Estimation of Aboveground Biomass in Zambia through Integration of GEDI, Sentinel-1 and Sentinel-2 Measurements

Accurate estimation of aboveground biomass (AGB) at regional scales remains challenging, particularly in diverse ecosystems like Zambia. This study developed an integrated approach combining spaceborne LiDAR (GEDI), synthetic aperture radar (Sentinel-1), and multispectral (Sentinel-2) measurements to generate footprints of 25 m diameter, 10 and 20 meter resolution maps of AGB across Zambia. Random Forest (RF) regression was employed to establish relationships between GEDI AGB product and Sentinel’s variables. Feature importance analysis revealed that Near-Infrared (NIR) and Shortwave Infrared (SWIR) bands were the most significant predictors, particularly in arid regions. The model performance varied significantly across different land cover types, with R2 values of 0.453, 0.178, and 0.054 for Savanna, Grasslands, and Woody savanna, respectively. However, the overall model validation showed moderate performance with R2 of 0.508 for AGB estimation. While these results demonstrate the potential of multi-sensor data integration for large-scale vegetation structure monitoring, they also highlight the challenges in achieving consistent accuracy across heterogeneous landscapes. Future research should focus on enhancing model performance through improved calibration strategies, better incorporation of ecosystem-specific characteristics, and collection of reliable ground measurements.