Predicting forest biodiversity from spaceborne GEDI LiDAR and machine learning: structural drivers and regional transferability across Pacific Northwest forests

Abstract

Remote sensing offers important opportunities for large-scale biodiversity monitoring, yet the capacity of spaceborne laser scanning to predict forest biodiversity remains insufficiently understood. This study evaluates relationships between forest structural complexity and tree species diversity using data from the GEDI mission. Structural metrics describing canopy height, vegetation density and vertical foliage distribution were linked to ground-based forest inventory data from 9,692 plots across the Pacific Northwest region of the United States. Tree biodiversity was quantified using species richness and two diversity indices. Structural complexity showed consistent positive relationships with biodiversity, with vegetation density and vertical foliage heterogeneity outperforming maximum canopy height as predictors. Machine-learning models explained approximately 17–22% of biodiversity variation under standard cross-validation. In contrast, spatial cross-validation revealed strong reductions in predictive performance, indicating limited geographic transferability. Including broad-scale climate variables produced modest improvements, while regionally calibrated models substantially increased accuracy, achieving up to 40% explained variance for species richness. Overall, the results demonstrate that spaceborne laser-derived forest structure contains meaningful biodiversity information, but effective large-area application requires regional calibration to account for spatial variation in forest composition and environmental conditions.