Mineralogical mapping using airborne imaging spectrometry data

Abstract

With the development of airborne, high spectral resolution imaging spectrometers, we now have a tool, that allows us to examine surface materials with enough spectral detail to identify them. Identification is based on the analysis of position and shape of absorption features in the material spectra in the visible and infrared (0.4µm to 2.5µm). These absorption features are caused by the interaction of Electro-Magnetic Radiation (EMR) with the atoms and molecules of the surface material. Airborne data were collected to evaluate these new high spectral resolution systems. The data quality was assessed prior to processing and analysis and several problems were noted for each data set (striping, geometric distortion, etc.). These problems required some preparation of the data. After data preparation, data processing methods were evaluated, concentrating primarily on the log residuals and hull quotients methods. The processing steps convert the data to a form suitable for analysis. The data was analysed using the Spectral Analysis Manager (SPAM) package, developed by JPL. Two Imaging spectrometers were evaluated. The AIS - 1 instrument was flown over an area in Queensland, Australia. Ground data and laboratory work confirmed the presence of anomalous areas detected by the instrument. The data quality was poor and only basic classification of the data was possible. Anomalies were classed as "GREEN VEGETATION", "DRY VEGETATION", "CLAY" or "CARBONATE" based on the position of the major absorptions observed. The second instrument, the GER - II was flown over an area of Nevada, USA. Ground data and laboratory work confirmed the presence of the anomalies detected by the instrument. The data quality was somewhat better. Identification of sericite, dolomite and illite was possible. However, most of the area could still only be classed in the broad groupings listed above. To conclude, the effectiveness of identification is limited to a large degree by the poor data quality. If the data quality can be improved, techniques can be applied to automatically locate and identify material spectra, from the airborne data alone.