REMOTE SPECTRAL MAPPING OF REGOLITH IN THE OLARY DOMAIN Ian C. Lau1, Alan J. Mauger2, Graham Heinson1 and Patrick R. James1 1 CRC LEME, School of Earth and Environmental Science, The University of Adelaide, North Terrace, Adelaide, SA 5005 2 PIRSA, Office of Minerals and Energy Resources, GPO 1671 Adelaide SA 5001 INTRODUCTION Until recently there has been little activity on the Olary Domain with regard to detailed regolith-landform mapping (Brown & Kernich 2002, Crooks 2002, Lawie 2001), with only regional studies being performed by Gibson (1996 & 1999) and Skwarnecki et al. (2001). Over the last decade the adjacent Broken Hill Domain has been the location for a considerable amount of regolith-focused research, including 1:25000 scale mapping (de Caritat et al. 2000) and more regional studies (Hill 2001, Gibson & Wilford 1996,). All of these studies have used digital imagery and geophysics to aid interpretation, demonstrating the value of remote sensing assisting in regional to prospect scale mapping projects. The acquisition of hyperspectral datasets with extensive coverage, such as the Musgrave (Stamoulis et al., 2001; Mauger et al. 2002) and Broken Hill (Robson et al. 2003) projects has brought an increased opportunity to provide regional mineral maps as a tool to aid company exploration. Robson et al. (2003) published preliminary interpretations of lithological mapping with stratigraphic, regolith, alteration, iron oxide distribution and mineral information being extracted from HyMapTM imagery over a large area of the Broken Hill Domain. Multispectral ASTER data has also been used to produce mosaiced mineral maps on a regional scale from the Curnamona Province (Hewson et al. 2003). In November 1998 HyVista Corporation acquired five overlapping 30km by 5km strips of HyMapTM data commissioned by MIM Exploration over the Olary and Mingary 1:100 000 map sheets. The 300 km2 of hyperspectral data included the White Dam Cu-Au-Mo Prospect, the Green & Gold and Wilkins Cu-Au workings as well as a range of regolith- landforms and rock types characteristic of the region. This project aims to investigate the spectral and mineralogical properties of the regolith around the White Dam Prospect using drill and surface material as well as evaluating the mapping potential of the hyperspectral imagery. LOCATION AND GEOLOGY The region of study is located approximately 25km north-east of Olary and is constrained by the coverage of the hyperspectral imagery (figure 1). The area contains a wide variety of landforms as well as regolith and geological units, extending from basement rocks of the Willyama Supergroup around the White Dam Prospect, south-west over the Barrier Highway to the MacDonald Corridor shear-zone and into a region of younger, Adelaidean rocks. The highly to moderately weathered Palaeoproterozoic Willyama Supergroup metasediments and felsic intrusive rocks have undergone five phases of deformation, where as the Neoproterozoic to Palaeozoic Adelaidean metasediments in the southern regions of the imagery, are less deformed and generally only slightly weathered. The basement rocks occur as inliers between Tertiary to Recent alluvial and colluvial sediments, which dominate the low lying areas. FIGURE 1. Locality map of region of study. METHOD BHEI gamma-ray data were used as a preliminary mapping tool to classify suspected lithologies and regolith-landform units. Further analysis of these classes was performed using a 25m digital elevation model (DEM) and digital orthoimagery, commissioned by the South Australian Government, to form part of a regolith-landform interpretation map product. CSIRO developed HYCORR software was used to atmospherically correct the five strips of HyMapTM data. The results from the initial correction were found to be sufficient for lithological discrimination but was not adequate for quantitative analysis of the mineral spectra. An improved atmospheric correction was performed using a combination of model based software (HYCORR) and an Empirical Line method using field and laboratory measurements of samples from bright and dark targets within the imagery. Vegetation and other unwanted pixels in the re-corrected imagery were masked to remove redundant data and to improve the unmixing process. Mineral abundance maps were constructed using mixture tuned match filtering (Harsanyi & Chang, 1994) following CSIRO techniques from the masked imagery (Quigley, 2001). The mineral maps were integrated with the gamma-ray derived data to improve the regolith-landform interpretation map. Field sampling was performed to validate the hyperspectral imagery results by analysing with an Analytical Spectral Devices (ASD) spectrometer and X-ray diffraction (XRD) of selected samples. RESULTS The gamma-ray derived classes were found to discriminate areas of exposed bedrock, which correlated with topographic highs observed in the DEM. The less altered Adelaidean rocks displayed a distinctive subdued radiometric response with little variation between lithological units. The older, deformed Palaeoproterozoic metasediments and intrusive rocks displayed characteristic responses, enabling preliminary lithological mapping and discrimination. Regolith-landform units were also able to be distinguished due to differing moisture content in alluvial regions and the signature strength in relation to the depth of the basement. Three hundred rock and soil samples were collected from various regolith-landform units found in the study region and spectrally measured using an ASD Fieldspec and a Portable Infrared Mineral Analyser (PIMA) instruments, which were referenced back to the airborne hyperspectral data. Preliminary results from the surface samples have identified kaolinite, white micas, hematite, goethite, smectite and carbonate minerals. The processed HyMapTM data produced comparable results to field samples. The predominant minerals found in the re-corrected data consisted of kaolinite, illite and muscovite as well as smectite (montmorillonite), hematite and goethite. Carbonate and MgOH minerals were not found due to the dominance of dry vegetation in the region causing broad absorptions in the 2.3µm region and low signal-to-noise over these wavelengths. Geo- rectified mineral maps were produced for the region around the White Dam Prospect for each HyMapTM strip. Difficulties were found with the production of seamless mosaics due to differing scene statistics and characteristics. FUTURE RESEARCH The study provided useful information for targeting field sites for a detailed follow up investigation involving XRD analysis and spectral measurements of subsurface regolith materials of drill-hole material from the White Dam Project area. Ongoing research is required on the relationship of the hyperspectral imagery to the information extracted from the airborne gamma-ray data and their role in understanding the regolith. It is anticipated that changes in the mineral chemistry observed in the spectra and XRD analysis will be reflected in the radiometric dataset. Further work is required to determine the validity of the atmospheric correction and subsequent results from the HyMapTM imagery through ASD spectrometer and XRD analysis of field samples. Technical difficulties regarding multi-swath hyperspectral data require further attention to allow the generation of seamless mosaiced mineral maps. ACKNOWLEDGMENTS The authors would like to acknowledge the support of the following persons and organisations for their support and collaboration on this project: CSIRO Exploration and Mining, CRC LEME, HyVista Corporation, Primary Industries and Resources of South Australia, the New South Wales Department of Mineral Resources, Geoscience Australia, Mount Isa Mines Exploration and Polymetals Ltd. REFERENCES Brown, A.D. & Kernich A., 2002. Luxemburg Regolith-landform map, CRC LEME. Unpublished report. Crooks, A., 2002. Regolith mapping on the Mingary 1:100,000 map area, Curnamona Province. MESA Journal. 25; Pages 42-45. de Caritat, P., Gibson, D., Hill, S.M., Killick, F., Lavitt, N., Papp, E., Tonui, E., Brachmanis, J., Dann, R., Debenham, S., Foster, K., Hill, L.J., Holzapfel, M., Jones, G.L., Maly, B.E.R., Senior, A., Shirtliff, G., Tan, K.P., Turner, M., West, D.S. & Willis, S.M. 2000. Regolith research in the Broken Hill region: an overview of CRC LEME activities. In Peljo, M. 2000. Broken Hill Exploration Initiative: Abstracts of papers presented at the May 2000 Conference in Broken Hill. AGSO Record 2000/10. Pages 12-15. Gibson, D.L., 1996 Curnamona 1:500,000 regolith-landform map. CRC LEME, Perth. Gibson, D.L. 1999. Explanatory notes for the Broken Hill and Curnamona province 1:500 000 regolith landform maps. CRC LEME Open File Report 77. Gibson, D.L., & Wilford, J., 1996, Broken Hill 1:500,000 regolith-landform map. CRC LEME, Perth. Hewson, R., Mah, A., Dunne, M. & Cudahy, T.J., 2003. Mapping Mineralogical and structural relationships with satellite-borne ASTER and airborne geophysics at Broken Hill. . ASEG 16th Geophysical Conference and Exhibition, Adelaide. February 2003, CD ROM. Harsanyi, J.C., & Chang, C.I., 1994, Hyperspectral image classification and dimensionality reduction: Anorthogonal subspace projection approach: IEEE Transactions on Geoscience and Remote Sensing, v. 32, Pages 779–785. Hill, S.M., 2001 Broken Hill Domain 1:100,000 regolith-landform map. CRC LEME, Perth. Mauger, A.J., Stamoulis, V. & Cocks, P.A., 2002. Hyperspectral airborne survey for geological mapping, Musgrave Ranges, South Australia. 5th International Airborne Remote Sensing Conference, Miami, Florida, May 2002. Pages 22-24, Lawie, D.C., 2001. Exploration geochemistry and regolith over the northern part of the Olary Domain, South Australia, Unpublished Ph.D thesis, University of New England. Quigley, M., 2001. Mineral Mapping. Data & Image Processing Workshop, 22 nd February 2001, 1st Geological Hyperspectral Focus Group, Adelaide, SA, Presented by Cudahy, T., Quigley, M. & Mason, P., CSIRO Exploration and Mining. Robson, D., Cocks, P. & Taylor, G., 2003. Calibration, processing and interpretation of hyperspectral data over the Broken Hill region. ASEG 16th Geophysical Conference and Exhibition, Adelaide. February 2003, CD ROM. Skwarnecki, M.S., Li Shu & Lintern, M.J., 2001. Geochemical dispersion in the Olary District, South Australia: Investigations at Faugh-a-Ballagh Prospect, Olary Silver Mine, Wadnaminga Goldfield and Blue Rose Prospect, CRC LEME Open File Report 113. Stamoulis, V., Mauger A.J. & Cudahy, T.J., 2001. Mapping mineral abundances of the Musgrave Block, South Australia, using airborne hyperspectral VNIR-SWIR data. In: IGARSS 2001; Scanning the present and resolving the future; proceedings. Milne-Tony (chairperson); Cechet-Bob (chairperson) International Geoscience and Remote Sensing Symposium. 2001, Vol. VII; Pages 3169-3171. 2001. Institute of Electrical and Electronics Engineers. New York, NY, United States. 2001.