Evaluation of cross-track illumination in EO-1 Hyperion imagery for lithological mapping

Hyperspectral remote sensing data is a powerful tool for discriminating lithological units and for the preparation of mineral maps for alteration studies. The spaceborne hyperspectral Hyperion sensor, despite its narrow swath width (similar to 7.5 km), possesses great potential with its 196 channels within the wavelength range 426.82-2395.50 nm. Although it has many advantages such as low cost and on-demand coverage, much uncertainty exists in the utility of its applications. For example, poor signal-to-noise ratio, the presence of sensor-specific defects and thicker atmospheric column due to its spaceborne platform makes certain environmental and geological applications difficult or impossible. In this article we demonstrate these calibration-related uncertainties, which are manifest from the preprocessing stage to the classification stage. In addition, the intimate mixing of minerals within specific targets, for example within individual outcropping lithological units or endmembers, adds uncertainty to our spectral discrimination results. The aim of this study was to develop and evaluate an approach for geological mapping of outcrops with Earth Observing- 1 (EO-1) Hyperion data. Atmospheric corrections and correction for cross-track illumination (CTI) variations (smile) were determined at different wavelength regions: the visible-near-infrared (VNIR; 420-1000 nm) and shortwave infrared (SWIR; 1000-2400 nm) regions. Our methodology was tested in a selected site at Central Anatolia, Turkey containing minimal vegetation cover. The results obtained from the image analyses were then compared and assessed with field observations and spectral measurements.