Infrared hyperspectral imagery gives new opportunities for night observations for military, or security purposes, and for
geological studies as rocks have specific infrared absorption bands. Generally, an optimized utilization of spectral
information requires to retrieve spectral emissivity, which involves atmospheric compensation and surface temperature
and emissivity separation (TES). This paper presents a new method dedicated to a future airborne hyperspectral sensor
that will operate in the 3-5.5 and 8-12 µm spectral ranges, at 2.2 km height. It combines neural networks in order to
characterize the required parameters for atmospheric compensation and a spectral smoothness approach for TES. The
network training is performed with radiance spectra simulated with MODTRAN4, and using ASTER emissivities, and
the TIGR atmospheric database. A sensitivity study based on experimental design is carried out in order to compare
impacts of atmospheric and surface parameters on radiance at several wavelengths. Atmospheric compensation and TES
methods are then presented and their accuracy is assessed. Sensitivity of the retrievals to instrumental characteristics
such as signal to noise ratio and radiometric calibration, is also studied.
The retrieval of surface emissivity and temperature from infrared radiances measured by an airborne hyperspectral sensor closely depends on the ability to correct the acquired data from atmospheric effects. In this paper we present a new atmospheric correction scheme based on sounding techniques and neural networks. A key problem of neural network is to select relevant entries and outputs. Therefore, a preliminary sensitivity analysis that takes into account atmospheric conditions as well as the surface emissivity and temperature variations is carried out. It shows that only the first three or four PCA coefficients of atmospheric profiles have a significant influence on the radiance measured in the 4.26 μm carbon dioxide and the 6.7 μm water absorption bands. But these coefficients allow to rebuilt temperature and water profiles with enough accuracy for the addressed problem. This lead us to develop two groups of neural networks, the first one to estimate the main PCA coefficients of temperature profile, and the second one to retrieve the related water PCA coefficients. The atmospheric profiles thus obtained are then used to derive the "ground" radiances. Eventually we evaluate the accuracy of surface temperature and emissivity obtained with the derived atmospheric profiles.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.