TECHNICAL PAPERS

MULTI-SPECTRAL DATA AND THEIR APPLICATIONS FOR GOLD EXPLORATION IN LATIN AMERICA

Presented at the Second International Gold Symposium, Lima, Peru, May, 1996.

Multi-spectral data collected by either satellite mounted, airborne or hand-held instruments, use reflected and re-radiated solar energy to determine the nature of material occurring at the earth’s surface. This multi-spectral data is a measure of solar reflectance and thermal emittance in specific parts of the electro-magnetic spectrum. The data ranges from visible and near infra-red light waves (VNIR), through to the short wave and thermal infra red (SWIR and TIR respectively). Spectra collected across these wavelengths can provide a unique finger print for various surface materials and thus have applications in geology, mineral exploration and mapping.

SPOT, Landsat Multi-Spectral Scanner (MSS) and Thematic Mapper are examples of satellite mounted data acquisition systems whereas Geoscan, GERIS, DAIS and AVIRIS are airborne instruments that have been used in geological applications. Each system has a unique spatial (i.e. footprint) and spectral resolution that effectively limits the instrument’s performance and its ability to discriminate detail. Satellite instruments provide broad coverage at low spatial and low spectral resolution and are useful as sources of data for preliminary or first pass regional reconnaissance work. Airborne systems provide greater spectral and spatial resolution and are applied in both regional and detailed prospect exploration. The more sophisticated the instrument, the greater is the range of applications for its data. However, this is offset by an increase in cost and decrease in availability.

The multi-spectral data in its raw form is digital but can be processed to produce false colour images of terrain surveyed. The most commonly used application of such satellite images is in the photogeological interpretation of significant structures, geological contacts and lineaments. However, much more information can be gained from spectral analysis of the data. Landsat TM data is finding widespread use as an exploration targeting tool in Latin America where spectral analysis highlights zones with high iron and clay contents. Many Peruvian gold exploration "hot spots" were initially defined on the strength of their high iron and clay spectral signature. However, some caution must be exercised in interpreting some of these zones because high iron-clay signatures do not always relate to hydrothermal alteration.

More sophisticated instruments are able to discriminate key hydrothermal minerals such as chlorite, kaolinite, sericite, alunite, jarosite, and silica. This is achieved by use of calibration to detect very minor spectral differences between such minerals. Thus, not only can prospective exploration targets be identified but also their hydrothermal alteration mineralogy can be mapped in detail from the air. In remote or topographically difficult regions, this provides a major exploration advantage. Both large and complex alteration zones associated with porphyry and epithermal styles of gold mineralisation and narrower, more subtle shear zone-hosted gold occurrences can be detected.

High sulfidation epithermal ore bodies, such as Yanacocha and La Coipa, and low sulfidation deposits, such as San Cristobal and Kori Kollo, have characteristic hydrothermal alteration assemblages that can be readily identified and mapped using high resolution multi-spectral data. Although much narrower, the quartz veins and chloritic wall rock alteration in the deposits of the Pataz district of Peru are also detectable given the appropriate multi-spectral tool.

For each specific ore body type, multi-spectral data can discriminate its associated hydrothermal alteration mineralogy. Thus, the problem is not of spectral but of spatial resolution. In situations like the Pataz vein-hosted gold mineralisation, veins and alteration selvedges are only metres wide but may be more than 1km in length. In photogeological analyses of data, features must have a dimension more than three times the pixel size to be recognisable. Thus, satellite data with a resolution of 30m cannot be expected to reveal such veins. However, in practice, the longer and wider veins do show up due to the extensive shedding of quartz away from the vein. High spatial resolutions of 5m or less can be achieved with airborne multi-spectral instruments and readily pick out both veins and alteration halos. However, such detail is gained at the expense of total regional coverage and may be beyond the budget of many exploration companies. Consequently, where a multi-spectral survey is being considered, care must be taken to ensure that the tool has both the spectral and the spatial resolution needed.

An important drawback in the use of low resolution data has been the impurity of the individual pixel spectra. The larger the pixel, the greater the variety of materials that will be present within the ground that it covers. The 30m square Landsat TM pixel for example might contain more than one rock type or multiple alteration minerals plus a range of vegetation types. The resultant spectrum obtained will not approximate to any of its constituent materials. With higher spectral resolution than offered by Landsat, it is possible to recognise in the data, spectral features that relate to specific materials. Thus end member materials can be identified. Recent developments in the field of spectral unmixing enable these end members to be quantified and sub-pixel sized features to be mapped. The significance of successful spectral unmixing analysis is that high spatial resolution surveys of 5m or less can be avoided, allowing high spectral resolution surveys to be flown on regional scales.

Multi-spectral data is finding increasing acceptance worldwide as an exploration tool but it must be remembered that it is just one of many tools available. There is no substitute for fieldwork and the good explorationist uses all the various tools available in the most appropriate order so as to focus and derive optimum value for his time in the field. Multi-spectral data is not the best prime survey tool in heavily vegetated regions of the world or where geology is obscured by extensive superficial cover. In such areas, airborne geophysical techniques such as magnetics, radiometrics, EM or radar provide more benefit. However, in arid and semi-arid regions, multi-spectral data provides a cost-effective survey alternative as a framework for subsequent exploration geochemistry and geophysics. Indeed, with the ability to discriminate individual minerals at sub-pixel scales, airborne multi-spectral data is effectively an airborne high-density geochemical/mineralogical tool

In summary, multi-spectral data has already proven to be very valuable in the arid parts of Latin America where Landsat imagery has been used for regional geological appraisals and to target large scale alteration zones. Airborne data from high resolution instruments has been successfully applied in both Chile and Brazil but is generally more difficult and expensive to obtain. However, the long term benefit of high spectral resolution data lies in its value as both a regional mapping and a targeting tool plus its downstream application to detailed mineral mapping of mineralised prospects. In remote and difficult terrains, the systematic and planned use of multi-spectral data provides a very cost effective answer to logistically difficult exploration.

1.0 INTRODUCTION

Multi- and hyperspectral remote sensing data are finding increasing use in many applications around the world, including mineral exploration. Such data acquired from satellite platforms has already played a major role in the identification of significant hydrothermal alteration systems. However, it has been found to be lacking in the detailed spatial and spectral resolutions necessary for the advancement of exploration to the area of mineral mapping. This paper will investigate the nature of multi- and hyperspectral data in the light of the requirements for gold and mineral exploration in general and in South America in particular. The current status of operating systems will also be reviewed and the ways in which their data can be analysed and presented will be discussed. Evaluation of the applicability and cost effectiveness of multi- and hyperspectral remote sensing will be reviewed in comparison to other, more traditional methods of regional exploration.

2.0 MULTISPECTRAL DATA

Spectral data collected by either satellite mounted, airborne or hand-held instruments, is reflected and re-radiated solar energy and is used to determine the nature of material occurring at the earth’s surface. This multi-spectral data is a measure of solar reflectance and thermal emittance in the electro-magnetic spectrum ranging in wavelength from 0.4mm in the visible, through the near and short wave infra red to 2.5mm, and out into the thermal infrared between 8 and 12mm. Downwelling solar radiance interacts with materials and minerals on the earth’s surface where it is either reflected, transmitted, absorbed or re-emitted. Specific materials reflect or absorb in different degrees for different wavelengths and each has its own unique spectrum. Thus, reflection and emittance spectra collected across these wavelengths can provide a unique finger print for various surface materials (Fig. 1). Laboratory spectra gathered from vegetation, rocks and minerals provide a reference for use in the identification of these same materials from either satellite or airborne data. However, only rarely is the data cell (pixel) sampled by these remote instruments comprised of a single material. Thus airborne and satellite spectra are typically impure or mixed with elements of vegetation, soil, mineral and cultural materials. Consequently, for geological applications, such data is best applied to terrane with minimal vegetation such that the mineral component of the various pixel spectra is maximised. Nevertheless, even without vegetation, the spectra of naturally occurring rocks and soils will be comprised of mineral mixtures.

Figure 1. Remote sensing reference chart showing some spectral curves of common materials such as water, vegetation, soil and a clay mineral relative to some multi- and hyperspectral instruments and their spectral coverage, band positions and widths.

The term multi- or hyperspectral data refers to data collected by an instrument in multiple wavelengths so as to generate a spectral curve such as those shown in figure 1. Satellite data such as Landsat TM is considered multispectral by virtue of recording data in 7 distinct wavelength bands, other instruments such as AVIRIS with 224 channels are termed hyperspectral. Instruments with relatively few broad bands widely spread such as Landsat TM are low resolution and have limited discriminatory powers. Hyperspectral instruments on the other hand, have many contiguous narrow channels and a far greater potential for mineral recognition. In evaluating the applicability of such data to gold exploration in Latin America, it is important therefore to consider both the nature of the terrane, the type and style of mineralisation and associated macro mineral assemblages that may characterise it, the stage of exploration and features required to be discriminated, and the types of spectral data available. Large parts of Latin America are heavily vegetated and are really not suitable therefore for exploration using multi- or hyperspectral data as a geological or mineral mapping tool. However, much of the western part of South America from Northern Peru southward to Central Chile and then across into Patagonia is arid or semi arid country and ideally suited to the remote collection of rock, soil and mineral spectra. All that remains to be determined is the nature of the mineralisation being sought and the ability of the currently available instruments to recognise the salient features of these deposits.

3.0 SPECTRAL CHARACTERISTICS OF GOLD DEPOSITS

Gold deposits around the world are many and varied but can be grouped into two main types, placer and hydrothermal. Gold bearing placer deposits are indistinguishable spectrally from non mineralised alluvial gravels and will not be discussed here. However, hydrothermal gold mineralisation, whether shear-zone hosted, banded iron formation related, epithermal high or low sulphidation types, sediment hosted, skarn or detachment related, is characterised by an attendant suite of alteration minerals which lend themselves very well to identification and discrimination by remote sensing techniques. Spatz (1996a & b) discusses in detail the remote sensing characteristics of volcanic and sediment hosted gold deposits and those related to detachments. However, these and the shear zone hosted occurrences are characterised by their own specific alteration styles and mineral assemblages.

Figure 2 shows a schematic view of alteration zones at various levels in a mineralising porphyry hydrothermal system. The various hydrothermal alteration assemblages are developed according to ambient temperature and pressure regimes during mineralisation. Mapping these assemblages at the surface allows the explorationist to understand the mineralising system and to vector towards mineralisation. These key mineral assemblages and the minerals that comprise them are listed in figure 3. Many of these minerals, in particular the clay minerals, appear very similar in the field and are difficult to discriminate. Nevertheless, they all have specific spectral features, in particular in the short wave infrared (SWIR) part of the spectrum (figure 4). Thus, the role of remote sensing and multi- and hyperspectral data in gold exploration lies in the ability of its various instruments to discriminate these minerals spectrally and thus to recognise and map key hydrothermal mineral assemblages (figures 2 & 3).

	Figure 2. Schematic representation of alteration zones and mineral assemblages within a porphyry/epithermal style mineralising system.
	Figure 3. Matrix showing alteration mineral assemblages and families important in mineralising systems (after Lyon, pers. comm.).
	Figure 4. Short wavelength infrared spectra of some typical hydrothermal alteration minerals relative to the band positions of the Geoscan MKII AMSS.

4.0 MULTI- AND HYPERSPECTRAL INSTRUMENTS

A range of currently operational multi- and hyperspectral instruments are listed in figure 1. This list does not include ground or hand held spectroradiometers such as the PIMA or GER group of instruments which are not typically used in a regional sense but tend to be applied more in support of airborne and satellite data. Nevertheless, the ground instruments are finding increasing use and should be included here because their data is hyperspectral, with many more channels and much narrower band passes than any of the airborne and satellite instruments.

Of these latter, Landsat TM has been the most widely used in geological and mineral exploration circles, particularly in western South America where enhancements which highlight clay-iron enrichment have been used to identify porphyry style hydrothermal alteration systems. Figure 5 is one such enhancement and shows the El Halcon porphyry copper prospect located 55km north of Copiapo in Chile. This enhancement uses the ratios of bands 5/4 and 3/1 to highlight iron-rich zone and 5/7 to discriminate clay rich areas. Unfortunately, due to its low resolution with just one broad band in the 2.2-2.5mm and 8-12mm wavelength ranges compared to instruments such as Geoscan MKII with eight and six channels respectively across the same range (figure 1), Landsat TM is not able to discriminate important individual hydrothermal alteration minerals. Geoscan multi-spectral SWIR data over El Halcon can map the distribution of important minerals such as alunite (yellow in figure 6) and can generate thematic assemblage maps such as that of the El Abra porphyry copper deposit in which silica can also be identified using the 8-12mm wavelength range (figure 7).

	Figure 5. Landsat TM image showing typical clay-iron enhancement of the El Halcon prospect, Chile.
	Figure 6. Geoscan MKII AMSS data over part of the El Halcon prospect showing alunite distribution in yellow (from Agar et al. 1994).
	Figure 7. Thematic image map of the El Abra porphyry copper deposit, Chile derived from Geoscan MKII AMSS data, showing argillic alteration in yellow and quartz tourmaline breccias as turquoise in a ring around a barren core to the deposit in red.

A further limitation of satellite data such as Landsat TM is their spatial resolution. Landsat TM data has a pixel size of 30m as compared to airborne instruments such as AVIRIS (20m), GERIS and Geoscan (both variable between 2 and 20m). Such low spatial resolution imposes limitations upon the ability of satellite instruments to identify narrow vein features. In vein-hosted gold mineralisation, veins and alteration selvedges are often only metres wide but may be more than 1km in length. In photogeological analyses of data, features must have a dimension more than three times the pixel size to be recognisable. Thus, satellite data with a resolution of 30m cannot be expected to reveal such veins. However, in practice, the longer and wider veins do show up due to the extensive shedding of quartz away from the vein. High spatial resolutions of 5m or less can be achieved with airborne multi-spectral instruments and readily pick out both veins and alteration halos. However, such detail is gained at the expense of total regional coverage and may be beyond the budget of many exploration companies. Consequently, where a multi-spectral survey is being considered, care must be taken to ensure that the tool has both the spectral and the spatial resolution needed.

The limitations of satellite instruments are well demonstrated when Landsat TM data over the Charters Towers area of Queensland, Australia, is compared with Geoscan MKI multi-spectral data over the same area (Figure 8). In this example, the difference in spatial resolution is evident from the relative degree of detail visible in the two data sets. Importantly, however, the Geoscan data also discriminates and maps the trace of sericitic alteration along a shear zone in granite. This image focussed the attention of exploration in the area and led to the eventual discovery and successful mining of over 300,000 ounces of gold in Ashton Mining’s Rishton project. Given similar spectral and spatial resolution, the very similar granite hosted vein deposits of the Pataz district (Vidal et al., this volume) and of the Nazca - Ocona district (Valdivia, this volume, Martinez, this volume) could equally well be discriminated.

Figure 8. A Landsat TM image in the background of the Charters Towers gold district, Queensland, Australia, with three separate Geoscan images of selected parts of the area superimposed. Note the Rishton shear zone highlighted in white in the lower Geoscan image and the greatly enhance spatial resolution of the Geoscan data.

Thus, satellite multi-spectral data such as Landsat TM is of low spatial and spectral resolution but yet has a large footprint allowing for wide coverage at low cost. Such data is ideal for and is widely used in first pass, broad regional studies. Airborne multi- and hyperspectral instruments however have much higher spatial and spectral resolution and are much more powerful in terms of their geological and mineral mapping potential. However, coverage is limited and commissioning of surveys can be costly. Thus, such data tends to be applied to more localised studies and, as described below, for highly focussed detailed mineral mapping of mineralised systems.

Ground systems have been designed and developed fundamentally to provide good quality ground spectra to assist in the calibration and application of the airborne and satellite systems noted above. However, in recent years, these ground systems have found increasing use as a prime mapping tool for fundamental mineral recognition in complex alteration zones and are being widely applied to drill cuttings and core as a means of logging alteration mineral assemblages.

5.0 DATA ANALYSIS

Multi-spectral data in its raw form is digital but can be processed to produce false colour images of terrain surveyed. The most commonly used application of such satellite images is in the photogeological interpretation of significant structures, geological contacts and lineaments. However, much more information can be gained from spectral analysis of the data. Landsat TM data is finding widespread use as an exploration targeting tool in Latin America where spectral analysis highlights zones with high iron and clay contents (figure 5). Many Peruvian gold exploration "hot spots" were initially defined on the strength of their high iron and clay spectral signature. However, some caution must be exercised in interpreting some of these zones because high iron-clay signatures do not always relate to hydrothermal alteration.

More sophisticated instruments such as Geoscan, GERIS and AVIRIS are able to discriminate key hydrothermal minerals such as chlorite, kaolinite, sericite, alunite, jarosite, and silica. This is achieved by use of calibration to detect very minor spectral differences between such minerals. Initially, broad alteration types such as argillic and propylitic assemblages can be discriminated simply by use of the relative position of absorption bands in key minerals. For example, chlorite, epidote and carbonate, typical minerals in propylitic alteration, have absorption features at longer wavelengths in the SWIR than do those associated with argillic alteration such as alunite, kaolinite and sericite (figure 4). Similarly, alunite is discriminated in El Halcon by a band difference image (figure 7) and many more minerals and alteration assemblages have been mapped in the same way (Agar et al., 1994). Thus, simple band difference imaging can localise broad alteration zones as in figures 9 and 11 for the La Coipa and El Hueso deposits respectively. However, careful calibration of the same airborne multispectral data against known uniform materials on the ground, such as water bodies for example, allows the recognition of relatively pure mineral spectra within the airborne data such as those for alunite and kaolinite in figures 10 an 12. Thus, not only can prospective alteration zones be targeted for exploration but also their hydrothermal alteration mineralogy can be mapped from the air. High sulfidation epithermal ore bodies, such as Yanacocha and La Coipa, and low sulfidation deposits, such as San Cristobal and Kori Kollo, have characteristic hydrothermal alteration assemblages that can be readily identified and mapped using high resolution multi-spectral data. Although much narrower, the quartz veins and chloritic wall rock alteration in the deposits of the Pataz and Nazca - Ocona districts of Peru are also detectable given the appropriate multi-spectral tool at a suitable spatial resolution. In remote or topographically difficult regions, this provides a major exploration advantage.

Figure 9. Geoscan MKII AMSS image over the La Coipa gold deposit showing argillic alteration in pink and propylitic alteration in yellow.	Figure 10. Geoscan MKII AMSS image over the La Coipa gold deposit showing alunite distribution in yellow and kaolinite as pale blue with typical airborne mineral spectra.
Figure 11. Geoscan MKII AMSS image over the El Hueso gold deposit showing argillic alteration in pink and propylitic alteration in yellow.	Figure 12. Geoscan MKII AMSS image over the El Hueso gold deposit showing alunite distribution in yellow with typical airborne mineral spectrum.

An important drawback in the use of multi- and hyperspectral data has been the impurity of the individual pixel spectra. The larger the pixel, the greater the variety of materials that will be present within the ground that it covers. Both the 30m square Landsat TM pixel and a 10m square Geoscan or GERIS pixel for example might contain more than one rock type or multiple alteration minerals plus a range of vegetation types. The resultant spectrum obtained will not approximate to any of its constituent materials. However, with higher spectral resolution than is offered by Landsat, it is possible to recognise in the data, spectral features that relate to specific materials. Thus end member materials can be identified. Recent developments in the field of spectral unmixing enable these end members to be quantified and sub-pixel sized features to be mapped (Boardman, 1993, Boardman & Kruse 1994, Kruse, 1996). Furthermore, very careful spectral analysis has enabled not only individual minerals to be discriminated in airborne AVIRIS data at 20m resolution, but permits the recognition and mapping of high and low temperature varieties of the same mineral (figure 13). Although high and low temperature alunites may allow the explorationist to localise the high heat flow and potentially mineralised parts of an ancient hydrothermal system, the image map produced in figure 13 is probably much further than one would ordinarily process spectral data within an ordinary exploration programme. Nevertheless, it serves to demonstrate the potential of airborne spectrometry as a mineral mapping tool. The significance of this, coupled with spectral unmixing analysis is that high spatial resolution surveys of 5m or less can be avoided, thereby allowing high spectral resolution surveys to be flown on regional scales.

Figure 13. Mineral map over the Cuprite alteration system, Nevada, USA derived from AVIRIS airborne hyperspectral data (Clark & Swayze 1995).

6.0 COST EFFECTIVENESS

Multi-spectral data is finding increasing acceptance worldwide as an exploration tool but it must be remembered that it is just one of many tools available. There is no substitute for fieldwork and the good explorationist uses all the various tools available in the most appropriate order so as to focus and derive optimum value for his time in the field. If multi- or hyperspectral remote sensing is to play a major role in mineral exploration, it must not only be seen to be technically sound but it must also be cost effective in comparison to alternative methods of exploration. Multi-spectral data is not the best prime survey tool in heavily vegetated regions of the world or where geology is obscured by extensive superficial cover. In such areas, airborne geophysical techniques such as magnetics, radiometrics, EM or radar provide more benefit and regional geochemistry remains the most frequently used first pass exploration technique in all terranes. Interestingly however, when multispectral remote sensing is compared to regional geochemistry and airborne geophysics, it can be seen to be provide a much higher sample density than any other method, 30m for satellite and probably 10m for airborne (figure 14a). Furthermore, hyperspectral data such as is acquired by the GER 63-channel Digital Airborne Imaging Spectrometer has far greater dimensionality than any of the other methods (figure 14b). Regional geochemistry may analyse for up to 30 elements or more but this is very rare.

Figure 14. Comparison of the relative resolution (a), dimensionality (b), and cost (c) of typical first pass exploration tools.

Finally, in terms of cost per unit area, taking into account simply the cost of acquisition of data and not the processing and interpretation, hyperspectral data is more than competitive with geophysical and geochemical surveys (figure 14c). Indeed, the figures for the geochemical data, which include not just the assay cost but also the cost of accessing and collecting the samples, are typical of a survey in the relatively accessible, flat lying Archaean gold areas of Western Australia (Mackay & Schnellman, 1989). In topographically difficult and remote parts of Andean South America, this cost is likely to be much higher. Similarly for airborne geophysical surveys, low level fixed wing surveys are impossible over much of the Andean belt and helicopter borne surveys are both more costly and higher risk. Thus, the cost effectiveness of airborne hyperspectral surveys for topographically difficult arid terranes such as the Central and southern Andes is extremely well demonstrated in comparison to alternative methods.

7.0 DISCUSSION & CONCLUSIONS

Multi-spectral data use reflected and re-radiated solar energy to determine the nature of material occurring at the earth’s surface. Spectra collected across visible and near infra-red, short wave and thermal infra red wavelengths provide a unique finger print for various surface materials such as rocks, minerals, soils and vegetation. The data is recorded by both airborne and satellite mounted instruments and is supported by high resolution ground instruments. Each system has a unique spatial and spectral resolution that effectively limits the instrument’s performance and its ability to discriminate detail. Satellite instruments provide broad coverage at low spatial and low spectral resolution and are useful as sources of data for preliminary or first pass regional reconnaissance work. Airborne systems provide greater spectral and spatial resolution and are applied in both regional and detailed prospect exploration.

The multi-spectral data in its raw form is digital but can be processed to produce false colour images of terrain surveyed. Landsat TM data is finding widespread use as an exploration targeting tool in Latin America where spectral analysis highlights zones with high iron and clay contents. Many Peruvian gold exploration "hot spots" were initially defined on the strength of their high iron and clay spectral signature. However, some caution must be exercised in interpreting some of these zones because high iron-clay signatures do not always relate to hydrothermal alteration. More sophisticated instruments are able to discriminate key hydrothermal minerals such as chlorite, kaolinite, sericite, alunite, jarosite, and silica. Thus, not only can prospective exploration targets be identified but also their hydrothermal alteration mineralogy can be mapped in detail from the air. In remote or topographically difficult regions, this provides a major exploration advantage. Both large and complex alteration zones associated with porphyry and epithermal styles of gold mineralisation and narrower, more subtle shear zone-hosted gold occurrences can be detected. High sulfidation epithermal ore bodies, such as Yanacocha and La Coipa, and low sulfidation deposits, such as San Cristobal and Kori Kollo, have characteristic hydrothermal alteration assemblages that can be readily identified and mapped using high resolution multi-spectral data. Although much narrower, the quartz veins and chloritic wall rock alteration in the deposits of the Pataz and Nazca - Ocona districts of Peru are also detectable given the appropriate multi-spectral tool.

An important drawback in the use of hyperspectral data has been the impurity of the individual pixel spectra. However, recent developments in the field of spectral unmixing enable end member minerals to be identified and sub-pixel sized features to be mapped. The significance of successful spectral unmixing analysis is that high spatial resolution surveys of 5m or less can be avoided, allowing high spectral resolution surveys to be flown on regional scales.

Multi-spectral data is finding increasing acceptance worldwide as an exploration tool but there is no substitute for fieldwork and the good explorationist uses all the various tools available in the most appropriate order. Multi-spectral data is not the best prime survey tool in heavily vegetated regions of the world or where geology is obscured by extensive superficial cover. However, in arid and semi-arid regions, multi-spectral data provides a cost-effective survey alternative as a framework for subsequent exploration geochemistry and geophysics. Indeed, with the ability to discriminate individual minerals at sub-pixel scales, airborne multi-spectral data is effectively an airborne high-density geochemical/mineralogical tool

In summary, multi-spectral data has already proven to be very valuable in the arid parts of Latin America where Landsat imagery has been used for regional geological appraisals and to target large scale alteration zones. Airborne data from high resolution instruments has been successfully applied in both Chile and Brazil but is generally more difficult and expensive to obtain. However, the long term benefit of high spectral resolution data lies in its value as both a regional mapping and a targeting tool plus its downstream application to detailed mineral mapping of mineralised prospects. In remote and difficult terrains, the systematic and planned use of multi-spectral data provides a very cost effective answer to logistically difficult exploration.

8.0 REFERENCES

AGAR, R.A., FRASER, N.R. & LOCKETT, N.H. 1994, "Geoscan Airborne Multispectral Scanner as an exploration tool applied to El Halcon Prospect, Chile": In Mining Latin America; Challenges to the mining industry; Inst. I.M.M., Chapman & Hall, London, pp. 151-164

BOARDMAN, J.W. 1993, "Automating Spectral Unmixing of AVIRIS Data Using Convex Geometry Concepts": Summaries of the Fourth Annual JPL Airborne Geoscience Workshop, JPL Pub. 93-26, Vol 1, AVIRIS Workshop, Jet Propulsion Laboratory, Pasadena, CA, pp.11-14

BOARDMAN, J.W. & KRUSE, F.A. 1994, "Automated spectral analysis; a geological example using AVIRIS data, north Grapevine Mountains, Nevada: In Proceedings, ERIM Tenth Thematic Conference on Geologic Remote Sensing, Environmental Research Institute of Michigan, Ann Arbor, pp. I-407 - I-418.

BUCHANAN, L.J., 1981. "Precious metal deposits associated with volcanic environments in the Southwest U.S." In:- Relations of tectonics to ore deposits in the southern Cordillera; Arizona Geol. Soc. Dig., v.XIV, pp. 237-262

CLARK, R.N., & SWAYZE, G.A., 1996. "Cuprite, Nevada AVIRIS 1995 Data: Tricorder 3.3 product": In:- Hyperspectral Remote Sensing Analysis Workshop, ERIM Eleventh Thematic Conference on Geologic Remote Sensing, Environmental Research Institute of Michigan, Las Vegas, Nevada.

EINAUDI, M.T., 1982. "General features and origins of skarns associated with porphyry copper plutons.": In:- Advances in the Geology of the Porphyry Copper Deposits: Southwestern North America, S.R.Titley ed., Univ. Arizona Press, Tucson, pp. 185-210.

KRUSE, F.A., 1996. "Mineral Mapping for Environmental Hazards Assessment Using AVIRIS Data, Leadville, Colorado, USA" In Proceedings, ERIM Eleventh Thematic Conference on Applied Geological Remote Sensing, Environmental Research Institute of Michigan, Ann Arbor, pp. II-142-150.

MACKAY & SCHNELLMAN PTY. LTD., 1989. "Costing Mineral Exploration" Unpublished report to Geoscan Pty. Ltd., Perth, Australia, 15p.

MARTINEZ, P., "Yacimientos auriferos relacionadas al batolito de la costa en la franja Nazca-Ocona, Ica y Arequipa." Proceedings, Peru: Second International Gold Symposium, Lima.

SPATZ, D.M., 1996a, "Remote sensing Characteristics of Precious Metal Systems: The Volcanic-Hosted Deposits." In Proceedings, ERIM Eleventh Thematic Conference on Geologic Remote Sensing, Environmental Research Institute of Michigan, Ann Arbor, pp. I-1 - I-12.

SPATZ, D.M., 1996b, "Remote sensing Characteristics of Precious Metal Systems: The Sediment-Hosted and Detachment Related Deposits." In Proceedings, ERIM Eleventh Thematic Conference on Geologic Remote Sensing, Environmental Research Institute of Michigan, Ann Arbor, pp. I-13 - I-22.

VALDIVIA, J., 1996. "Geologia estructural de las vetas auriferas en la mina Ishihuinca, Caraveli, Arequipa." Proceedings, Peru: Second International Gold Symposium, Lima.

VIDAL, C.E., PAREDES, J., MACFARLANE, A.W. & TOSDAL, R.M., 1996. "Geologia y metalogenia del distrito Minero Parcoy, provincia aurfiera de Pataz, La Libertad." Proceedings, Peru: Second International Gold Symposium, Lima.

9.0 FIGURE CAPTIONS

Figure 1. Remote sensing reference chart showing some spectral curves of common materials such as water, vegetation, soil and a clay mineral relative to some multi- and hyperspectral instruments and their spectral coverage, band positions and widths.

Figure 2. Schematic representation of alteration zones and mineral assemblages within a porphyry/epithermal style mineralising system.

Figure 3. Matrix showing alteration mineral assemblages and families important in mineralising systems (after Lyon, pers. comm.).

Figure 4. Short wavelength infrared spectra of some typical hydrothermal alteration minerals relative to the band positions of the Geoscan MKII AMSS.

Figure 5. Landsat TM image showing typical clay-iron enhancement of the El Halcon prospect, Chile.

Figure 6. Geoscan MKII AMSS data over part of the El Halcon prospect showing alunite distribution in yellow (from Agar et al. 1994).

Figure 7. Thematic image map of the El Abra porphyry copper deposit, Chile derived from Geoscan MKII AMSS data, showing argillic alteration in yellow and quartz tourmaline breccias as turquoise in a ring around a barren core to the deposit in red.

Figure 8. A Landsat TM image in the background of the Charters Towers gold district, Queensland, Australia, with three separate Geoscan images of selected parts of the area superimposed. Note the Rishton shear zone highlighted in white in the lower Geoscan image and the greatly enhance spatial resolution of the Geoscan data.

Figure 9. Geoscan MKII AMSS image over the La Coipa gold deposit showing argillic alteration in pink and propylitic alteration in yellow.

Figure 10. Geoscan MKII AMSS image over the La Coipa gold deposit showing alunite distribution in yellow and kaolinite as pale blue with typical airborne mineral spectra.

Figure 11. Geoscan MKII AMSS image over the El Hueso gold deposit showing argillic alteration in pink and propylitic alteration in yellow.

Figure 12. Geoscan MKII AMSS image over the El Hueso gold deposit showing alunite distribution in yellow with typical airborne mineral spectrum.

Figure 13. Mineral map over the Cuprite alteration system, Nevada, USA derived from AVIRIS airborne hyperspectral data (Clark & Swayze 1995).

Figure 14. Comparison of the relative resolution (a), dimensionality (b), and cost (c) of typical first pass exploration tools.