Maximizing the quantitative utility of airborne hyperspectral imagery for studying plant physiology: An optimal sensor exposure setting procedure and empirical line method for atmospheric correction

Abstract

Proper calibration of airborne hyperspectral imagery is essential for maximizing the quantitative utility of remotely-sensed imagery, especially when distinguishing subtle changes in spectral curves related to specific plant physiological properties (e.g. chlorophyll and water). Many studies use the empirical line approach with reference reflectance taken from dark and bright targets to calibrate airborne images. However, few have paid attention to the issue of sensor oversaturation due to the exposure setting of the imaging sensor, and no studies have investigated the effects this has placed on image calibration. With limited radiometric resolution, a sensor would become saturated by energy reflected from bright targets when its exposure is set to maximize signals reflected from a feature of interest, for example vegetation. This would result in large bias in the reflectance calibration process, and should be addressed for enormous amounts of high spatial and spectral resolution data that have been increasingly taken by manned or unmanned aircraft. In this study, we test the exposure setting of a hyperspectral sensor for maximizing vegetation signal and investigate potential reference targets in an airborne scene, and propose a more suitable airborne hyperspectral imaging and an empirical line atmospheric correction procedure by taking into account: 1) imaging sensor exposure setting, 2) spectral extrapolation, 3) sensor saturation of targets’ signal, and 4) optimal materials and grey levels for field reference reflectance for the empirical line method. The imaging experiment was conducted over a grassland field with the Micro-Hyperspec VNIR sensor. Using field hyperspectral data to validate the calibration results, we found that our proposed empirical line calibration approach improved the reflectance accuracy significantly. Vegetation indices calculated from the calibrated spectra were able to estimate chlorophyll content with success. Our work offers insights into image calibration and describes a feasible method to maximize quantitative utilities of airborne Hyperspectral imagery for vegetation studies.