Scalable Inverse Allometry for UAS Forestry: Utilizing National Forest Inventory Datasets to Predict Diameter from Drone-Derived Heights and Tree Density

Unmanned Aerial Systems (UAS) have revolutionized forest mapping, yet the direct measurement of diameter at breast height (DBH) remains a significant "bottleneck" for remote-sensing-based forestinventories. This study developed a robust, three-tiered allometric engine to predict DBH using only tree height (HT) and stand density (TPA) derived from Forest Inventory and Analysis (FIA) data across the Western United States. Utilizing a dataset of 27,822 trees, species-specific master equations were calibrated for 21 species, followed by geographic adjustment factors across four EcoRegions and a non-parametric bias correction layer for large-diameter cohorts. Results indicated high model performance, with Adjusted R square values reaching 0.87 for several species and polynomial architectures emerging as the superior fit for 62% of the species studied. Regional analysis revealed significant allometric plasticity, with adjustment factors requiring shifts of up to 13.3% to account for environmental variability. Notably, the non-parametric correction successfully mitigated a systematic under-prediction of over 10 inches in old-growth cohorts (>35"), where radial growth persists despite height stagnation. By providing a mathematical bridge between canopy metrics and ground-level diameter, this framework enables the automation of volume, biomass, and carbon estimation, offering a scalable path toward high-frequency, low-cost precision forestry.