A Case Study of Six Regions in Gyeonggi Province-

다국어 초록 (Multilingual Abstract)

Aboveground biomass is a key indicator for quantifying forest carbon stocks and their changes, and spatially continuous estimation methods are needed for forest carbon monitoring. However, conventional estimation approaches have limitations in representing fine-scale spatial heterogeneity within forests because of constraints related to data update cycles and spatial representativeness. This study estimated aboveground biomass in six regions of Gyeonggi Province, South Korea, by generating a 10 m continuous canopy height map using GEDI, Sentinel-1/2, terrain, and land cover data, and by integrating it with National Forest Inventory-based forest type map attributes. The applicability and bias characteristics of the GEDI-based estimation approach were evaluated through comparisons with national biomass estimates and airborne LiDAR-based biomass estimates.
The GEDI-based canopy height estimation model showed a performance of R² = 0.70 and RMSE = 4.37 m. Although the GEDI-based aboveground biomass estimates were generally higher than both national and airborne LiDAR-based estimates in terms of absolute magnitude, their spatial patterns showed strong agreement with airborne LiDAR-based biomass, with correlation coefficients of 0.84–0.89 at the 10 m resolution. High-difference areas, where the discrepancy between GEDI-based and airborne LiDAR-based estimates was large, had higher proportions of 4th–5th age classes, high canopy density, broadleaf and mixed forests, and steep slopes. These results indicate that the bias in GEDI-based biomass estimation may be associated with specific forest structural and topographic conditions.
This study demonstrates the potential of integrating open-access remote sensing data with national forest inventory information to spatially map aboveground biomass in Korean forests. This approach can complement sample-based forest inventory data by representing within-forest spatial variability and can improve comparability and consistency across regions through a standardized estimation procedure. To improve the accuracy of GEDI-based biomass estimation, further research should incorporate independent validation data, such as field measurements and airborne LiDAR, and develop correction schemes that account for forest structural and topographic conditions.