GEOSCAN Search Results: Fastlink


TitleThe impact of spectral band characteristics on unmixing of hyperspectral data for monitoring mine tailings site rehabilitation
DownloadDownloads (Preprint)
LicencePlease note the adoption of the Open Government Licence - Canada supersedes any previous licences.
AuthorLévesque, J; Staenz, K; Szeredi, T
SourceCanadian Journal of Remote Sensing vol. 26, issue 3, 2000 p. 231-240,
Alt SeriesEarth Sciences Sector, Contribution Series 20042784
PublisherInforma UK Limited
Mediapaper; on-line; digital
File formatpdf
AreaCopper Cliffe Mine; Sudbury
Lat/Long WENS -81.5000 -81.0000 46.5000 46.0000
Subjectsremote sensing; mapping techniques; Compact Airborne Spectrographic Imager (CASI); IKONOS 2
Illustrationstables; flow charts; graphs; satellite images; histograms
Released2014 07 31
AbstractThis paper demonstrates the impact of band characteristics on spectral unmixing of Compact Airborne Spectrographic Imager (casi) reflectance data acquired in the visible near-infrared spectral range over the Copper Cliff mine tailings site near Sudbury (Ontario, Canada). Spectral unmixing is used to monitor the rehabilitation status. For this purpose, the bands were reduced systematically from 65 to 33, 17, 8 and 4 bands using the original casi bandwidth of 8.7 nm full width at half maximum (FWHM). An interpolated data set using a cubic spline was derived from the 65-band casi spectra to study the effect of the bandwidth. The bandwidth at FWHM was varied between 8.5 nm and 76.5 nm in increments of 8.5 nm for six selected band positions across the casi wavelength range. High spatial resolution Ikonos 2 multispectral sensor image data with less bands and larger bandwidths than casi were simulated in order to investigate the combined effect of number of bands and bandwidth. These data cubes were unmixed with a constrained linear approach involving the endmembers green vegetation, lime, oxidized tailings, and two different water types. A comparison of the unmixing results indicates a mean absolute difference of up to 0.02 (standard deviation: ± 0.05) between the endmember fractions of the simulated data versus the 65-band benchmark data if the bands are selected according to physical spectral properties. The errors are much larger if the bands are not positioned properly. In general, the unmixing of the simulated data, including the four broad-band Ikonos simulation case, produced similar results as with the full 65-band casi as long as the bands are positioned with respect to physical spectral properties. Such a result, especially for the broad-band simulations, was achieved mainly because the endmembers are spectrally very distinct and do not show subtle differences.

Date modified: