The effect of spatial resolution on radiometric and geometric performances of a UAV-mounted hyperspectral 2D imager

Abstract The effect of spatial resolution on the radiometric and geometric performances of hyperspectral sensors is an essential issue in remote sensing that urgently needs to be investigated, especially for low-altitude remote sensing principles and applications. Using an unmanned aerial vehicle (UAV)-mounted miniature hyperspectral 2D imager (Cubert UHD 185) system, a series of hyperspectral images of several reflectance targets (5%, 20%, 30%, 40%, 60% and 65%) were imaged in hovering flight at various spatial resolutions (ground sampling distances (GSDs)) from 1.2 cm to 4.8 cm, with intervals of 0.4 cm, which correspond to flight altitudes from 30 m to 120 m in increments of 10 m. Subsequently, the effect of spatial resolution on radiometric and geometric performances was evaluated in terms of the change in reflectance and geometric recognition ability of the shape of targets at visible to near-infrared wavelengths. This paper provides a set of methods for assessing the effect of spatial resolution on radiometric and geometric performance, including a radiative transfer model simulation for imaging quality performance, the geometric recognition loss degree (GRLD) for measuring image geometry recognition ability, and a trend projection analysis for developing continuous distribution images of radiometric and geometric performances. The results show that when the size of the target is not less than 50 (row) × 50 (column) pixels in a Cubert hyperspectral image, the absolute error (AE) and the root mean square error (RMSE) of the reflectances of its central pixel are both less than 0.05. Additionally, as the spatial resolution decreased, the AEs of the target reflectances in visible bands increased and then stabilized, and those in the red-edge band and near-infrared bands first increased slowly and then decreased rapidly because an increasing number of pixels were influenced by the surrounding area; thus, the shapes of the spectral curves of the sample area became increasingly similar to those of the surrounding area. This study provides a guide for selecting an appropriate spatial resolution for UAV remote sensing to improve operational efficiency. The reflectance and geometric quantitative losses at different spatial resolutions are conducive to parameter inversion in quantitative remote sensing and spatial resolution transformation and enrich the knowledge of low-altitude UAV hyperspectral remote sensing.