Contrasting Local and Global Modeling with Machine Learning and Satellite Data: A Case Study Estimating Tree Canopy Height in African Savannas

This paper investigates the use of machine learning and satellite imagery to estimate tree canopy height across African savannas. The authors compare local and global modeling approaches, analyzing how spatial heterogeneity and ecological variation affect predictive performance. Using remote sensing data and machine learning techniques, the study demonstrates tradeoffs between globally generalized models and locally specialized models for ecological monitoring. The work contributes to environmental machine learning and provides insights for biodiversity conservation, carbon accounting, and large-scale ecosystem assessment in data-constrained regions.