Performance assessment of onboard and scene-based methods for Airborne Prism Experiment spectral characterization

Petra D'Odorico; Luis Guanter; Michael E Schaepman; Daniel Schläpfer

doi:10.1364/AO.50.004755

Performance assessment of onboard and scene-based methods for Airborne Prism Experiment spectral characterization

Appl Opt. 2011 Aug 20;50(24):4755-64. doi: 10.1364/AO.50.004755.

Authors

Petra D'Odorico¹, Luis Guanter, Michael E Schaepman, Daniel Schläpfer

Affiliation

¹ Remote Sensing Laboratories, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland. petra.dodorico@geo.uzh.ch

PMID: 21857698
DOI: 10.1364/AO.50.004755

Abstract

Accurate spectral calibration of airborne and spaceborne imaging spectrometers is essential for proper preprocessing and scientific exploitation of high spectral resolution measurements of the land and atmosphere. A systematic performance assessment of onboard and scene-based methods for in-flight monitoring of instrument spectral calibration is presented for the first time in this paper. Onboard and ground imaging data were collected at several flight altitudes using the Airborne Prism Experiment (APEX) imaging spectrometer. APEX is equipped with an in-flight characterization (IFC) facility allowing the evaluation of radiometric, spectral, and geometric system properties, both in-flight and on-ground for the full field of view. Atmospheric and onboard filter spectral features present in at-sensor radiances are compared with the same features in reference transmittances convolved to varying instrument spectral configurations. A spectrum-matching algorithm, taking advantage of the high sensitivity of measurements around sharp spectral features toward spectrometer spectral performance, is used to retrieve channel center wavelength and bandwidth parameters. Results showed good agreement between spectral parameters estimated using onboard IFC and ground imaging data. The average difference between estimates obtained using the O(2) and H(2)O features and those obtained using the corresponding filter features amounted to about 0.3 nm (0.05 of a spectral pixel). A deviation from the nominal laboratory instrument spectral calibration and an altitude-dependent performance was additionally identified. The relatively good agreement between estimates obtained by the two approaches in similar spectral windows suggests they can be used in a complementary fashion: while the method relying on atmospheric features can be applied without the need for dedicated calibration acquisitions, the IFC allows assessment at user-selectable wavelength positions by custom filters as well as for the system on-ground.