Scale-Wavelength Decomposition of Hyperspectral Signals - Use for mineral classification & quantification in a reflective-thermal study of a lignite mining site -

Abstract

Intense lignite mining led to several environmental problems. One is acid mine drainage (AMD), the acidification of soils and waters due to sulphide mineral weathering and mobilisation of heavy metals. The Thesis aims at the spatial mapping of AMD relevant minerals. Data acquisition encompassed a hyperspectral airborne survey in the reflective (HyMap) and a multispectral survey in the thermal (AHS) domains, coupled with spectroscopic reference measurements and laboratory analysis of collected samples. Overlaying spectral absorption features in mineral or soil mixtures can result in very complex spectral signatures. Thus, a multi-scale analysis of spectra at several levels-of-detail and more versatile feature describing parameters would be beneficial. This can be realised by the continuous wavelet transform (CWT). The thesis develops two spectral analysis methods on this multi-scale decomposition approach: The Wavelet Transform Feature Analysis for quantification of mixture components and the Wavelet Coherence Mapper for classification of selected minerals.