Accuracy assessment of LAI, PAI and FCOVER from Sentinel-2 and GEDI for monitoring forests and their disturbance in Central Germany

Forest monitoring benefits from biophysical parameters such as Leaf Area Index (LAI), Plant Area Index (PAI) and fractional vegetation cover (FCOVER) and can be obtained from optical and LiDAR remote sensing, such as Sentinel-2 (S2) and the Global Ecosystem Dynamics Investigation (GEDI). While GEDI-derived products consider all phyto-elements, those from S2 refer to green elements only. Apart from individual accuracies, systemic deviations among products are thus expectable. However, products from S2 and GEDI lack inter-comparison. We evaluated S2 and GEDI-derived LAI, PAI and FCOVER with digital hemispherical photography observations (DHP) in a forest disturbance hotspot in Germany across various forest conditions, including vital stands, standing deadwood and clearings. We found moderate to high agreement with in-situ data, with highest accuracy for S2-derived LAI (R2 = 0.54) and GEDI-derived FCOVER (R2 = 0.73). Agreements between S2 and GEDI products were low, which we attribute to systematic influences of woody components, GEDI’s limitations in sloped terrain, and saturation of optical signals in dense canopy. In conclusion, findings suggest that while GEDI is effective in dense canopies, S2 products are beneficial for monitoring forest recovery. We also see potential for synergistic use in monitoring standing deadwood for habitat mapping and fire risk assessment.