TECHNICAL PAPERS

GIS INTEGRATION OF AIRBORNE MULTISPECTRAL, GEOPHYSICAL AND OTHER MINERAL EXPLORATION DATA AT THE EL HALCON PORPHYRY COPPER PROSPECT, COPIAPO, CHILE

Presented at the Twelfth International Conference and Workshops on Applied Geologic Remote Sensing, Denver, Colorado, 17-19 November 1997.

ABSTRACT

A major difficulty in advancing mature exploration projects, explored over several years by many different companies, is the integration of disparate datasets. These must be presented at the same scale and in the same format such that they can be readily modelled and interpreted.

A methodical data compilation and analysis technique which allows the extraction of the maximum amount of geoscientific data from each dataset is required. An exploration model and strategy can then be applied to produce an output dataset which will direct the exploration effort. This can be achieved by entering the data into a Geographical Information System (GIS). Such an approach was used at the El Halcon porphyry copper prospect near Copiapó, Chile.

All of the various datasets were digitised where necessary and were converted to a common format. Geoscan Airborne Multispectral Scanner data were processed using both algorithms and advanced spectral unmixing routines. Interpretation of geophysical data in conjunction with geological and remote sensing data revealed previously unclear relationships and when combined pointed to deficiencies in the data coverage. Once these deficiencies are rectified and the newly acquired data integrated into the GIS and analysed within the context of an exploration model, drill targets will then be generated with a much higher degree of confidence than was previously possible.

Integration of disparate datasets into a common format within a GIS allowed these datasets to be interpreted both in isolation and interactively with others. This technique extracted the maximum amount of geoscientific information from all of the available data and allowed many more parameters to be tested. It can therefore be seen that a GIS can be a powerful decision support tool for project management.

1.0 INTRODUCTION

A major difficulty in advancing mature exploration projects explored over several years by many different companies, is the integration of disparate datasets. These must be presented at the same scale and in the same format such that they can be readily modelled and interpreted.

The El Halcon porphyry copper prospect, which is located near Copiapó, Chile, has been explored over many years by several different organisations. The data produced from this work is mostly in the form of paper maps which are at various scales and which cover different parts of the prospect. These datasets include geology, ground magnetics, resistivity and induced polarisation. The only available digital data at the commencement of this study was Geoscan Mk2 Airborne Multispectral Scanner (AMSS) data which was acquired in May 1992.

This scanner was a narrow band remote sensing device which could record up to 24 channels from 46 available spectral bands within the range of 0.49 mm to 12.0 mm. These spectral bands comprised 10 in the visible/near infrared (VNIR), 8 in the shortwave infrared (SWIR) and 6 in the thermal infrared (TIR) portions of the electromagnetic spectrum. The El Halcon data were collected from an altitude of 15 670 feet above ground level which resulted in a ground resolution of 10 m and a swath width of 9.9 km. The band specifications were as presented in Table 1.

Table 1. Geoscan MK2 band positions and band widths.

2.0 GEOSCAN AMSS DATA

2.1 PREPROCESSING AND GEOREFERENCING

The immediate environs of the El Halcon prospect were covered by two partial scanner flight lines oriented east west, with an overlap of approximately ten percent. Initial image processing involved replacement of line dropouts by averaging two lines on either side of the faulty line to calculate the replacement line values. The angular pixel distortion at the edge of each scan line, known as the tan q effect, caused by the incidence angle of the scan, was also corrected. Further processing involved running noise reduction procedures in order to minimise instrument noise, aircraft component noise and atmospheric backscatter effects. The specific routines used were dark subtraction using each band minimum, a minimum noise fraction (MNF) transformation to segregate noise, followed by an inverse MNF transform to reduce noise and the Internal Average Relative Reflectance (IARR) Calibration routine.

Once the data had been calibrated, ground control points were chosen from 1:50 000 scale topographic maps to provide a ground control framework from which to warp and geocode the data using a triangulation warping routine. Triangulation warping routines are ideally suited to airborne data due to its variable and localised distortion as a consequence of undesirable aircraft motion such as yaw, pitch and roll. Nearest neighbour resampling was used as it does not alter the digital number of the pixel being resampled (Jensen, 1986). The overlapping portions of each run were carefully stitched together to provide a seamless mosaic image of the prospect.

2.2 ADVANCED PROCESSING

The data were then processed in order to provide mineralogical information as to the style of alteration present. Previous work had identified argillic and quartz sericite style alteration during geological mapping and processing of the Geoscan data using band differencing techniques had outlined areas consistent with the presence of argillic, sericitic and propylitic style alteration (Agar et. al., 1994). The technique employed here was to choose alteration minerals identified on the ground and also those which are associated with porphyry copper deposits and compare their laboratory spectra with each pixel spectrum within the SWIR bands. The first stage of this technique is to resample or convolve the laboratory spectra to the Geoscan Mk2 band positions. Examination of Figure 1 reveals that these minerals fall into four separate groupings given these band positions. One mineral from each group was therefore selected to be used in a spectral feature fit routine where a scale is calculated for the absorption feature depth, which acts as a measure of abundance for each material (Clarke et. al., 1987, 1991). The image and reference spectra are compared for each material using a least squares routine and both a scale and a root mean square error value are calculated. The spectral fit is then the scale divided by the error. Comparison of reference spectra with spectra from areas with the best fit indicated a good match for each of the four groups.

Figure 1. Selected Laboratory Spectra Convolved To Geoscan Mk2 Band Positions.

These spectra were then used to carry out a linear spectral unmixing routine over the area shown in Figure 2a. This process provides a measure of the relative abundances of each identified end member material. The reflectance at each pixel for each wavelength is assumed to be a linear combination of reflectance of each end member material present within the pixel and in direct proportion to its abundance within that pixel (Boardman, 1989). In a similar way to the fit routine, the unmix produces a scale and error value. The error was divided into the scale to provide an unmix fit. The unmix fit and the spectral feature fit were then multiplied together to produce the final unmix images. This revealed a very good correspondence between the top 35% of the final unmix image (Figure 2b) with the reference spectra for the sericite, opaline silica, illite and montmorillonite grouping. Less confident matches were produced for the other mineral groupings (Figures 3 and 4).

Figure 2. Area Subject To Unmixing Routines a) Geoscan Band 3 Greyscale Mosaic Image b) Unmix Image For Sericite, Illite, Montmorillonite and Opaline Silica Group.

Figure 3. Expanded View Of Unmix Images a) Alunite, Kaolinite and Pyrophyllite Group b) Sericite, Illite, Montmorillonite and Opaline Silica Group.

Figure 4. Expanded View Of Unmix Images a) Calcite, Chlorite and Epidote Group b) Jarosite.

2.3 GEOLOGICAL INTERPRETATION

Geological maps from previous work, which had identified a body of Tertiary granite (Tg) of essentially granodioritic composition intruding into a sequence of andesitic volcanics, were scanned and imported into a GIS. The geological boundaries were digitised on screen using the scanned raster images as a background. These vectors were then overlayed on Geoscan data algorithms (Agar et. al., 1994), imported into the GIS and designed to highlight lithological contacts and structure. The geological contacts were then modified using the imagery as a background. In this way, different algorithms could be used as a backdrop to the vectors being edited allowing different geological features to be highlighted. Interpretation of detailed or complex areas was readily facilitated by zooming in and out as required. A dynamic link between the GIS and the image processing software allowed standard image processing techniques such as contrast stretching to be carried out interactively with the interpretation and database building procedure. During this process, attributes were added to the spatial data, thus building the textual database at the same time as the spatial data was being produced and edited. Zones of silicification, including quartz tourmaline breccias, identified during ground mapping, were also interpreted and modified from Geoscan TIR band images (Figure 5). Cultural features identified in the imagery were also added to the database during the interpretation process, as were drillhole locations and associated logging and assay information.

Figure 5. Geoscan TIR Image With Geology in White, Drillhole Locations, Alteration Boundaries and Veining/Breccias in Black.

3.0 GROUND GEOPHYSICAL DATA

Geophysics, comprising ground magnetics, resistivity and induced polarisation was carried out by a previous explorer over part of the prospect. Unfortunately, the digital data from this work was not available for this study. It was deemed extremely important to relate the geophysics to the geology and remote sensing interpretation in order to come to a better understanding of the prospect geology and mineralisation potential. The available geophysical data were presented as contoured intensity, colour coded paper maps related to a local grid. In this format it is very difficult to relate the geophysical data to any other data as a consequence of the local grid and fixed scale. To overcome this problem, the maps were scanned and imported into an image processor. The relationship between the local grid and the standard cartographic datum and projection for this area was determined. A network of local grid points on the scanned maps were chosen and their coordinates in the Provisional South American Datum were calculated. These coordinate pairs were then used to warp the scanned images to the reference projection and hence the correct orientation, while at the same time reducing the distortion inherent in paper maps due to initial variations in the printing procedure, stretching and scale changes due to age and storage conditions and any rescaling during scanning. The resultant images were then edited to remove unwanted map presentation detail, leaving only the colour coded, contoured geophysical data. These images were then imported into the GIS for interactive analysis with other datasets.

Although in this format the geophysical data cannot be manipulated to enhance interpretation as it could be in its raw digital format, it could now be viewed and overlayed at the same scale and orientation as the other datasets and be readily rescaled as required. In this way, previously unused or difficult to use analogue maps can become valuable digital datasets, albeit less useful than original digital data.

4.0 DISCUSSION AND CONCLUSIONS

As with many GIS projects, the compilation of the database is by far the most time consuming part of the project. Once this is completed, the analysis and interpretation of relationships between the various datasets can be rapidly carried out.

Database compilation at the El Halcon prospect has resulted in a clarification of the relationship between the geophysics and geology. Higher induced polarisation values for the most part correspond to the alteration strength in a similar fashion to the magnetics, where a ring of low values coincide with the limit of the alteration. These lows may also have a relationship with the identified silicified zones. An extension of the geophysical coverage would be required to better evaluate this relationship. The resistivity data is somewhat ambiguous in that at one location a resistivity low corresponds to a zone of strong alteration, whereas another resistivity low lies outside the alteration envelope. However, this location is covered in colluvial material which may be masking alteration underneath this cover, thus providing a possible explanation for this lack of correlation. It is also apparent that the drillhole with the highest copper values (H3, Figures 5 & 6) is situated where there is a high magnetic response and, if the resistivity is extrapolated, may be in a resistivity low. It also corresponds to a zone of strong alteration, whereas the other drillholes are outside the strongly altered zone.

Figure 6. Ground Magnetics Coverage Outlined in Grey With Geology Over Sericite, Illite, Montmorillonite & Opaline Silica Unmix Image.

None of these relationships were readily apparent prior to combining multiple datasets within the GIS. Furthermore, examination of the combined data has very clearly outlined a deficiency in the geophysical coverage of the prospect area (Figure 6). More geophysics would enhance the understanding of these relationships and would greatly improve drill targeting.

Combining all of the available data within the GIS has defined relationships between the various datasets which were not clearly apparent previously. Interpreting the data within the framework of a geological model has revealed where important data is lacking and hence where to next direct the exploration effort. In this sense it can be seen that the methodical compilation of all available geoscientific data within a GIS can maximise the information extraction from each data component, and when combined can act as a powerful decision support tool.

5.0 REFERENCES

R.A. AGAR, N.R. FRASER, & N.R. LOCKETT, "The Geoscan Airborne Multispectral Scanner as an Exploration Tool Applied to the El Halcon Prospect, Chile." In Proceedings Mining Latin America, Santiago, May 1994.

J.W. BOARDMAN, "Inversion of imaging spectrometry data using singular value decomposition". In Proceedings, IGARSS’89, Twelfth Thematic Canadian Symposium on Remote Sensing, v4, p. 2069-2062, 1989.

R.N. CLARK, T.V. KING, & N. GORELICK, "Automatic continuum analysis of reflectance spectra". In Proceedings, 3rd Airborne Imaging Spectrometer (AIS) Workshop, JPL Publication 87-30, p. 138-142, 1987.

R.N. CLARK, G.E. SWAYZE, A. GALLAGHER, N. GORELICK, & F.A. KRUSE, "Mapping with imaging spectrometer data using the complete band shape least squares algorithm simultaneously fit to multiple spectral features from multiple materials". In Proceedings, 3rd Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop, JPL Publication 91-28, p. 2-3, 1991.

J.R. JENSEN, Introductory Digital Image Processing, Prentice-Hall, Englewood Cliffs, NJ, p. 379, 1986.

6.0 FIGURE CAPTIONS

Figure 1. Selected Laboratory Spectra Convolved To Geoscan Mk2 Band Positions.

Figure 4. Expanded View Of Unmix Images a) Calcite, Chlorite and Epidote Group b) Jarosite.

Figure 5. Geoscan TIR Image With Geology in White, Drillhole Locations, Alteration Boundaries and Veining/Breccias in Black.

Figure 6. Ground Magnetics Coverage Outlined in Grey With Geology Over Sericite, Illite, Montmorillonite & Opaline Silica Unmix Image.