Modelling Forest Structure and Aboveground Biomass Dynamics in Southwestern Nigeria Using GEDI LiDAR and Multi-Sensor Fusion

This study modelled canopy height and aboveground biomass (AGB) dynamics across Southwestern Nigeria (2020–2025) by integrating GEDI LiDAR metrics with multi-sensor predictors, including optical, radar, topographic, and environmental variables. Using machine learning, the research quantified forest degradation and associated carbon loss in this data-scarce tropical region. Model performance was moderate but consistent with regional-scale studies, with Random Forest providing the best canopy height estimates in 2020 (R² = 0.486) and XGBoost yielding the most accurate AGB results in 2025 (R² = 0.475). The findings reveal a significant ecological decline: mean canopy height decreased by 16.63% (from 8.36 m to 6.97 m), while AGB declined from 64.89 t/ha to 45.18 t/ha, representing a 30.37% reduction across the region. This reduction indicates a substantial decline in carbon storage capacity, with biomass equivalent decreasing from 111.61 tCO₂/ha in 2020 to 77.71 tCO₂/ha in 2025. Analysis showed that canopy height is primarily driven by moisture-sensitive SWIR variables and disturbance indices, whereas AGB is influenced by a broader combination of structural, climatic, and soil factors. Notably, biomass loss consistently exceeded canopy height reduction, suggesting structural thinning, where forests lose density and carbon despite relatively stable vertical structure. These results demonstrate the effectiveness of GEDI-based multi-sensor fusion for monitoring forest degradation and carbon dynamics. The observed biomass declines highlight significant implications for carbon storage and underscore the urgent need for improved forest conservation and management strategies in Southwestern Nigeria.