Rocklea Dome

Last modified by Laukamp, Carsten (Mineral Resources, Kensington WA) on Jul 13, 2017

This page aims to provide an overview of avialble publications, reports and data of the Rocklea Dome case study that was undertaken by Maarten Haest during his stay with the Western Australian Centre of Excellence for 3D Mineral Mapping, led by Tom Cudahy. The publically available data set of the Rocklea Dome area, located in the Hamersley Basin of Western Australia, comprises remote and proximal hyperspectral, geochemical and XRD data that were used to characterise the resource potential of the Rocklea Dome Channel Iron Deposit. For sumary slides see C3DMM_RockleaDomeCID_keyslides_short.pptx

If you're just after the data, please go straight to Chapter 4 "Currently publically available data".


The Rocklea Dome 3D Mineral Mapping project was funded by the Western Australian Government through their support to the Western Australian Centre of Excellence for three-dimensional mineral mapping (C3DMM) in Kensington and by Murchison Metals Ltd. M. Verrall helped with acquisition of the XRD spectra and his introduction to QXRD is greatly appreciated. M. Cardy, A. Hackett, and S. Travaglione are acknowledged for the acquisition of the infrared spectroscopic data and the preparation of samples for XRD analysis. This work profited from fruitful discussions with CSIRO colleagues C. Ong, A. Rodger, E. Ramanaidou, and M. Wells and with Murchison Metals Ltd. geologists J. Johnson and S. Peterson. The Geological Survey of Western Australia covered part of the costs for the diamond drilling through their Exploration Incentive Scheme co-funded Exploration Drilling program. Ben Caradoc-Davies is thanked for helping C3DMM to make the Rocklea Dome hyperspectral dataset available via the SISS and Rob Woodcock is thanked for providing funding to assess and prepare the dataset for publication.

Executive Summary:

The Rocklea Dome 3D Mineral Mapping project was conducted by the Western Australian Centre of Excellence for 3D Mineral Mapping from 2009 to 2012 to showcase the opportunities offered by existing hyperspectral remote and proximal sensing technologies for comprehensive minerals systems analyses.

Key achievements of the Rocklea Dome 3D Mineral Mapping project are (Haest et al., 2012 a,b; 2013):

  • Quantification of iron oxide phases  and associated mineralogy derived from hyperspectral data and validated using X-ray diffractometry and geochemistry:
    • iron (oxyhydr-)oxide content: RMSE of 9.1 wt % Fe
    • Al clay content: RMSE 3.9 wt % Al2O3
    • hematite/goethite ratio: RMSE 9.0 wt % goethite
    • Spatial characterisation of vitreous vs. ochreus goethite
    • Defining the Tertiary channel boundary using key mineralogical parameters, such as the kaolin crystallinity
    • Modelling the iron ore resource of the Rocklea Dome CID
    • Identification of drill holes that were sunk into barren (i.e. bedrock) lithologies, suggesting that about a third of drill holes could have been saved
    • Detailed characterisation of clay mineralogy that is associated with distinct domains of the CID and its cover (i.e. kaolin group vs. Al-smectites vs. Fe-smectites)
    • Characterisation of mineral assemblages in the Quaternary cover of the Tertiary channel (e.g. calcrete)
    • Improvement of quality of mineral maps by application of vegetation unmixing methods

All of the above points showcase how hyperspectral data can be used for the whole of mine life cycle, from exploration to resource characterisation.

1       Geological Setting:

Haest et al. (2012b) provide a detailed overview of the geology of the Rocklea Dome and the formation of CID, which are briefly summarised here. The bedrock geology of the Rocklea Dome area comprises of an Archaean age monzogranite pluton and cross-cutting mafic and ultramafic intrusives that form part of the Pilbara Craton. The pluton is overlain by Archaean to Proterozoic metasedimentary and volcanic rocks of the Hamersley Province, which now envelope the central dome of monzogranite (Fig. 1; Thorne & Tyler, 1996). The intrusion of the monzogranite formed an exposed double, shallow plunging anticline with a hinge zone trending approximately WNW-ESE.

A meandering tertiary palaeochannel passes over the Archean and Proterozoic rocks, which locally contains Channel Iron ore Deposits (CID). The southern portion of the Beasley River CID crosscuts the north-western part of the Rocklea Dome. Channel iron ore was also drilled along 8 km strike length of a palaeochannel on the eastern side of the Rocklea Dome, which was referred to by Haest et al. (2012a,b; 2013) as the Rocklea Dome CID (Fig. 1). The bedrocks and Tertiary channel are covered partly by a wide range of regolith units (e.g Quaternary alluvium), variably covered by green vegetation and dry vegetation (mostly Spinifex grass and bushes).

Channel iron deposits consist typically of a mixture of Fe-Ox pelletoids and ferruginised wood fragments below 10 mm in size (Morris & Ramanaidou, 2007). In CID systems, the base of the channel sections consists of a clay horizon of variable composition. The CID is capped in places by calcrete and silcrete.

2       Analytical Methods:

2.1      VNIR-SWIR drill core spectroscopy

Reflectance spectra of the rock chips (RC) and diamond drill cores (RKD) were measured using CSIRO’s HyChips™ system (cf. Huntington et al., 2004),which comprises an Terraspec™-based spectrometer (ASD) system. An automated X-Y table moves the drill core tray in a snake-like pattern below the ASD optical fibre at a distance of ~6 to 13 cm (depending on sample type, i.e., diamond core or drill chips) while the spectral data are collected. Each sample spectrum is collected from a 1 × 1 cm area. Four light globes are positioned 40 cm above and at a small angle (off the backscatter/specular angle) to the measurement/sample point. In addition to hyperspectral data, high spatial resolution (0.1 mm pixel) images are collected from the core or chip tray, as well as the sample height in the tray measured using a laser profilometer. Reflectance spectra were calibrated using a Teflon/Spectralon ™ panel (see Haest et al., 2011a for more details).  ASD™ spectra were collected in the visible to near-infrared (VNIR: 380−1,000 nm) and shortwave-infrared (SWIR: 1,000−2,500 nm), with sampling intervals of 1.4 nm in the VNIR and 2.0 nm in the SWIR and a wavelength accuracy of ±1 nm. The spectral resolution is 5 nm in the VNIR and between 11 and 12 nm in the SWIR. The ASD radiance spectra of each sample are first converted to apparent bidirectional reflectance using the Teflon signal, which is collected at the beginning/ end of each drill core/drill chip tray measurement cycle.

2.2      Remote sensing

Airborne VNIR–SWIR data were collected using the Airborne Multispectral Scanner (AMS), which is an earlier version of the HyMap™ system (Cocks et al., 1998). The AMS system collects 96 bands over the VNIR-SWIR, excluding the atmospheric bands between ~1000 nm and ~1400 nm and between ~1800 and ~1950 nm, respectively. For each spectral band, the average band spacing is 15 nm and the average full width at half maximum is 17 nm. The AMS data over Rocklea Dome were collected from the 31st of July to the 2nd of August 2000, producing a set of in total 14 flight lines, with a combined length of ~280 km at a pixel size of approximately 7 m. All flight lines were flown in a north–south direction and have a width of ~3.5 km. Atmospheric correction was done using MODTRAN5 (Berk et al., 2004, 2006) and the SODA technique (Rodger, 2011), based on a combination of the AMS at-sensor radiance with in-scene flight parameters (e.g. latitude, longitude, sensor height, etc.). For more details about georeferencing and mosaiking the single flight lines see Haest et al. (2013).

2.3      X-ray fluorescence analysis (XRF)

(modified from Haest et al., 2012a)

XRF analysis of RC samples (1m interval) "was conducted by Kalassay Ltd., using a Bruker Pioneer X-ray fluorescence instrument, equipped with an end window 4 kW rhodium X-ray tube. Analyses included the weight percentages of Fe, P, S, SiO2, Al2O3, Mn, CaO, K2O, MgO, and TiO2. The sample portion was dried at 105°C for 12 or for 1 hr, depending on whether the sample was wet or dry, respectively. Samples were then crushed to a nominal 90% passing 75 μm. The sample powders were fused in a Herzog automated (RF energized) fusion furnace and cast into 40 mm diameter beads using a 12:22 flux containing 5% sodium nitrate. Matrix corrections were applied using a calculated alpha correction for this combination of flux, tube, and instrument geometry. Previously determined weight ranges were used for both the sample and the flux weight. Lab duplicates, internationally certified reference materials, and reference materials of the same ore type were used as standards and Kalassay Ltd. reported a precision better than 0.01% for all analyses. Duplicate samples were also sent to the Amdel laboratory in Cannington, with good correlation observed."

2.4      Loss on ignition (LOI)

(modified from Haest et al., 2012a)

For measurements of the loss on ignition (i.e., LOI), a "predried portion of all samples was heated in an electric furnace to 1,000°C. Goethite will release its strongly bonded water and its OH groups between 260° and 425°C (Strezov et al., 2010), organic matter is completely ignited by 550°C (Dean, 1974), aluminosilicate clay materials will decompose between 530° and 605°C (Strezov et al., 2010), and inorganic carbon will be oxidized and lost as CO2 between 700° and 850°C (Dean, 1974). A single LOI measurement at 1,000°C will therefore potentially include the mass loss of all these components".

2.5      Sample Storage

Drill core trays, field samples and XRF standards are all stored at the Australian Resources Research Centre (ARRC), 26 Dick Perry Avenue, Kensington, WA 6151. Samples can be viewed and investigated at the ARRC, using local analytical facilities. The custodian of the samples is Carsten Laukamp.

3       Software and Processing Methods

3.1      Processing of hyperspectral drill core data

Hyperspectral drill core data were processed using The Spectral Geologist software (TSGTM) by interpreting the abundance, composition and/or crystallinity of selected mineral groups and species using the Multiple Feature Extraction Method. A list of scripts applied to the hyperspectral drill core and rock chip data can be found in: drill hole data/RC_hyperspectral_geochem/GeoscienceProductDescriptions_ProximalHyperspectral.xlsx.

3.2      Image Processing

The processing strategy for generating geoscience products from AMS data, such as the Kaolin Crystallinity (Table 1) is based on the Queensland Next Generation Mineral Mapping Project (Cudahy et al., 2008) and builds on the quality control of the acquired data. In the case of image processing of hyperspectral remote sensing data, well calibrated radiance-at-sensor or surface reflectance data are required. Levelling and statistics-based methods were avoided as these introduce undesirable scene-dependencies with the result that image products from different areas are not comparable. Physics-based reduction models were applied to the remote sensing data, using the image processing software ENVITM. Complicating effects were removed in their order of development (e.g. first instrument, then atmospheric, followed by surface effects) through either offsets or normalization.

3.3      The Multiple Feature Extraction Method

(modified from Laukamp et al., 2010)

A key step in the development of geoscience products is the extraction of mineralogical information from the calibrated infrared reflectance spectra. In hyperspectral proximal (e.g. HyLoggingTM) and remote (e.g. AMS) sensing technologies the visible-near (VNIR: ca. 350 – 1350 nm), short-wave (SWIR: ca. 1350 – 2500 nm) and thermal (TIR: ca. 8000 – 12500 nm) infrared part of the electromagnetic spectrum are used to infer abundance and composition of various geological materials. The mineralogical information is captured in the reflectance spectra in absorptions features, which are based on the physicochemical characteristics of the various minerals. Feature extraction methods (Cudahy et al., 2008) can be used to determine the mineralogy of a sample material. The advantage of the multiple feature extraction method (MFEM) is that the associated scripts are not biased on a training dataset or library spectra, but are based only on the visible and/or infrared active functional groups of minerals (see Laukamp et al., 2011, for more details). This allows the same scripts to be applied to remote sensing and proximal hyperspectral data, which enables a straightforward integration of for example subsurface (e.g. HyLoggingTM) and surface data (e.g. HyMap) in 3D modelling software packages. Complications, such as spectrally overlapping materials, are removed by the application of thresholds. Interferences of mineralogical information with other surface materials such as vegetation can be evaluated by using a multiple linear regression model for unmixing vegetation from hyperspectral remote sensing data (Rodger & Cudahy, 2009).


4       Currently publically available data

These are the links to the Rocklea Dome dataset:

There are about 130 RC holes. For details (e.g. bore hole geometry), please refer to Haest et al. (2012a,b).

Table 1 Specifications of stored data

main directory

sub directory

file name

type of data




Digital terrain model




Digital terrain model of 100K mapsheet Hardey 2252





Digital elevation model



Digital elevation model

drill hole data



table describing multiple feature extraction scripts applied to hyperspectral data for interpretation of mineralogy















spectral and geochemical data exported from TSG















remote sensing data



AMS product "2200D", showing the relative abundance of Al-clays





AMS product "2200W", indicating compositional changes of Al-smectites and white micas (AlVIAlIV(Fe,Mg)-1Si-1)





AMS product "2250D", showing the relative abundance of chlorite, epidote and/or biotite





AMS product "Carbonate abundance", showing the relative abundance of carbonates





AMS product "Kaolin crystallinity"










AMS product "2320D", vegetation unmixed



AMS product "Al-clay abundance index", vegetation unmixed



AMS product "Ferric Oxide Abundance Index", vegetation unmixed



Digital elevation model


Rocklea Dome exercise


Exercises for analysis of HyLogging data



Suggested answers to exercises for analysis of HyLogging data



PPT-presentation summarising Rocklea Dome exercise and results


5       Publically available teaching material

Maarten and Carsten generated a student exercise about the application of hyperspectral drill core and chips data for iron ore resource characterisation using The Spectral Geologist Software (TSGTM).

Exercise: StudentExercises_Rocklea.docx

Answers: Answers_CIDexercises.docx  

PPT for teaching: MinSpec_Workshop_7RockleaDomeTSG_HandsOn.pptx

Maarten generated a 3D model in Gocad and Leapfrog. Related files are available from Carsten Laukamp.

Related peer-reviewed articels and conference presentations:

Haest, M., Cudahy, C., Rodger, A., Laukamp, C., Martens, E., Caccetta, M. (2013): Unmixing vegetation from airborne visible-near to shortwave infrared spectroscopy-based mineral maps over the Rocklea Dome (Western Australia), with a focus on iron rich palaeochannels.- Remote Sensing of Environment, 129, 17-31.

Haest, M., Cudahy, T., Laukamp, C., Gregrory, S. (2012): Quantitative mineralogy from visible to shortwave infrared spectroscopic data - I. Validation of mineral abundance and composition products of the Rocklea Dome channel iron deposit in Western Australia.- Economic Geology, 107, 209 - 228.

Haest, M., Cudahy, T., Laukamp, C., Gregrory, S. (2012): Quantitative mineralogy from visible to shortwave infrared spectroscopic data - II. 3D mineralogical characterisation of the Rocklea Dome channel iron deposit, Western Australia - Economic Geology, 107, 229 - 249.

Haest, M., Cudahy, T., Rodger, A., Laukamp, C., Martens, E., Caccetta, M. (2012): Creating a seamless 3D mineral map: applications to channel iron ore exploration in the Rocklea Dome.- Proceedings of the 34th IGC, 5. - 10.08.2012, Brisbane, Australia, p1943.

Haest, M., Cudahy, T., Laukamp, C. (2012): Application of infrared spectroscopy-based mineralogy to channel iron ore resource evaluation.- Resource Evaluation and Mining 2012, 30.03.2012, Perth, 29-33.

Haest, M., Cudahy, T., Laukamp, C., Ramanaidou, E., Gregory, S., Stark, C., Podmore, D. (2011): Characterisation of bedded and channel iron ore resources using HyLoggingTM.- To: Iron Ore Conference 2011, 11.-13.07.2011, Perth, Australia, p. 249 - 256.

Haest, M., Cudahy, T., Laukamp, C., Gregory, S. (2011): Creating a Seamless, Semi-Quantitative 3D Mineral Map with Airborne and Drill Core Infrared Spectroscopic Data.- 7th EARSeL Workshop on Imaging Spectroscopy, Edinburgh, Scotland, 11-13.04.2011.

Huntington, J., Whitbourn, L., Laukamp, C., Schodlok, M., Yang, K., Haest, M., Quigley, M., Mason, P., Berman, M., Green, A., (2011):  Probing Australia’s Subsurface with New 1, 2 and 3D Hyperspectral Logging Technologies for “Whole-of-Mine-Life” Mineralogical Characterisation.- Extended Abstracts of the AIG Exploration Technologies Meeting 2011, Perth, Australia, 40-47.

Haest, M.1, Cudahy, T., Cardy, M., Hackett, A., Laukamp, C. (2010): Quantification of Hylogging data for the development of a 3D pysicochemical model of channel iron ore.- NVCL Symposium, Canberra, Australia, 8. - 9. July 2010.

Haest, M., Laukamp, C., Cudahy, T., Gessner, K., Gregory, S. (2010): Hyperspectral insights in the 3D Mineralogy of a channel iron ore deposit: implication to exploration, mining and ore genesis.- In: Geological Society of Australia, 2010 Australian Earth Science Convention (AESC) 2010, Earth Systems: change, sustainability, vulnerability. Abstract No 98 of the 20th Australian Geological Convention, p. 155 - 156.

Other references

Berk, A., Anderson, G. P., Acharya, P. K., Bernstein, L. S., Muratov, L., Lee, J., et al. (2006). MODTRAN (TM) 5: 2006 update — art. no. 62331F. In S. S. L. P. E. Shen (Ed.), Algorithms and technologies for multispectral, hyperspectral, and ultraspectral imagery, XII Pts 1 and 2. (pp. F2331–F).

Berk, A., Cooley, T. W., Anderson, G. P., Acharya, P. K., Bernstein, L. S., Muratov, L., et al. (2004). MODTRAN5: a reformulated atmospheric band model with auxiliary species and practical multiple scattering options. In K. P. C. A. C. M. R. P. R. H. S. N. I. Schafer (Ed.), Remote sensing of clouds and the atmosphere, Ix. (pp. 78–85).

Cocks, T., Jenssen, R., Stewart, A., Wilson, I., & Shields, T. (1998). The HyMap™ airborne hyperspectral sensor: The system, calibration and performance. 1st EARSEL Workshop on Imaging Spectroscopy, Zurich, October 1998 ( Hymap_specs.pdf)

Cudahy, T.J., and Ramanaidou, E.R., 1997, Measurement of the hematite: goethite ratio using field visible and near‐infrared reflectance spectrometry in channel iron deposits, Western Australia: Australian Journal of Earth Sciences, v. 44, p. 411−420.

Cudahy, T., Jones, M., Thomas, M., Laukamp, C., Caccetta, M., Hewson, R., Rodger, A., Verrall, M. (2008): Next Generation Mineral Mapping: Queensland airborne HyMap and satellite ASTER surveys 2006-2008.- CSIRO report P2007/364, 161pp.

Dean, W.E., 1974, Determination of carbonate and organic-matter in calcareous sediments and sedimentary-rocks by loss on ignition—comparison with other methods: Journal of Sedimentary Petrology, v. 44, p. 242−248.

Huntington, J., Mauger, A., Skirrow, R., Bastrakov, E., Connor, P., Mason, P., Keeling, J., Coward, D., Berman, M., Phillips, R., Whitbourn, L., Heithersay, P., and AusIMM, 2004, Automated mineralogical logging of core from the Emmie Bluff, iron oxide copper-gold prospect, south Australia, Pacrim 2004 Congress, 2004: Parkville Victoria, Australasian Institute of Mining and Metallurgy Publication Series, p. 223−230.

Laukamp, C., Cudahy, T., Caccetta, M., Chia, J., Gessner, K., Haest, M., Liu, Y.C., Rodger, A. (2010): The uses, abuses and opportunities for hyperspectral technologies and derived geoscience information.- AIG Bulletin, 51(Geo-Computing 2010 Conference, Brisbane, September 2010): 73-76.

Laukamp, C. (2011): Short Wave Infrared Functional Groups of Rock-forming Minerals.- CSIRO report EP115222.

Morris, R. C., & Ramanaidou, E. R. (2007). Genesis of the channel iron deposits (CID) of the Pilbara region, Western Australia. Australian Journal of Earth Sciences, 54, 733–756.

Ramanaidou, E.R., Morris, R.C., and Horwitz, R.C., 2003, Channel iron deposits of the Hamersley Province, Western Australia: Australian Journal of Earth Sciences, v. 50, p. 669−690.

Rodger, A. (2011). SODA: A new method of in-scene atmospheric water vapor estimation and post-flight spectral recalibration for hyperspectral sensors: Application to the HyMap sensor at two locations. Remote Sensing of Environment, 115, 536–547.

Rodger, A. & Cudahy, T. (2009): Vegetation corrected continuum depths at 2.20 mm: An approach for hyperspectral sensors.- Remote Sensing of Environment, 113: 2243-2257.

Strezov, V., Ziolkowski, A., Evans, T.J., and Nelson, P.F., 2010, Assessment of evolution of loss on ignition matter during heating of iron ores: Journal of Thermal Analysis and Calorimetry, v. 100, p. 901−907.


    Add new comment