TECHNICAL PAPERS

GEOSCAN AIRBORNE MULTISPECTRAL SCANNERS AS EXPLORATION TOOLS FOR WESTERN AUSTRALIAN DIAMOND AND GOLD DEPOSITS

In 'Geophysical Signatures of Western Australian Mineral Deposits', Geology and Geophysics Department (Key Centre) & UWA Extension, The University of Western Australia, Publication No.26, 1994, pages 435-447.

The use of remote sensing in mineral exploration has evolved from basic photogeology to interpretation of more sophisticated satellite and airborne mutispectral data sets. Although the mineral mapping capabilities of Geoscan airborne multispectral scanners have been demonstrated for well exposed and arid terrains, the question remains as to their effectiveness in deeply weathered regimes such as Western Australia.

Geoscan airborne multispectral data have been collected for a number of Western Australian mineral deposits and prospects. The applicability of the data to exploration for similar deposits, based upon a series of simple image processing treatments derived according to the known reflectance spectra of associated alteration minerals, has been assessed.

Spectral characteristics of ultramafic rock such as kimberilte or lamproite are recognisable at Blue Well, but not over the Aries kimberlite pipe which has a spectral response dominated by the reworking and removal of desert varnish from detritus derived from its surrounding sandstone host. Shear zone hosted gold mineralisation in the deeply weathered Yilgarn Block is characterised by spectral signatures related to the presence of sericite, silica and iron oxides. The same minerals and spectral variations are also apparent in Carlin style mineralisation at Kazput Pool.

In each example studied, simple image processing treatments are used to enhance the spectral characteristics of the deposit. These signatures are dominated by the present day surface mineralogy of the deposit and are the result of the interaction between hydrothermal and supergene alteration. In an exploration mode, a method has been successfully developed and cost effectively appiled for first pass testing for the basic spectral criteria of each geological model, followed by more detailed spectral discrimination of key mineral assemblages and early geochemical follow-up.

1.0 INTRODUCTION

The use of remote sensing mineral exploration began some 60 years ago with hand held cameras being pointed out of aircraft windows, and has since evolved through stereoscopic aerial photography to sophisticated space age technology and satellite borne, multispectral digital imaging systems. When the Landsat multispectral scanner (MSS) began operations in the early 1970s, remote sensing geology took an enormous leap forward but still functioned largely by way of photogeological interpretation of hard copy. Since then, the Landsat Thematic Mapper (TM) instruments have provided the geological user with information relating to iron oxides and clays, but the data remain coarse and of general purpose with low spectral and spatial resolution, requiring sophisticated statistical processing techniques not readily understood by the average geological user.

In the early 1980s, an Australian exploration company, Carr Boyd Minerals Ltd, recognised the limitations of available and planned satellite remote sensing systems as applied to the needs of mineral exploration, and combined with the CSIRO to develop an airborne system specifically for these needs. Geoscan Pty Ltd grew from this research project and developed its Mkl Airborne MultiSpectral Scanner (AMSS) in 1984. Since then, Geoscan has designed and developed a second AMSS system also specifically for use in mineral exploration. Both the Mkl and Mkll AMSS systems are capable of producing high spatial resolution, remotely sensed data (pixel sizes from 3 to 20 m) and each has an enhanced spectral resolution. The Mkl records data in 13 channels configured as in Figure 1 and Table 1, whereas the MklI records in 24 channels (Fig. 2, Table 1). Importantly, both instruments collect spectral information from three parts of the electromagnetic spectrum: the visible and near infrared (VNIR), short wave infrared (SWIR), and thermal infrared (TIR); providing information relating to iron oxides, clays and carbonates, and silicates, respectively (Figs 1 & 2).

Figure 1: Selected mineral and rock spectra relative to Geoscan AMSS MkI band positions in: (a) the visible near infra-red (VNIR), (b) the short wave infra-red (SWIR), (C) the thermal infra-red (TIR).	Figure 2: Selected mineral and rock spectra relative to Geoscan AMSS MkII band positions in: (a) the visible near infra-red (VNIR), (b) the short wave infra-red (SWIR), (C) the thermal infra-red (TIR).

The use of Geoscan AMSS instruments for mineral exploration has been reported in a number of studies (Honey & Daniels, 1986; Gunn et al., 1989; Derriman & Agar, 1990; Lyon & Honey, 1990; Huntington et al., 1991; Agar, 1992). However, there is limited published work on geological remote sensing in the deeply weathered terrains of Western Australia (Drury & Hunt, 1988,1989; Lyon, 1990; Fraser, 1991). This paper shows how spectral information gathered by the Geoscan 13 channel MkI or 24 channel Mkll AMSS instruments can be used to identify and map known mineral prospects, and evaluates the potential of this technology as a primary exploration tool in Western Australia.

TABLE 1: Geoscan AMSS Mkl and MkIl band specifications.

2.0 DATA ACQUISITION, CHARACTERISTICS AND PROCESSING

Before embarking upon a detailed evaluation of Geoscan data, it is important to understand the basic principles behind its acquisition because they impact on data processing techniques. Reflected or emitted electromagnetic radiation reaching the AMSS is reflected by a high speed rotating mirror onto the detectors. The gain and offset for each detector are set during a trial run over the survey area to maximise the sensitivity of the system. The offset is adjusted to produce a mid range digital number (DN) of 127 and the signal gain or amplification is set to maximise the 8 bit dynamic range of 0 to 255 so that nothing of interest is saturated (DN = 255) or below detection (DN = 0).

Both Geoscan AMSSs are characterised by a high signal to noise ratio (SNR) which allows their data to be interpreted without pre-processing. The DN recorded for each channel in each pixel comprises three components: the reflected energy, an atmospheric backscatter component, and an additive electronic component. Atmospheric effects on the data are corrected statistically. The additive electronically induced effects are uniform for each channel within any spectrometer and are effectively eliminated by the use of straightforward mathematical difference or ratio treatments of bands within the same spectrometer. Thus, in testing for the mineral kaolinite, which has a spectrum showing a pronounced reflectance minimum coincident with Geoscan AMSS Mkll band 14 (Fig. 2), treatments 13-14 or 15-14 could be applied and both would produce relatively large values wherever kaolinite was present. After differencing or ratioing, the processed data are adjusted to maximise the dynamic range of the data so that large positive values will approach DN=255 and be bright, whereas large negative values will be close to DN=0 and dark.

By displaying three separate band treatments through the three colour guns (red - R, green - G, blue - B) of an image processor, it is possible to test three spectral relationships simultaneously, and thus sequentially eliminate other minerals that may have some common features. For example, both sericite and kaolinite have a common absorption feature at 2.2 um (Geoscan AMSS Mkll band 14, Fig. 2) but in the case of kaolinite, the feature has a strong left hand asymmetry centred on band 13 such that displaying band diflerences 13-14 16-14 13-16 (RGB) would show sericite as white (combination of strong positive in all colours) and kaolinite as yellow-green (slight positive in red, strong positive in green, strong negative in blue).

3.0 DIAMOND EXPLORATION

3.1 ARIES KIMBERLITE PIPE

The Aries kimberlite pipe is located in the Phillips Range, central Kimberley (Fig. 3). It is a diamondiferous pipe which was discovered by traditional drainage sampling methods by a Triad-Freeport joint venture in 1985. The surface expression of the kimberlite comprises three depressions of up to 20 m depth which are clearly visible on aerial photographs but which are indistinguishable from any other depression (Fig. 4a). However, both the kimberlite and drainage "tail" are clearly visible in selected treatments of Geoscan Mkl AMSS data (Towie et al., 1991).

The "spectral signature" of the pipes referred to by Towie et al. (1991) is shown in Figure 4b and was interpreted initially as showing the presence of montmorillonite at the surface because of relative absorption in band 8, as indicated by the white response against a dark blue background in treatment 9/8 7/8 6/8 RGB. However, in reality, the signature is not that simple as a reference to the SWIR spectrum for montmorillonite relative to Mkl AMSS band positions will indicate (Fig. 1). Although the data do indicate an absorption feature around band 8, the ratio of band 9/ 8 should be less than one and hence dark on the image. By generating a series of simple difference treatments to examine the relative reflectance across the four SWIR bands for the area over the kimberlite, an approximate spectral shape can be generated which bears little or no resemblance to any of the typical clay spectra shown in Figure 1. Indeed, when the reflectance in bands 7, 8 and 9 are viewed relative to band 6 in a treatment which wouid highlight any, or all, of the common clay minerals ( 6-7 6-8 6-9 RGB, Fig. 4c), the kimberlite remains dark. This suggests that there are few or no clay minerals at the surface.

Figure 3: Location map of Western Australia showing localities referred to in text.

Vegetal materials including cellulose, protein and lignin all have absorption features in the SWIR and care must be taken not to confuse their responses with those of the kimberlite lobes. By using band 4, located at 0.85 um over a part of the VNIR spectrum where vegetation is highly reflective, the distinctive SWIR response can be compared with the distribution of vegetation in the area. Figure ~ (treatment 9/8 7/8 4 RGB) demonstrates that the spectral signature over the kimberlite lobes comprises an element of vegetation (white) but also a significant portion including the drainage "tail" which is yellow and therefore not a vegetal effect. The kimberlite's host sandstones are bright in Figure ~ (treatment 6-7 6-8 6-9 RGB), indicating a high surface clay content, probably kaolinitic, which most likely reflects the presence of a kaolinite based desert varnish covering the sandstone (Lyon, 1990). Desert varnish is also typically high in iron-oxide material and this can be checked spectrally using the Mkl AMSS VNIR bands 1 to 5. Figure 4e shows a treatment 3-2 3-4 3-5 RGB which is designed to enhance iron oxide minerals and, in this case, where most of the country rocks and soils appear to be relatively rich in iron oxide, the kimberlite lobes show little if any response at all, further suggesting that they are not coated in desert varnish.

Field evaluation of the above spectral interpretation, supported by laboratory spectral and XRD analysis of field samples, showed that the surrounding sandstones were indeed coated with a kaolinitic desert varnish and that the kimberlite lobes were covered by a quartz rich sandy soil with very little clay content (Hatch, 1991; Hatch et al., 1992). Thus, although the Geoscan Mkl AMSS data do clearly distinguish the Aries kimberlite lobes, they do so indirectly as a result of the scanner's ability to recognise a clean, desert varnish free detrital sand covering the preferentially eroded kimberlite, and derived from the kimberlite's own host rock.

Figure 4: Geoscan AMSS MkI imagery over the Aries kimberlite pipe showing treatments: (a) 3 2 1 RGB (simulated true colour) (b) 9/8 7/8 6/8 RGB (c) 6-7 6-8 6-9 RGB (d) 9/8 7/8 4 RGB (e) 3-2 3-4 3-5 RGB, relative to (f) geological map of the image area to the same scale.

3.2 BLUE WELL PROSPECT

The Blue Well diamond prospect is located on Sturt Meadows Station, northwest of Leonora in the Yilgarn Cralon of Western Australia (Fig. 3). The prospect developed from a speculative regional Geoscan AMSS Mkll survey flown by Australian Ores and Minerals Ltd over the whole of the Leonora 1:250 000 map sheet in November 1989. Systematic first-pass appraisal of all the data was carried out using standard RGB treatments 17 8 2 and 11-16 11-17 11-18 which enhance ultramafic rocks as deep blue and white, respectively. The Blue Well prospect stood out in this first-pass evaluation as a large, subcircular feature with an apparent dispersion train away to the northeast (Fig. 5b-c) in an area where ultramafic rocks had hitherto not been recorded.

The aim of follow up image processing using the SWIR treatment 13-14 13-15 13-16 RGB (Fig. 5d) was to discriminate montmorillonitic and kaolinitic clays. The strong white colour of the subcircular feature implies that the clay present is montmorillonite, and the dark background in both SWIR treatments (Fig. 5c-d) indicates that there is a relative paucity of clay minerals in the surrounding areas. In the TIR, treatments 20 21 22 RGB and 23-20 22-20 21-20 (Fig. 2e-f) can be used to discriminate silica as cyan and white, respectively, as a consequence of the relatively low emission of quartz at 9.0 um (i.e. in the region of Geoscan AMSS Mkll band 20). In this case, the target area contrasts with its surroundings, which indicate an extensive siliceous surface covering.

Figure 5: Geoscan AMSS MkII imagery over the Blue Well prospect showing treatments: (a) 4 3 2 RGB (simulated true colour) with geological map (b)17 8 2 RGB (c) 11-16 11-17 11-18 RGB (d) 13-14 13-15 13-16 RGB (e) 20 21 22 RGB (f) 23-20 22-20 21-20 RGB.

Field checking of the target area showed it to comprise a fine sandy loam surrounded by sub cropping granite and silcrete partially covered by fine grained quartz sand (Fig. 5a). Sampling of the surface loam found no diamond indicator minerals but the prospect was nevertheless drilled on the strength of encouraging ground magnetic results (Australian Ores and Minerals, 1991). Two drillholes terminated in tremolite-actinolite-chlorite schists, geochemically equivalent to basaltic komatutes, after passing through a montmorillonite rich colluvium (Australian Ores and Minerals, 1991). Thus, the Geoscan AMSS Mkll was able to recognise the weathering products of a hitherto unrecognised ultramafic enclave within deeply weathered granites of the Yilgarn Craton. Although in this example the ultramafic rock discovered was not a kimberlite, the image processing principles are nevertheless valid as a targeting tool when applied to diamond exploration.

4.0 GOLD EXPLORATION

4.1 JILLAWARRA, ARCHEAN SHEAR ZONE HOSTED GOLD

The Jillawarra mining centre is situated within the arcuate Mingah Range Archaean greenstone belt in the Murchison Province of the Yilgarn Craton, 80 km WNW of Meekatharra (Fig. 3). The greenstone belt comprises a series of ultramafic, mafic, intermediate and felsic volcanic rocks and metasedimentary rocks, and a series of abandoned mine workings are aligned along an east west shear zone within the belt (Fig. 6). The area has been explored using airborne electromagnetics (INPUT), aeromagnetics and ground EM surveys in search of volcanogenic massive sulphide (VMS) base metal deposits and for gold using a regional heavy mineral sampling program (CRAE, 1988).

Figure 6: The geology of the Mingah Range greenstone belt and Jillawarra mining centre showing the location of the Geoscan imagery displayed in Figure 7. After CRA Exploration Pty Ltd (1988).

Geoscan AMSS Mkl data covering this area were processed as part of a regional reconnaissance gold exploration programme by Australian Ores & Minerals Ltd in 1989. The Jillawarra mining centre attracted attention because it stood out as a distinct linear zone defined in white on treatment 6-7 6-8 6-9 RGB, suggesting significant hydrothermal alteration along a shear zone. This is similar to the case at Hadleigh Castle, near Charters Towers, Queensland, where Geoscan data led to the discovery of a significant gold resource (Gunn et al, 1989; Ashton Mining Ltd, 1993). In this area, positive indications of silicification and ferruginisation were recognised using band differences 11-10 and 3-1, respectively. The Jillawarra area has since been reflown using the higher resolution AMSS Mkll which is discussed here.

Figure 7: Geoscan AMSS MkII imagery over the Jillawarra mining centre showing treatments: (a) 3 2 1 RGB (simulated true colour) (b) 17 8 2 RGB (c) 11-14 11-16 11-18 RGB (d) 13-14 13-15 13-16 RGB (e) 6-7 6-8 6-9 RGB (f) 20 21 22 RGB.

The Mkll treatment 11-14 11-16 11-18 RGB, which is the equivalent of the Mkl 6-7 6-8 6-9 RGB treatment, depicts the shear zone and mining centres as well as an area to the west of the mining centre as white (Fig. 7c). This other area was considered attractive as an exploration target because it appeared to show a continuation and splaying of the Jillawarra shear zone. It also showed a much broader zone of alteration (cyan and white colours, Fig. 7c) and a response in treatment 13-14 13-15 13-16 RGB indicative of a sericitic zone (white) and a propylitic zone (cyan) which is also evident over the abandoned workings of the mining centre (Fig. 7d). When viewed in a lithological discriminator such as 17 8 2 RGB (Fig. 7b), this target area appears as a uniform deep blue, indicative of ultramafic rocks which have been mapped in that area (CRAE, 1988). Interestingly, the main shear zone to the north of the splay is much narrower and suggests only propylitic alteration. In terms of prospectivity, the spectral information over the target area is further enhanced by the iron oxide discriminator 6-7 6-8 6-9 RGB (Fig. 7e), which shows the mining centre and broader zone along the splay to be strongly white and the main shear zone to be locally white, indicative of the presence of iron oxides. Furthermore, the treatment 20 21 22 RGB, which enhances silica as cyan, in addition to showing widespread cyan related to surface detrital material, also shows a clear linear zone of cyan, unrelated to drainage, along the splay structure with another similar zone along the main shear and associated with the indicated iron oxides.

Collectively, these spectral variations present an attractive exploration target when placed in their geomorphic and structural context. Geomorphologically, the target area is located in an erosional regime underlain by saprolitic material, yet the spectral information over the area shows a pattern and a mineralogy unrelated to the weathering profile, but consistent with a structural control related to a shear zone. The shear zone is known to be mineralised in parts, as evidenced by the abandoned workings, which show the same spectral characteristics as the exploration target area. Reconnaissance fieldwork on the new target identified sericitic alteration, silicification and the presence of jarosite within the shear zone, and returned encouraging assays of up to 14.5 g/t Au from sericite-jarosite schist with anastomosing quartz veins, and surface soil geochemical anomalies of between 10 and 100 times background (Perez, 1 990a).

4.2 KAZPUT POOL, PROTEROZIC CARLIN-STYLE MINERALISATION

The gold mineralisation at Kazput Pool in the Ashburton area of Western Australia (Fig. 3) has been likened to a Carlin style occurrence and has returned low grade gold values of around 1.5 g/t Au over modest intervals (6-9 m) in both trenches and drill core (Border Gold, 1993). The mineralisation is associated with pyritic and silicic alteration in dolomitic siltstones of the Duck Creek Dolomite and sandstones of the underlying McGrath Formation. Typical Carlin style mineralisation is characterised by an association with carbonate or carbonaceous rocks, orebodies which are conformable with stratigraphy and extend out from a steeply dipping fault, and relatively simple rock alteration, mainly silicification and argillisation (predominantly sericite and/or kaolinite) (Dickson et al., 1979).

Figure 8: Geoscan AMSS MkII imagery over the KazputPool prospect showing treatments: (a) 23 17 3 RGB (b) 11-14 11-15 11-16 RGB (c) 13-15 14-15 16-15 RGB (d) 13-14 13-16 16-15 RGB (e) 19 20 23 RGB, relative to (f) geological map of the image area to the same scale.

Exploration at Kazput Pool has delineated a geochemically anomalous zone, about 8 km in length, primarily along the contact between sandstones of the McGrath Formation and the overlying Duck Creek Dolomite (Fig. 8f; Border Gold, 1993). Geoscan AMSS Mkll data over the prospect show the contact between the sandstones (brown) and dolomites (cyan) in the lithological discriminator 23 17 3 RGB (Fig. 8a). TIR band 23 has been used specifically because carbonate and dolomitic rocks have relatively low emittance at that wavelength (11-12 um, Fig. 2c). A clear spectral anomaly is apparent in the SWIR band difference treatment 11-14 11-15 11-16 RGB, which shows the dolomitic rocks as deep blue, the sandstones as orange-brown, and an anomalous area of strong white and cyan surrounded by a diffuse white halo in the contact zone (Fig. 8b). Further investigation of the spectral anomaly using treatments 13-15 14-15 16-15 RGB and 13-14 13-16 16-15 RGB provides indications of the surface mineralogy in that area (Fig. 8c-d). The former treatment shows the outcropping anomalous area, or target zone, as white in a reddish halo, indicating relatively low reflectance in band 15 which suggests that sericite could be present (Fig. 2b). The latter treatment also shows the target area as white but with a yellow halo, confirming a sericitic mineralogy for the outcropping area, with a broadening of the absorption feature through bands 14 and 13 which is consistent with a montmorillonitic clay in the weathering products (Fig. 2).

Thermal data shows the dolomitic rocks as yellow in treatment 19 20 23 RGB (Fig. 8e), which also highlights silica as blue. There is quite clearly a large amount of silica present in association with the dolomitic rocks, either as a weathered patina or as detritus. However, close inspection of the spectral anomaly reveals linear blue features on relatively high ground associated with the SWIR spectral anomaly. These are unrelated to drainage, and therefore possibly reflect localised silicification or quartz veining.

Although discovered using regional stream sediment geochemistry, this prospect is clearly recognisable in AMSS data and can be shown to have a spectral signature and structural setting consistent with a Carlin style conceptual model. The work of Perez (1990b) shows how this, and several other similar prospects, were successfully targeted using the Geoscan AMSS Mkll as part of a regional reconnaissance exploration programme.

5.0 DISCUSSION AND CONCLUSIONS

The value of any technology or method to mineral exploration can only be determined according to its ability to provide required results in a cost effective manner. The use of remote sensing in mineral exploration has been largely restricted to geological mapping in the early stages of reconnaissance exploration, with only limited use as an exploration targeting technique. In very well-exposed arid terrains, clay-iron spectral signatures from Landsat TM imagery can be used to locate exploration targets (Loughlin, 1991; Abrams et al., 1983; Podwysocki et al., 1983). However, the low spectral and spatial resolution of Landsat TM data make it less effective in many situations, creating many false anomalies (Crosta & Rabelo, 1993) and missing smaller and more subtle alteration signatures (Coopersmith, 1991).

The examples described here demonstrate that a variety of ore deposit types from a range of weathering environments in Western Australia can be recognised using Geoscan data. Furthermore, important information about their mineralogy can be interpreted from the same data. Although it could be argued that two of these examples were discovered in hindsight, the treatments used are simple and designed to enhance specific key features that would reasonably be expected by an explorationist as indicators of mineralisation. In each case, the key to extrapolating from known to unknown is in the understanding of the likely surface expression of the target and testing for it without discounting other, potentially more obvious, possibilities. In the case of the Aries kimberlite, for example, to test the dataset simply for an exposed kimberlite (ultramafic rock) within sandstone would not have recognised the pipes. However, a treatment designed to test for clean, desert varnish free sand derived from the local sandstone, and deposited in hollows within it, clearly does work. Thus, the explorationist must be alert to the possibilities of targets having signatures somewhere between the two extremes. Undoubtedly, there will be some natural hollows filled in the same way which are not over eroded kimberlite. However, a combination of this technique with a search for coincident aeromagnetic anomalies, for example, would create a powerful exploration strategy for target generation and determining priorities.

At Aries, the weathered kimberlite does not show through the overburden. However, at Blue Well in the Yilgarn Craton, the montmorillonitic signature of the weathered ultramafic rock persists through a deep weathering profile to the surface and demonstrates the value of the data over areas of poor outcrop and extensive colluvium. Sericite, a common and very important alteration mineral, also persists through deep weathering profiles and, at Jillawarra, is discriminated within saprolite by the Geoscan data. Silica can also be persistent through deep weathering profiles but silicification can be confused with quartz float and silica sands. Careful interpretive work as at Jillawarra and Kazput Pool, can discriminate those signatures related to geological structures and/or alteration, from those of detrital origin. Indeed, recognition of subtle structures in association with key mineral assemblages can enhance the prospectivity of a target and provide a new dimension to a known mineralised system. This is evident at Jillawarra where the main shear zone was known and recognisable from abandoned workings and aeromagnetic data but a hitherto unrecognised splay, with a prospective hydrothermal alteration pattern, was interpreted from AMSS data and led to a new exploration prospect.

From a remote sensing perspective, mineral deposits can be described and broadly tested mineralogically in terms of their iron oxide, clay, carbonate and silica contents. Targets which satisfy the conceptual spectral parameters can then be viewed more critically, using the same data to determine probable mineral assemblages, relationships or patterns, and important lithological and structural associations. A prospectivity ranking can therefore be developed that can be supplemented with other existing information such as airborne magnetic or radiometric data. At that stage, the only missing element is geochemistry and sampling can now be carried out in a systematic manner over those targets most likely to produce success, thereby optimising time and effort in regional geochemical exploration and reducing costly land acquisition to those spectral targets with anomalous geochemistry. This approach has been shown to be cost-effective in Nevada (Coopersmith, 1991; De Largie et al., 1993; Lyon et al., 1993) and Western Australia (Biggs, 1989; Derriman & Agar, 1990), and led to the discovery of the Hadleigh Castle mine and other resources in Queensland (Gunn et aL, 1989; Ashton Mining Ltd, 1993).

6.0 ACKNOWLEDGEMENTS

The author is extremely grateful to Mark Derriman, who collected most of the Open File material from the Mines Department, Ros Perez of Ashton Mining Ltd and Adrienne Meakins of CRAE for helpful discussions and information on the Jillawarra mining centre, and Jo Cannon for patience and understanding over the production of the colour plates and figures.

7.0 REFERENCES

ABRAMS M.J., BROWN D., LEPLEY L. & SADOWSKI R., 1983. Remote sensing for porphyry copper deposits in southern Arizona. Economic Geology 78, 591-604.

AGAR R.A., 1992. Geological applications and limitations of airborne multi-spectral scanners; their availability, cost-effectiveness and future. In Remote Sensing and Spatial Information, the Functions, the Payback, the Future. 6th Australasian Remote Sensing Conference, Wellington, Volume 2, pp.259-273.6th Australasian Remote Sensing Conference Committee, Wellington.

ASHTON MINING LTD., 1993. Annual Report 1992. Ashton Mining Ltd, Melbourne, 48 pp.

AUSTRALlAN ORES AND MINERALS LTD, 1991. E37/197 - Blue Well, Final Report. Mines Department of Western Australia, Open File Report 6041, 59 pp.

BIGGS E.R., 1989. Costing of a theoretical mineral exploration programme. Mackay & Schnellman Pry Ltd, Report on behalf of Geoscan Pry Ltd, 12 pp. (unpublished).

BORDER GOLD, 1993. Border Gold Ltd. Australian Stock Exchange Prospectus. Border Gold Ltd, Perth, 100 pp.

CCOPERSMITH H.O., 1991. A Geoscan AMSS discovery at Sentinel, Humboldt County, Nevada. In Proceedings, &h Thematic Conference on Geologic Remote Sensing, unpaginated addendum. Environmental Research Institute of Michigan, Denver.

CRA Exploration Pty Ltd, 1988. Jillawarra-Chesterfield, Annual Report for 1988. Mines Department of Western Australla, Open File Report 13700, 7-8.

CROSTA A.P. & RABELO A., 1993. Assessing Landsat TM for hydrothermal alteration mapping in central-western Brazil. In Proceedings, 9th Thematic Conference on Geologic Remote Sensing, Pasadena, Volume 2, pp.1053-1061. Environmental Research Institute of Michigan, Pasadena.

DE LARGIE D.A., LOCKETT N.H., AGAR R.A., KROENKE M. & LYON R.J.P., 1993. An integrated Landsat-MSS, aeromagnetic, air-photographic and Geoscan AMSS approach to gold exploration in Nevada. In Proceedings, 9th Thematic Conference on Geologic Remote Sensing, Pasadena, Volume 2, pp.1097-1110. Environmental Research Institute of Michigan, Pasadena.

DERRIMAN M.D.J. & AGAR R.A., 1990. Gold and base metal exploration in the Pilbara Craton, Western Australia, using the Geoscan Airborne Multi-Spectral Scanner. In Proceedings, 5th Australasian Remote Sensing Conference, Perth, 2, pp.924-927.5th Australasian Remote Sensing Conference Committee, Perth.

DICKSON F.W., ROBERT O.R. & RADTKE A.S., 1979. The Carlin Gold Deposit as a product of rock-water interactions. In Ridge J.D. (ed.), The International Association on the Genesis of Ore Deposits Fifth Quadrennial Symposium, Proceedings Volume 2. Bureau of Mines and Geology, Nevada, Report 33, 101-108.

DRURY S.A. & HUNT G.A., 1988. Remote sensing of laterized Archaean greenstone terrain: Marshall Pool area, North Eastern Yilgarn Block, Western Australia. Photogrammetnc Engineenng and Remote Sensing 54, 1717-1725.

DRURY SA. & HUNT G.A., 1989. Geological uses of remotely sensed reflected and emitted data of laterized Archaean terrain in Western Australia. International Journal of Remote Sensing 10, 475-497.

FRASER S.J., 1991. Discrimination and identitication of ferric oxides using satellite thematic mapper data - A Newman case-study. International Journal of Remote Sensing 12, 635-641.

GUNN M.J., HONEY F.R., LYON R.J.P. & FURNELL R.G., 1989. Geoscan AMSS leads to gold mineralisation at Crossroads, Queensland. In Proceedings, NQ Gold '89 Conference, Townsville, p139. Australasian Institute of Mining and Metallurgy, Melbourne.

HATCH A., 1991. A remote sensing study of kimberlite and lamproite pipes in the Kimberley region, Western Australia. B.Sc. (Honours) thesis, The University of Western Australia, Nedlands, 76 pp. (unpublished).

HATCH A., TAYLOR W.R. & AGAR R.A., 1992. A remote sensing study of kimberlite and lamproite pipes using Geoscan airborne multi-spectral scanner imagery: examples from the Kimberley, Western Australia. In 11th Australian Geological Convention, Earth Sciences, Computers and the Environment. Geological Society of Australia, Absfracts 32, 309-310.

HONEY F.R. & DANIELS J.L., 1986. Rock discrimination and alteration mapping for mineral exploration using the Carr Boyd Geoscan airborne multi-spectral scanner. In Proceedings, 5th Thematic Conference on Geologic Remote Sensing, pp.267-278, ERIM, Reno.

HUNTINGTON J.F., CRAIG M.C., CHURCHILL J.N. & GREEN A.A., 1991. Airborne mineral mapping with the GERIS and Geoscan aircraft scanners for mineral exploration and mapping. In Abstracts, 8th ASEG Conference and Exhibition. Advances in Mineral Exploration, pp.18-21. Australian Society for Exploration Geophysicists, Sydney.

LOUGHLIN W.P., 1991. Principal component analysis for alteration mapping. In Proceedings, 8th Thematic Conference on Geologic Remote Sensing, pp.293-306. ERIM, Denver.

LYON R.J.P., 1990. Effects of weathering, desert varnish etc. on spectral signatures of mafic, ultramafic and felsic rocks, Leonora, West Australia. In Proceedings, 5th Australasian Remote Sensing Conference, Perth, Volume 2, pp.915-923. 5th Australasian Remote Sensing Conference Committee, Perth.

LYON R.J.P. & HONEY F.R., 1990. Direct mineral identification (DM1) with Geoscan MKII advanced multi-spectral scanner (AMSS). In Proceedings, Imaging Spectroscopy of the Terrestrial Environment. Society of Photo-Optical Instrumentation Engineers, Proceedings Series 1298, 50-61.

PEREZ R., 1990a. A32283. Jillawarra Project, Australian Ores & Minerals Ltd., relinquishment report for E511232. Mines Department of Western Australla, Open File Report 4763, 36 pp.

PEREZ R., 1990b. A353 Ashburton reconnaissance joint venture, final technical report, Ashton Gold (W.A.) Ltd. Mines Department of Western Australla, Open File Report 4763,14 pp.

PODWYSOCKI M.H., SEGAL D.B. & ABRAMS M.J., 1983. Use of multispectral scanner images for assessment of hydrothermal alteration in the Marysville, Utah, Mining Area. Economic Geology7s, 675-687.

TOWIE N.J., MARX M.R., BUSH M.D. & MANNING E.R., 1991. The Aries diamondiferous kimberlite pipe, central Kimberley block, Western Australia. In Extended Abstracts, 5th International Kimberlite Conference. CPRM, Special Pubhcation 2191, 435-436.

8.0 FIGURE CAPTIONS

Figure 1: Selected mineral and rock spectra relative to Geoscan AMSS MkI band positions in: (a) the visible near infra-red (VNIR), (b) the short wave infra-red (SWIR), (C) the thermal infra-red (TIR).

Figure 2: Selected mineral and rock spectra relative to Geoscan AMSS MkII band positions in: (a) the visible near infra-red (VNIR), (b) the short wave infra-red (SWIR), (C) the thermal infra-red (TIR).

Figure 3: Location map of Western Australia showing localities referred to in text.

Figure 4: Geoscan AMSS MkI imagery over the Aries kimberlite pipe showing treatments: (a) 3 2 1 RGB (simulated true colour) (b) 9/8 7/8 6/8 RGB (c) 6-7 6-8 6-9 RGB (d) 9/8 7/8 4 RGB (e) 3-2 3-4 3-5 RGB, relative to (f) geological map of the image area to the same scale.

Figure 5: Geoscan AMSS MkII imagery over the Blue Well prospect showing treatments: (a) 4 3 2 RGB (simulated true colour) with geological map (b)17 8 2 RGB (c) 11-16 11-17 11-18 RGB (d) 13-14 13-15 13-16 RGB (e) 20 21 22 RGB (f) 23-20 22-20 21-20 RGB.

Figure 6: The geology of the Mingah Range greenstone belt and Jillawarra mining centre showing the location of the Geoscan imagery displayed in Figure 7. After CRA Exploration Pty Ltd (1988).

Figure 7: Geoscan AMSS MkII imagery over the Jillawarra mining centre showing treatments: (a) 3 2 1 RGB (simulated true colour) (b) 17 8 2 RGB (c) 11-14 11-16 11-18 RGB (d) 13-14 13-15 13-16 RGB (e) 6-7 6-8 6-9 RGB (f) 20 21 22 RGB.

Figure 8: Geoscan AMSS MkII imagery over the KazputPool prospect showing treatments: (a) 23 17 3 RGB (b) 11-14 11-15 11-16 RGB (c) 13-15 14-15 16-15 RGB (d) 13-14 13-16 16-15 RGB (e) 19 20 23 RGB, relative to (f) geological map of the image area to the same scale.