GEOSCAN Search Results: Fastlink


TitleScene-based spectral response function shape discernibility for the APEX imaging spectrometer
AuthorBrazile, J; Neville, R A; Staenz, K; Schläpfer, D; Sun, L; Itten, K I
SourceIEEE Geoscience and Remote Sensing Letters vol. 3, no. 3, 1658017, 2006 p. 414-418,
Alt SeriesNatural Resources Canada, Contribution Series 20181653
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Mediapaper; on-line; digital
File formatpdf
Subjectsgeophysics; remote sensing
ProgramCanada Centre for Remote Sensing Divsion
AbstractScene-based spectrometer calibration is becoming increasingly interesting due to the decreasing cost of computing resources as compared with laboratory calibration costs. Three of the most important instrument parameters needed for deriving surface reflectance products are per-band bandwidths, i.e., full-width at half-maximum, band centers, and spectral response function (SRF) shape. Methods for scene-based bandwidth and band center retrieval based on curve matching in the spectral regions near well-known solar and atmospheric absorption features have been investigated with satisfying results. The goal of this work is to establish the feasibility of per-band SRF shape discernibility. To this end, at-sensor radiances in multiple application configurations have been modeled using Moderate-Resolution Atmospheric Transmission (MODTRAN) 4 configured for the currently being built Airborne Prism Experiment (APEX) imaging spectrometer in its unbinned configuration (i.e., optimized for spectral resolution). To establish SRF shape discernment feasibility, per-band MODTRAN 4 spectral "filter response function" files have been generated for five common theoretical shapes using APEX nominal bandwidth and band center specifications and are provided as MODTRAN 4 input for the instrument model. In several application configurations, the typically used Gaussian SRF is used as reference and compared with radiances resulting from hypothetical instruments based on the four other shapes to detect differences in selected spectral subsets or "windows" near well-known Fraunhofer features. A relative root-mean-square metric is used to show that discernment in some cases is directly feasible, and in others, feasible if noise reduction techniques (e.g., along-track averaging of homogeneous targets) are possible.

Date modified: