Multi-Scale BRDF Modeling and Scale Effect Analysis Using UAV Multi-Angle Remote Sensing

Unmanned aerial vehicles (UAVs) are increasingly used in close-range remote sensing due to flexible multi-angle observations and high spatial resolution. However, fine-scale surface heterogeneity challenges accurate modeling of reflectance anisotropy. Most existing BRDF models are designed for medium or low resolutions data and may not perform well with high resolution UAV imagery. This study examines the scale dependence of semi-empirical BRDF models across grassland, bare soil, and wheat using UAV-acquired multi-angle spectral imagery with a 5 cm ground sampling distance (GSD). Data were resampled to six coarser resolutions (0.25 m to 8 m) to assess performance at multiple scales. Results show BRDF accuracy depends strongly on resolution and surface type. Grassland and bare soil fit best at 1 m, while wheat performed better at coarser resolutions due to vegetation structure. Solar zenith angle affected light incidence and scattering, and the absence of hotspot correction caused systematic errors. The optimized RTKLS_C model improved simulation in hotspot regions. This study highlights the importance of scale-aware BRDF modeling for UAV remote sensing and guides improvements in complex surface environments.