Optimizing UAV hyperspectral imaging for urban tree chlorophyll and leaf area index retrieval

Abstract

Uncrewed aerial vehicle (UAV) based hyperspectral imaging offers a flexible method for monitoring urban trees. However, its effect on estimating biochemical and biophysical parameters is still unknown. This article examines how spatial and spectral resolution, solar zenith angle (SZA), and diffuse solar irradiance (SKYL) affect chlorophyll content (Cab) and leaf area index (LAI) estimation using narrow-band indices (NBIs) through three-dimensional radiative simulations. The results show that spatial resolution minimally affects Cab estimation but significantly impacts LAI, with finer resolutions improving correlation with NBIs. In contrast, spectral resolution has little effect on LAI but greatly influences Cab, with a 2-nm resolution providing stronger correlations, while resolutions coarser than 6 nm are less sensitive. The Cab estimation prefers oblique SZAs, while LAI favors nadir SZAs. SKYL has little effect on Cab and minor impact on LAI. Sunlit pixels outperform shaded ones for Cab estimation, even at 2-m resolution, while entire-crown pixels show the highest LAI correlation. Different NBI strategies significantly affect LAI estimation but not Cab. A consistent conclusion emerges from the analysis of correlations between UAV hyperspectral imagery, with varying spatial and spectral resolutions, and corresponding Cab field measurements. This suggests that the knowledge revealed by the radiative transfer model is applicable to real-world conditions and improves understanding of natural processes without direct measurements.This article enhances the understanding of the influence of observation configurations on Cab and LAI estimation, offering insights to optimize UAV-based hyperspectral imaging and guide future satellite sensor development for tree monitoring.