Abstract
Precision agricultural monitoring is now widely deployed in flat annual croplands, and yet these same methods and data outputs fail to capture the complexity of 3D, vertical crop-yielding surfaces such as fruit orchards. This study presents a novel and cost-effective drone photogrammetry approach that employs consumer-grade equipment and production-ready software to inform orchard management decisions. Flying at unconventionally low altitudes (12-30 m), we employ oblique flight patterns to capture vertical surfaces of trellised apple trees. Using precisely georeferenced image stacks, we construct high-resolution photogrammetry point clouds that allow us to quantify orchard management targets such as flowers and fruitlets. This allows us to adapt forest ecology analysis methods (e.g. lidR package [1]) to an agricultural setting, and to achieve the equivalent of 3 mm "Ground Sampling Distance"(GSD) with industry-standard 1-2 cm georeferencing accuracy. Field experiments in South Tyrol's Vinschgau-Venosta apple orchards during critical flowering periods (April 2025) demonstrated next-day "bloom density"mapping capabilities. Workflows yielding aligned outputs were validated on a different drone and multispectral payload to detect 20-40 mm fruitlets (June-July 2025). Our approach bridges the gap in spatial and temporal resolution between more mature and spatially extensive 2D drone photogrammetry and slower but more precise ground-based approaches, and is well-suited for spatially complex agricultural systems. Overall, we show that drone-based monitoring can reduce labor and fuel inputs needed for tree-level monitoring in trellised orchard systems with minimal capital investment (i.e., equipment costs significantly below standard farm machinery).