Impact of Radiometric Variability on Ultra-High Resolution Hyperspectral Imagery Over Aquatic Vegetation: Preliminary Results

Over the last two decades, advancements in airborne imaging spectroscopy have prompted the exploitation of lightweight drones for detailed vegetation assessment at unprecedented resolutions. Yet, surface reflectance anisotropy and view-illumination effects may bias spectra extracted from push-broom scanners and derived spectral indices (SIs), particularly over aquatic vegetation, thus impacting the retrieval of biophysical and biochemical vegetation parameters. In this study, the impact of illumination conditions (overcast versus clear sky) and angular configurations (i.e., solar and viewing angles) on radiometric variability of centimetric resolution drone data was empirically investigated over four different aquatic plant species, representing different growth forms and canopy structures. Nadir-normalized reflectance spectra, broadband SIs, and the spectral angle distance to proximal leaf reflectance were used for characterizing and quantifying radiometric variability at canopy and leaf levels. Our findings demonstrated a decrement in reflectance under diffuse light conditions, especially in highly reflective domains within Green (520–580 nm) and Near-Infrared (700–850 nm) ranges, and a marked angular reflectance anisotropy in high absorption spectral regions (i.e., 450–500 nm and 630–700 nm) for aquatic vegetation. The normalized difference vegetation index (NDVI) showed overall lower sensitivity to incoming light variability and angular configurations compared to other tested SIs, whereas the water adjusted vegetation index (WAVI), suitably designed for aquatic vegetation, was less affected by angular anisotropy in floating plants. Indeed, radiometric variability exhibited a dependence on aquatic plant features, i.e., leaf orientation, canopy structure, and affinity with water (as canopy background).