GEOPHYSICS, VOL. 77, NO. 5 (SEPTEMBER-OCTOBER 2012); P. WC123–WC132, 20 FIGS., 2 TABLES.

10.1190/GEO2011-0514.1

Targeting nickel sulfide deposits from 3D seismic

reflection data at Kambalda, Australia

Milovan Urosevic1, Ganesh Bhat2, and Marcos Hexsel Grochau3

ABSTRACT over the ultrasalty lake surface produced seismic data of very
high quality. Processing focused on computation of accurate sta-
The greenstone belts of the Yilgarn Craton, Western Austra- tic and dynamic corrections, whereas imaging was helped by the
lia, host numerous Archaean gold, nickel, and iron ore deposits. existing geologic model. Advanced volumetric interpretation
These deposits typically are found in complex geologic struc- supported by seismic forward modeling was used to guide map-
tures hidden by a deep, heterogeneous, and often conductive ping of the main lithological interfaces and structures. Forward
regolith profile. This added complexity limits the depth of modeling was carried out using rock properties obtained from
penetration for the potential field methods, but at the same time ultrasonic measurements and one borehole, drilled in the proxi-
opens new revenue possibilities through the application of seis- mity of the 3D seismic volume. Using this information, geo-
mic methods. To explore this opportunity, we acquired high- metric constraints based on the typical size of ore bodies
resolution, experimental, 3D seismic data over Lake Lefroy found in this mine and a simple window-based seismic attribute,
in Kambalda, Western Australia. The main objective was to several new targets were proposed. Three of these targets sub-
map exceptionally complex, deep structures associated with sequently have been drilled and new zones of mineralization
Kambalda dome. Survey design used 3D ray tracing to improve were intercepted. The case study presented demonstrates that
the distribution of the common reflection points across ultrama- high-quality, high-resolution, 3D seismic data combined with
fic-basalt contacts which host numerous small, high-grade nick- volumetric seismic interpretation could become a primary meth-
el sulfide deposits. A combination of small explosive sources, odology for exploration of deep, small, massive sulfide deposits
high-shot/receiver density, and exceptionally good coupling distributed across the Kambalda area.

INTRODUCTION associated with excessively complex geologic structures. Conse-

quently, these deposits are very hard to discover by traditional geo-
Traditionally, nickel exploration was based on geologic and geo- logic or geophysical methods.
chemical methods, such as lithogeochemical and hydrogeochemical The recent boom in nickel mining has prompted mining compa-
analyses (Barnes et al., 2004). Since the 1970s, more and more geo- nies to adopt various combinations of surface and downhole ex-
physical exploration methods, like gravity and electromagnetics,
ploration methods. Known massive sulfide deposits occurring at
have been adopted by the industry (Dentith et al., 1994; Mutton
shallow depths were more or less mined out in the Kambalda region
and Peters, 2004). However, these methods normally are not effec-
tive for depths greater than 400 m, particularly in Australian con- and possibly also around the world. Consequently, miners are now
ditions. Of particular interest to mining companies are massive seeking deep-seated concealed ore bodies. However, target depths
sulfide (MS) deposits, which are low-volume but high-grade depos- of 600 to 1200 m makes exploration reliant on drilling, which is
its. Unfortunately, massive sulfide ore bodies occur at considerable risky and uneconomical. Hence, Consolidated Minerals decided
depths in Archean igneous and metamorphic terrains, and often are to experiment with the seismic reflection method and assess its

Manuscript received by the Editor 19 December 2011; revised manuscript received 30 April 2012; published online 6 September 2012.
1
Curtin University, Department of Exploration Geophysics, and Deep Exploration Technologies Corporative Research Centre (DETCRC), Perth, Western
Australia. E-mail: [email protected].
2
Consolidated Minerals Limited, Perth, Western Australia. E-mail: [email protected].
3
Curtin University, Department of Exploration Geophysics, Perth, Western Australia. E-mail: [email protected].
© 2012 Society of Exploration Geophysicists. All rights reserved.

WC123

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 WC124 Urosevic et al.

applicability for exploration of deep and complex MS deposits in volcanogenic massive sulfides category. The currently accepted hy-
Kambalda, in the eastern goldfields district of Western Australia. pothesis is that during the Archaean era, “lava rivers” were spilled
Such an attempt was encouraged by several successful experimental over this area on top of preexisting basalt rocks. The Ni sulfide par-
surveys reported by Greenhalgh and Mason (1997), Calvert and Li ticles clogged up together to form MS deposits. Because their den-
(1999), Milkereit et al., (1991, 1996, 2000). Moreover, successful sity is higher than the surrounding silicate minerals, most of the
case histories in the Kambalda region, which is thoroughly studied nickel sulfides were deposited at the bottom of lava channels near
by explorers for nickel and gold deposits (Neumayr et al., 2004) the basalt contact. Some of the sulfide masses were remobilized
were reported by Stolz (2003) and Urosevic et al. (2005). during later hydrothermal metamorphic events. As such, most of
The main objective of this study was to assess the true potential the massive sulfide mineralization deposits are of a small size. A
of 3D seismic methodology for delineation of the Kambalda dome typical ore body size will be 3 m thick and occupy an area of
and its complex nickel-bearing structures. The expectation of this 600 m2 . For a specific gravity value of 4000 kg∕m3 and nickel as-
study was to improve the knowledge of subsurface geology related say of 5%, a body of that size will contain approximately 360 tons
to massive sulfide deposits between depths of 300 and 1200 m. For of nickel. At the current market value, such a find would be worth
that purpose, we designed a high-resolution, mini-3D seismic sur- several million dollars. However, a target of such a size is hard to
vey. Available 3D geologic models, core sample tests, and borehole find by drilling, particularly at several hundreds of meters to 1.5
information were used in the design, and also later during the pro- kilometer depths.
cessing and interpretation stages. Advanced volumetric interpreta- The Kambalda nickel sulfide deposits occur on either side of a
tion tools were used to delineate main structures, map rock contacts tectonically controlled, doubly plunging anticline known as the
of relevance for the mineralization, and, in the last stage, to indicate Kambalda Dome (Stone and Masterman, 1998). Due to structural
seismic anomalies that could relate to nickel sulfide deposits. The complexities in the area, exploration for massive sulfides becomes
methodology used resulted in new discoveries. very challenging at greater depths (600 m and deeper). On the po-
sitive side, tests on drill core showed that the elastic properties of
LOCAL GEOLOGY mineralized rocks stand out from their host rock, and are expected to
Nickel sulfide deposits in the Kambalda area are closely asso- provide sufficient contrast in acoustic impedance to produce readily
ciated with Archean komatiite, a type of ultramafic rock (Lesher, detectable reflection events (Figure 1). Some sparse measurements
1989; Cowden and Roberts, 1990). The identification of interfaces of high-grade Ni ore (5.0 to 5.5%) extracted from two boreholes
between the Lunnon basalt, Kambalda Komatiite, Devon Consols close to the northern edge of seismic volume are presented in
basalt, and various intrusive rocks is very important in the nickel Figure 1a. The nickel sulfide samples clearly stand out from the
exploration process. Most individual ore shoots are located either host rocks in the impedance density crossplot. A significant number
on or very close to the contact between Lunnon basalt and the of core tests were completed for BH-336, which penetrated a low-
overlying ultramafic rock termed “Kambalda komatiite” (Stone grade Ni ore 0.5 m thick. A pseudoreflectivity log was created for
and Archibald, 2004). Kambalda sulfide deposits belong to the BH-336, which shows that the low-grade Ni ore plots within the
background trend (Figure 1b). Full-waveform-sonic and density
logs only were available for one surface borehole (LD105). Unfor-
tunately, this borehole was drilled at the southern edge of the seis-
mic volume and dipping 60° away. Consequently, it did not intersect
any of the rock interfaces of interest within the seismic volume.
However, the borehole intersected two very thin (0.2 to 0.3 m) sul-
fide lenses at shallow depths. These thin lenses have high density
(3.5 and 4 g∕cm3 , respectively) which has contributed to an ele-
vated reflectivity as marked by the red arrow in Figure 2. The com-
puted synthetic seismogram for LD105 suggests that sulfides in
contact with the host rock could create an elevated reflectivity.
However, the sonic log is of poor quality and does not permit further
generalizations. We can suggest that the Kapai Slate when in contact
with basalts is likely to produce good reflection. Significant velocity
variations appear to be present within the Kambalda komatiites and
Paringa basalt.

SEISMIC DATA ACQUISITION

The 3D seismic survey was conducted in 2007 by Geoforce Pty
Ltd. Three-dimensional design and logistics were provided by the
Figure 1. Core sample test results for (a) velocity versus density, Department of Exploration Geophysics, Curtin University. Data
using samples from two boreholes that intercepted high-grade was acquired above Beta Hunt nickel mine on Lake Lefroy, south
ore. Ni ore stands out, which gives rise to a good contrast between of the Kambalda township, as shown in Figure 3a. The 3D area was
ore and the host rocks. (b) Samples taken from boreholes drilled nested between abandoned gold mine pits, man-made mullock
close to 3D survey area; pseudoimpedance log constructed from
a large number of core samples measured in BH-336 that inter- heaps, dikes, and old and new mine infrastructure (Figure 3b).
cepted 0.5 m of low-grade sulfides. BH-336 is located over The main causeway connecting St. Ives mine camp to Kambalda
500 m away from seismic volume. cut through the 3D seismic area. The total area covered by the

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 Targeting Ni deposits from 3D seismic WC125

shot/receiver lines was approximately 3.5 km2. The shot-line se-

paration varied from a nominal separation of 100 m to 50 m,
and less (down to 10 m) where patches overlapped. Receiver line
separation was kept to around 90 m. Four patches, each consisting
of six receiver lines with a variable number of channels (up to 500),
were used to cover the 3D area (Figure 3c). Nominal receiver se-
paration was 10 m and shot separation was 20 m. Small explosive
charges (110 g) were deployed in 1.2 to 1.5 m deep holes. On the
hard ground, away from the salt lake, a free fall weight drop
(375 kg) was used to generate seismic energy. The nominal fold
for 5 × 10 m bins was design to be around 70. The fold analysis
and the processing used a 10 × 10 m bin size to double the fold.
To analyze the common reflection point (CRP) distribution over
the Kambalda Dome, we used 3D ray tracing. Comparing the
CRP to a conventional common midpoint (CMP) fold helped the
final survey design. It was necessary to introduce additional source
lines running approximately in a south–southwest — north–north-
east direction to improve the CRP fold (Figure 4). A higher bin size Figure 2. Full elastic synthetic seismogram computed for borehole
was used for the CRP display due to the fact that some rays were LD-105 using offset range 0–900 m. The rock codes are: PBS —
Paringa basalt, KSL — Kapai Slate, DCB — Devon Console
lost in the computation process. Therefore, the surface and the un- basalt, KAM — Kambalda komatite, UNK — Unknown rock
derground fold maps only can be compared in a qualitative sense. It type, LUN — Lunnon basalt. Stratigraphic tops and bottoms are
is clear that the displacement between the CMP and CRP points can marked with orange and gray zones, respectively. White zones stand
be significant, and that the CRP fold is more appropriate for illu- for unknown rock units. White arrows show thin high-density sec-
tions. Red arrow shows elevated reflectivity, resulting probably
mination of dipping structures such as the Kambalda Dome. from a combined effect of thin bed interference and the presence
The final 3D design, however, was not followed exactly due to a of high density lenses. Sonic log is of a low quality and should
large storm and subsequent surface flooding of the salt lake midway be taken with caution.
through the acquisition of patch 3. It took over three months for the
lake to dry and for DEG to repair the equipment damaged by the
hypersaline water. Upon recommencement of the survey, modifica-
tions had to be made to the source and receiver
lines to accommodate ground conditions.
The final distribution of the source and receiver
lines and the related CMP fold are shown in
Figure 5. The average shot density achieved
across the 3D survey was 630 shots∕km2 . The
receiver density was around 550 receivers∕
km2 , some 80 receivers∕km2 less than designed.
These changes produced lower fold in some parts
of the survey. The final area covered by the 3D
fold is approximately 2.2 × 1.2 km. Inlines are
oriented in southeast–northwest direction, cross-
lines are running southwest–northeast.
Data processing and interpretation was carried
out at Curtin University. Acquisition and proces-
sing stages took about 12 months to be com-
pleted, and in August 2008 the seismic volume
was available for the interpretation stage.

DATA PROCESSING
Noise generated by the underground drilling
activities and heavy vehicles of Beta Hunt nickel
mine were, to a large extent, avoided by careful
recording that included frequent breaks in
acquisition until the noise level was acceptable. Figure 3. Three-dimensional seismic experimental survey was carried out close to Kam-
Consequently, excessive ambient noise bursts balda, some 600 km west of Perth. Aerial photo shown in (a). Salt lake is shown in blue
contaminated less than 5% of the shots recorded. (flooded at the time). Broun is the elevated regolith surface. Most of the 3D area was
located on the salt lake (Lake Lefroy) and as shown in (b) it is surrounded by: Aban-
The data acquired after flooding included doned gold mines (AGM), Mine dumps (MD), dikes (D), main causeway, or mine road
reacquiring half of patch three (west of cause- (MR). Receiver and source lines are labelled as R-line and S-line, respectively. Four
way, Figure 3c). The repeated geophone stations overlapping patches were used for this survey, as shown in (c).

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 WC126 Urosevic et al.

did not exactly match previous positions due to additional mining One of the key objectives in hard rock seismic data processing in
activities and ground disturbance, resulting in a misalignment be- Australia is to “homogenize” the regolith; that is, to solve for time
tween old and new stations. Noise bursts and misaligned geophone delays caused by heterogeneous near-surface low-velocity zones
stations for the repeated patch made the application of surface (Urosevic et al., 2007). In the case of the Beta Hunt 3D survey,
consistent amplitudes less effective. Consequently, true relative- which was shot mainly across the salt lake, the regolith heteroge-
amplitudes — strictly speaking — were not preserved because, neities were only moderate in comparison to other areas in the Kam-
after initial amplitude compensation, we applied a long window balda region. A 3D raw shot record before and after application of
(1000 ms) AGC operator. However, coupling of the source and re- refraction statics is shown in Figure 7. The total magnitude of statics
ceivers was exceptionally good because over 90% of the 3D area (shot þ receiver) topped 54 ms over the salt lake and was up to
was acquired over hypersaline clay, close to suspension in places. 70 ms away from the lake due to elevated shot statics (Figure 8).
Consequently, shot records that were partially scaled (spherical Shot statics were significantly greater over the hard ground in com-
divergence correction and gain) were not too dissimilar from shots parison to shots fired on the salt lake. Accurate computations of the
with the AGC applied (Figure 6). The processing flow is depicted in delay times through the regolith were necessary to retain frequency
Table 1. content of the stacked section and achieve reflection continuity. A
geologically meaningful velocity field was estimated through con-
stant velocity stacks (CVS) computed along the full line length in
the inline and crossline directions at an interval of 200 m. An ex-
ample of CVS panels for the central crossline is shown in Figure 9.
It is relatively easy to form a mental picture of the underground
structures and derive discrete velocity functions from CVS panels.
Subsequent velocity runs included interactive semblance analysis
centered around CVS-derived velocity functions. The maximum de-
parture of the semblance-derived velocities from CVS guide func-
tions was limited to
200 m∕s. Therefore, the main purpose of
semblance analysis was to fine-tune CVS derived velocities.
A selection of four inlines, extracted from the cube after applica-
tion of residual statics and AGC before stack, is shown in Figure 10.
Very strong reflectivity originating from Beta West complex is seen

Figure 4. The surface fold map using the distribution of CMP

points (top left), underground fold map using the distribution of
CRP points (top right), and CRP to CMP difference map (bottom).
Displacements between the two along the structure are proportional
to the dip angle.

Figure 6. Three-dimensional shot record after: (a) application of

spherical divergence and gain corrections and (b) after AGC
1000 ms operator was applied in addition. Receiver lines are la-
Figure 5. The final CMP fold map for bin size 10 × 10 m. Source belled as RL-1 to RL-6. The two records are similar apart from
lines shown in black are mainly running in southwest–northeast some noisy sections like the one at the start of RL-3 and RL-6,
direction. Receiver lines (multicolored) are oriented southeast– which are overcalled by the application of AGC. This effect is
northwest. Note that only the central part of the 3D area has a fold diminished by the subsequent application of deconvolution and
over 80. filtering.

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 Targeting Ni deposits from 3D seismic WC127

at around 250 ms. The difference in overall reflectivity between the design that used high data density, and (3) high-precision data
western and eastern part of the section is apparent. The amplitude processing, particularly static and dynamic corrections.
ratio of strong to weak reflectors is of the order of 20∶1. It also
appears that the reflection continuity could benefit from further DATA INTERPRETATION
scaling. This was not performed, but the inspection of 3D stacks Application of seismic technology for the mineral industry is not
at various stages of the processing suggested that the adopted pro- a common practice, hence typical mining software designed for re-
cessing flow has not significantly affected relative amplitudes. Se- source calculations, such as Micromine, Vulcan, and Surpac, are not
lected in and crosslines, after application of DMO and 3D depth suitable for interrogating 3D seismic data cube. To overcome this,
migration, are shown in Figure 11. High data quality enables early we loaded our data into an oil-type interpretation package capable
identification of a known Beta shoot Ni deposit and a less explored of volumetric interpretation. The issue then becomes how to inte-
deep Alpha shoot deposit. High data quality is attributed to (1) next grate 3D geologic models, other geophysical measurements, and
to perfect coupling of source and receiver over the salt lake, (2) 3D mining information with a seismic cube. The use of local mine co-
ordinates in production further complicates seismic data analysis.
Apart from geologic models, which are important for guiding
Table 1. Basic 3D processing flow. the initial interpretation, it is necessary to calibrate seismic to assign
geologic meaning to the measured reflectivity, achieve proper cali-
bration, and speed up the interpretation process. Because only one
Geometry specification/binning sonic log, of low quality and away from the seismic cube, was avail-
Trace edit/reverse/mute able, we loaded drillhole trajectories and intercepts (or tops) corre-
sponding to relevant geologic interfaces to help calibrate
Shot energy equalization, spherical divergence correction, and seismic data.
6DB/s gain
First-arrival picking, editing, and refraction static calculations VOLUMETRIC VERSUS CONVENTIONAL
Signal and noise analysis INTERPRETATION
Application of spiking deconvolution (80 ms operator, 1000 ms
window), followed by broad band-pass filtering There are two different methodologies available to seismic inter-
preters. One approach is to interpret 3D volumes based on 2D slices,
Constant velocity stacks: Selected inlines and crosslines (200 m
increments) interpreting line-by-line. This is called “conventional” interpreta-
tion and has been used for decades in the oil industry. Another ap-
Application of NMO
proach, called “volumetric,” is to interpret 3D volumes in a real 3D
Stack 1, pick 3D horizon for residual statics way. This is performed by applying opacity and transparency filters,
Residual reflection static computation displaying and rotating objects, using special visualization techni-
Stack 2, with surface consistent residual reflection statics applied ques (like stereoscopy), and so on. The latter approach allows a
Dip moveout correction (3D DMO, shot domain), inverse NMO much more comprehensive understanding of the volume under in-
Final velocity analysis vestigation. In the present study, we interpret the Kambalda seismic
volume using the volumetric interpretation approach.
DMO stack (scaled, AGC 1000 ms)
Initially, structural patterns from seismic data were analyzed, and
F-XY deconvolution then the semblance attribute was computed, aiming to enhance dis-
DMO FXY stack continuities. Despite testing various parameters, semblance did not
Poststack time migration work well due to the presence of no data zones and associated low
Depth conversion S/N in some parts of the cube. Consequently, a simple “Perigram”
Depth migration (low-pass filtered envelope, multiplied by the cosine of the instan-
taneous phase) and amplitude were the most useful attributes to map
faults at all scales (Figure 12). Investigating the seismic volume in
vertical planes at different directions, and in horizontal and dipping
slices, made it possible to reveal full diversity and complexity of
underground structures. Normal, reverse, dextral, and other fault
patterns were identified based on breaks in continuity of several
discrete reflections, or changes in dips of successive reflectors,
or based on a termination of a reflective package (Figure 12).
The major (known) faults, such as the Alpha Island fault, were in-
terpreted first. Despite the decrease in seismic quality toward the
edges of the seismic volume where the fault is located, it was still
possible to identify and interpret this fault by examining depth slices
and considering its predominantly dextral movement component.
Figure 13a shows the previously interpreted Alpha Island fault in-
ferred from drilling and the new interpretation based on seismic
Figure 7. A 3D shot record before (left) and after (right) application
of refraction statics. Equalized amplitudes shown. Note that statics data. Both interpretations show similar direction (north–northeast
have significant magnitude across receiver line 2 (white arrow). The — south–southwest) in the area covered by the seismic volume,
shot static also is large, as marked by the horizontal yellow arrow. but they differ in dip angle and fault complexity. We interpreted

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 WC128 Urosevic et al.

this fault as a system of two subverticals, gently diverging faults

toward the surface that are slightly offset with respect to
the previous interpretation (Figure 13b). A significant number of
small-scale faults can be interpreted from the seismic cube. Some
of these faults are visible only in a few time or depth slices and then
disappear. Some fault types, like reverse faults, are better seen in
vertical cross sections as shown in Figure 14.
Following structural interpretation, the most relevant geologic
interfaces were interpreted in 3D. They were recognized and

Figure 8. Color-coded shot and receiver statics displayed along the

lines. Shot lines are running southwest–northeast, receiver lines
southeast–northwest. A total maximum static of 70 ms is encoun-
tered only when shots were positioned on the hard ground (dikes,
mullocks), away from the salt lake. Over the salt lake, the total static
is not exceeding 54 ms, which is of low-to-moderate magnitude for
this region.

Figure 9. Constant velocity stacks for the central crossline 65. De- Figure 11. Extracted lines from seismic cube: (a) Inlines (top) and
rivation of a geologically meaningful velocity model is relatively (b) crosslines. Line interval is 20 m. High data quality enables early
straightforward from these panels. Identification of artifacts also identification of a known Beta shoot Ni deposit and less explored
is much easier than when using semblance analysis. CVS analysis Alpha shoot. Crossline sections appear smoother due to different
was repeated after the application of residual static corrections and horizontal scale. Average inline is approximately 2.4 km in length,
DMO. about twice as long as the average crossline.

Figure 10. Selected inlines after preprocessing and application re- Figure 12. Inlines 87 and 89 depth sections: (a) Amplitude section
sidual reflection statics. High-reflectivity patches are present in each and (b) Perigram section (low pass envelope scaled by instanta-
section. The amplitude variation is significant, as is the difference neous phase, corrected for 90 phase shift. Subtle faults easily are
between strong and weak events. identified in Perigram section as indicated by the arrows.

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 Targeting Ni deposits from 3D seismic WC129

interpreted on seismic data using existing geologic models, litho- from 10 to 0 m. As expected, the resultant reflectivity is roughly
logical markers from drillholes, core sample tests, and forward- proportional to the thickness of the ore body and a strong amplitude
modeling experiments. The main geologic units interpreted are anomaly is not expected for nickel mineralization that is less than
the Devons Consol basalt, Kambalda Komatiite, intrusive rocks into 43 m in thickness.
the Kambalda Komatiite, and intrusive rocks into basalts. Opacity The seismic response of a 10 m thick nickel sulfide body not in
filters were particularly effective in highlighting sill and dike intru- constant contact with and dipping away from the ultramafic-Lunnon
sions into ultramafic rocks. The seismic signature of those inter- basalt contact is more complex due to the interference of multiple
faces has predominantly high amplitudes with
good lateral continuity. Some of the intrusives
mapped in the seismic volume are shown in
Figure 15.

TARGETING MASSIVE SULFIDE

DEPOSITS
The Beta Hunt active underground mine head-
ings are located between 700 to 1000 m vertically
below the surface. Mine workers, haulage trucks,
and other heavy mine machinery approach the
mine headings and ore bodies by driving down
about 9 km distance through a winding network
Figure 13. Alpha Island fault plane inferred from drilling is marked with blue arrow
of declines. Productive MS ore bodies have a (dark plane, partially shaded in white, outlined by a thin pink line). The evidences
thickness ranging from 0.5 m to 5 m, and very of the actual fault plane (thick red line) position can be seen in two different depth slices:
rarely to 10 m. Strike and dip extension of these (a) 768 m and (b) 1266 m. We reinterpret Alpha fault as consisting of the two planes with
ranges between 5 to 50 m. Most of the ore bodies similar directions, but diverging toward the surface. In fact, it is possible to see that
strong amplitudes are displaced in accordance with the dextral movement, as shown
were discovered by drilling an array of diamond by the black dashed line in (b). Other faults also can be seen as indicated with brown
drillholes into favorable geologic targets from sui- arrows in (a). Yellow discs denote position of various intrusives, as intercepted by under-
table underground drill chambers as shown in Fig- ground drillholes.
ure 16. Some targets are generated by downhole
transient electromagnetic surveys. At these work-
ing depths, discovering or missing any massive sulfide ore bodies
becomes commercially significant. Therefore, additional informa-
tion provided by seismic data can help increase the drilling success
rate. Similarly, precise information about the orientation of geologic
structures and lithology is critically important to the development of
underground mine headings.
The main objective of increasing geologic knowledge of this area
using seismic data was successfully achieved in the first phase of
seismic interpretation. Encouraged by the quality of seismic images,
we decided to direct further efforts toward characterization of seis-
mic amplitude anomalies. The idea was to investigate if we can infer
a seismic pattern or find appropriate seismic attributes that could Figure 14. Reverse faults (arrows) with vertical displacement visi-
contain an imprint of a typical nickel sulfide mineralization at this ble only in a vertical cross section. Crossline 115 displayed en-
larged. Many faults of different types pervade 3D seismic cube.
site. Thus, our ultimate goal became to attempt to target the location
of nickel sulfide deposits directly from the seis-
mic data. One of the foreseen difficulties is re-
lated to fast rock velocities that results in a
long seismic wavelength, as is typical for hard
rock environments. It was clear that thin nickel
bodies will not be resolved by a wavelet of
80 m or longer. However, the detection of nickel
ore bodies will be helped by their high reflectiv-
ity. The final seismic response will depend on the
ore thickness, and also the geometry of the
structures. To investigate this, we performed sim-
plified forward-modeling experiments using
typical thickness and geometries of nickel ore
Figure 15. Sills and dikes intruded into ultramafic rocks (Kambalda komatite) are (a)
bodies. Figure 17 shows the seismic response extracted with the application of opacity filters and (b) subsequently mapped. The seis-
of a nickel-wedge, located at the ultramafic- mic signature of those interfaces has predominantly high amplitudes with good lateral
basalt contact. The thickness of the body reduces continuity, extending for several hundreds of meters, typically 500–800 m.

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 WC130 Urosevic et al.

reflecting interfaces (Figure 18). The interference pattern changes because this interface is the most likely place for nickel sulfide de-
from constructive to destructive, and for a specific geometrical con- posits to occur. Very high reflectivity originating from intrusives
dition, a 10-m thick nickel body may not be detected by seismic. In into ultramafics (often porphyries) was excluded from the targeting
all other cases, constructive interference occurs and the reflections process. Discrimination of intrusives was largely based on signifi-
from nickel deposits are amplified to a variable degree. If the nickel cant divergence from the interpreted Lunnon interface and areal ex-
body is at a favorable separation/distance from the ultramafic/basalt tent of typical nickel bodies encountered at this mine.
interface, then constructive interference is at its maximum and a A set of seismic attributes was computed in an attempt to further
strong seismic anomaly is produced. For noisy data, mapping these enhance nickel deposits. None of the standard instantaneous attri-
variations may be challenging. butes helped the detection process. Instead, we used the average
Having a better understanding of the seismic signature of various maximum amplitude over a 100 m window, 60 m above,
nickel ore body geometries, we investigated the seismic response and 40 m below the Lunnon basalt and by limiting their size to
close to stratigraphic markers (intercepts) of existing drillholes 20–100 voxels, some 78 potential nickel targets were identified
(Figure 16). No boreholes were drilled from the surface within (Figure 19). An improved target selection was subsequently
the 3D volume. A good correlation was observed between the main achieved by computing the root mean square (rms) amplitude for
stratigraphic markers and our interpreted contacts such as the ultra- a narrow window encompassing only 10 m above and 4 m below
mafic/Lunnon basalt interface. Further interpretation was then the Lunnon basalt, and limiting the spatial extent of the potential
guided by expected reflectivity in the vicinity of this interface mineralizations to 30 × 30 m. Figure 20 shows the rms amplitude
around the ultramafic-Lunnon basalt interface. There are two
known deposits as marked by the black arrows. Four other potential
targets were identified in this section of the seismic volume. Three
targets (white circles) shown in Figure 20 were drilled from under-
ground drill chambers. Individual drillhole details, ore intercepts,
and assay results are shown in Table 2. Note that these diamond
core drillholes have positive dip, which means that the targets were
intercepted from underneath, because a drill chamber is located at
deeper level than are the ore bodies.

DISCUSSION
Massive sulfide deposits occurring at shallow depths in the Kam-
balda area have now been mined, and the main objective of explora-
Figure 16. Arrays of diamond drillholes made from suitable under- tion programs is to discover deeper targets to extend the life of
ground drill chambers to locate small size massive sulfide ore existing mines. Consequently, the main objective of this experimen-
bodies in Beta Hunt nickel mine.
tal seismic program was to investigate the potential of 3D seismic
reflection methods for mapping complex deep structures that could
host massive sulfides. Apart from the issues typical for the applica-

Figure 17. A simplified geologic model containing a nickel-wedge

body situated at the ultramafic-basalt interface (top) and corre-
sponding zero-offset seismic response computed with stress- Figure 18. Seismic response of a constant thickness, dipping nickel
velocity, finite-difference code, and 75 Hz Ricker wavelet (bottom sulfide body located close to the ultramafic-basalt interface. The
image). The elastic parameters used were: UM (V P ¼ 5600, same modelling parameters were used as in Figure 17. The inter-
ρ ¼ 2.9 g∕cm3 ); Lunnon basalt (V P ¼ 5500, ρ ¼ 2.85 g∕cm3 ); ference pattern changes from constructive (black arrow) to destruc-
Ni (V P ¼ 4500, ρ ¼ 4.5 g∕cm3 ). The nickel wedge is beyond tive (orange) as a function of the separation distance between the
the resolution limit, hence the reflectivity is roughly proportional nickel body and the ultramafic-basalt interface. Overall, a moderate
to the thickness of the body. It is expected that very thin nickel amplitude response is expected unless nickel body is at the favorite
bodies (below 3 m thickness) will not produce strong amplitudes. separation distance (5–10 m from the interface).

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 Targeting Ni deposits from 3D seismic WC131

tion of hard rock seismic in brown fields, specific to Beta Hunt was mine. A 3 m thick by 30 m strike by 20 m dip is considered a modest
the challenge of detecting deep seated concealed ore bodies. ore body at Beta Hunt. The value of such a discovery is well above
The Beta Hunt Nickel Mine hosts numerous small sized massive five million dollars, and most of the proven mineralization is more
sulfide ore bodies. In quantities, they represent good value for the or less of this value. Of course, there will be large costs involved in
ore recovery from these depths; however, cost can be reduced by
using already established infrastructure should nearby ore bodies
be located. The probability of hitting such small bodies by under-
ground fan drilling is small. Hence, the role of high-resolution seis-
mic for targeting ore bodies underneath Lake Lefroy may be crucial
for further development of Beta Hunt and expansion of nickel
mining in the Kambalda region.
Further improvements of the methodology presented here could
be made by attempting true relative amplitude processing. If any
time in the future, high-quality full-waveform sonic and density
logs become available within the 3D seismic volume, it would
be possible to achieve proper correlation and inversion of the seis-
mic data. The impedance cube should bring additional value and
further increase the precision in targeting nickel sulfide deposits.
Three-dimensional surveys of the size of 9 km2 or larger would en-
able further refinements of the interpretation process.

Figure 19. Potential “geobodies” or “geoanomalies” (bright colors)

mapped in 3D volume, in the vicinity of ultramafic/Lunnon basalt CONCLUSION
interface (brown surface). The selection of potential targets is based
on two criteria only: (1) the average maximum amplitude in a win- The Beta Hunt experimental survey was the first 3D survey ever
dow computed around the interface of interest, and (2) the expected acquired over Lake Lefroy. A combination of several factors, such
size of a mineralization. Known Ni occurrences (borehole intersec- as high data density, very good source/receiver coupling, deploy-
tions) are represented by tiny white dots within white ellipses. ment of small explosive charges, and high precision data processing
produced a high-resolution, high-quality seismic data cube. The 3D
volumetric seismic interpretation project was successful in achiev-
ing the primary objectives of mapping the main rock units down to
2-km depth. The Alpha Island fault system, one of the most impor-
tant structural elements in the area, is now precisely mapped and its
complexity better understood. Many small-scale faults and offsets
were identified within the Kambalda structural domain. The knowl-
edge gained from these structural models will be useful for future
mine infrastructure design and development.
High-precision correlation of seismic data was not possible be-
cause no surface boreholes have been drilled into the 3D seismic
cube. All of the information available for the interpretation was ob-
tained from core sample tests, underground drilling (unlogged), and
borehole logs outside the seismic volume. Despite the lack of ex-
plicit correlation, numerous potential massive sulfide targets were
identified within the seismic volume during the interpretation pro-
Figure 20. The rms amplitude map computed in a window around cess. The use of numerical modeling proved to be important for
ultramafic/basalt interface. Known Ni occurrences are marked by
black arrows. Proposed new targets are labelled by circles. White predicting the seismic response produced by small nickel bodies
circles are seismic anomalies verified by underground drilling as and subsequently devising a targeting strategy. Simple seismic
mineralized bodies (see also Table 2). Yellow circle has not been attributes in combination with the “Ni size criterion” (i.e., the
verified by drilling at present. restriction imposed on the lateral extent of seismic amplitude

Table 2. Drilling results and assays for three boreholes projected to verify seismic anomalies.

Hole_id Hole type Date drilled Azim Dip Depth From To Intercpt_m Grade_NiS%

BE23-061 core 24-Oct-08 327 38 55.4 20.6 23.7 3.1 7.97

BE23-061 core 24-Oct-08 327 38 55.4 23.7 27.2 3.5 2.64
BE23-062 core 25-Oct-08 334 31 85 61.2 64.8 3.6 5.38
BE23-064 core 26-Oct-08 51 60 48 20 23.7 3.7 6.81

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user
 WC132 Urosevic et al.

anomalies) proved to be useful in delineating new potential targets. REFERENCES

A total of 78 seismic exploration targets initially were interpreted
and ranked according to the level of confidence. Among these tar- Barnes, S. J., R. E. T. Hill, C. S. Perring, and S. E. Dowling, 2004, Litho-
geochemical exploration for komatiite-associated Ni-sulfide deposits:
gets, three of the highly ranked were drill tested. All holes inter- Strategies and limitations: Mineralogy and Petrology, 82, 259–293,
cepted massive nickel sulfide mineralization. These discoveries doi: 10.1007/s00710-004-0051-7.
Calvert, A. J., and Y. Li, 1999, Seismic reflection imaging over a massive
revealed the extension of the known ore bodies, increasing the sulfide deposit at the Matagami mining camp, Quebec: Geophysics, 64,
ore resources. Interpreted exploration targets could play a very im- 24–32, doi: 10.1190/1.1444521.
portant role in improving an otherwise very low drilling success rate Cowden, A., and D. E. Roberts, 1990, Komatiite- hosted nickel sulfide
deposits, Kambalda: Australasian Institute of Mining and Metallurgy
in the Kambalda area. The case study presented demonstrates that Monograph, 14, no. 1, 576–582.
high-quality, high-resolution, 3D seismic data combined with a vo- Dentith, M. C., K. F. Frankcombe, and A. Trench, 1994, Geophysical
signatures of Western Australian mineral deposits: An overview: ASEG
lumetric seismic interpretation could play a very important, if not Special Publications, Extended Abstracts, 1, 29–84.
crucial, role in future discoveries of deep, small, MS deposits dis- Greenhalgh, S. A., and I. M. Mason, 1997, Seismic imaging with application
tributed across the Kambalda Dome. to mine layout and development, in A. G. Gibins, ed., Geophysics and
geochemistry at the millenium: Proceedings of Exploration 97, GEOFX
Started as an experimental survey, this seismic cube can now be Publishers, 585–598.
“revisited” any time in future mine operations to assist with ore Jones, S., 2007, Structural analysis of Beta-hunt, Kambalda: Consolidated
Minerals unpublished internal report.
body targeting and mine development, particularly after new knowl- Lesher, C. M., 1989, Komatiite-associated nickel sulfide deposits: Reviews
edge is acquired through further underground drilling assays, core in Economic Geology, 4, 44–101.
sample tests, and improved geologic models. Malehmir, A., and G. Bellefleur, 2009, 3D seismic reflection imaging of
volcanic-hosted massive sulfide deposits: Insights from reprocessing
Halfmile Lake data, New Brunswick, Canada: Geophysics, 74, no. 6,
B209–B219, doi: 10.1190/1.3230495.
ACKNOWLEDGMENTS Milkereit, B., E. Adam, A. Barnes, C. Beaudry, R. Pineault, and A. Cinq-
Mars, 1991, An application of reflection seismology to mineral explora-
We would like to thank staff of Consolidated Minerals: Rob Hen- tion in the Matagami area, Abitibi belt, Québec, in current research, Part
C: Geological Survey of Canada, Paper 92-1C, 13–18.
derson, Exploration Manager, Tim Craske, former Chief Explora- Milkereit, B., E. K. Berrer, A. R. King, A. H. Watts, B. Roberts, E. Adam,
tion Geologist, Gary Snow, Former Geology Manager and Brett D. W. Eaton, J. Wu, and M. Salisbury, 2000, Development of 3D seismic
exploration technology for deep nickel-copper deposits — A case history
Davies, General Manager for providing us with the opportunity from the Sudbury basin, Canada: Geophysics, 65, 1890–1899, doi: 10
to work in this project. We also thank Abraham Whaanga, Angela .1190/1.1444873.
Storkey, Brett Davies, General Manager Geology, Sarah Jones, and Milkereit, B., D. Eaton, J. Wu, G. Perron, M. Salisbury, E. Berrer, and G.
Morrison, 1996, Seismic imaging of massive sulfide deposits, part II: Re-
Gemma King for the technical discussions, as well as Elizabeth flection seismic profiling: Economic Geology, 91, 829–834, doi: 10.2113/
Florkiewicz for providing dxf surfaces. We would like to thank gsecongeo.91.5.829.
the Center for High-Definition Geophysics (CHDG) at the Depart- Mutton, P., and B. Peters, The Waterloo and Amorac nickel deposits, Wes-
tern Australia: A geophysical case history: ASEG Special Publications,
ment of Exploration Geophysics, Curtin University, for conceiving Extended Abstract, 1–14.
and carrying out this project. We are grateful to Anton Kepic, di- Neumayr, P., J. L. Walshe, L. Horn, K. Petersen, C. Deyell, K. Moran, D.
Howe, K. Connors, N. Stolz, R. Morrison, and S. G. Hagemann,
rector of the CHDG, and now Program Leader for the Deep Ex- 2004, Hydrothermal alteration footprints and gold: CRC Conference,
ploration Technologies Corporative Research Center (DETCRC), 165–169.
and Peter Williams, currently of HiSeis Pty Ltd, who initiated seis- Stolz, E. M., 2003, Direct detection of gold bearing structures at
St. Ives, WA — DHEM versus DHMMR: Exploration Geophysics, 34,
mic surveying and designed data acquisition strategy in hipper-sal- 131–136.
ine conditions of Lake Lefroy in 2005. We thank Geoforce Pty. Ltd. Stone, W. E., and N. J. Archibald, 2004, Structural controls on nickel sulfide
ore shoots in Archaean Komatiite, Kambalda, WA: The volcanic trough
for carrying out very challenging data acquisition program at the controversy revisited: Journal of Structural Geology, 26, 1173–1194, doi:
lake. We thank Sinem Yavuz for helping preparation of figures. 10.1016/j.jsg.2003.11.014.
We are grateful to associate editor A. Malehmir and the reviewers Stone, W. E., M. Heydari, and Z. Seat, 2004, Nickel tenor variations between
Archaean komatiite-associated nickel sulfide deposits, Kambalda ore
G. Bellefleur, P. Hatherly, C. Krawczyk, and one more anonymous field, Western Australia: The metamorphic modification model revisited:
reviewer whose comments and advice are greatly appreciated, and Mineralogy and Petrology, 82, 295–316, doi: 10.1007/s00710-004-0045-
5.
were crucial for improving the quality of the manuscript. Finally, we Stone, W. E., and E. E. Masterman, 1998, Kambalda nickel deposits: Mono-
are grateful to Landmark graphics, a Halliburton company, for pro- graph, 22, 347–356.
viding us with 3D processing and interpretation packages within the Urosevic, M., A. Kepic, E. Stolz, and C. Juhlin, 2007, Seismic exploration of
mineral deposits in Yilgarn Craton, Western Australia: Proceedings of Ex-
Landmark-Curtin educational and research grant. We also are grate- ploration 07.
ful to Hampson & Russell, a CGGVeritas company, and ION Urosevic, M., E. Stolz, and S. Massey, 2005, Seismic exploration of complex
mineral deposits — Yilgarn Craton, Western Australia: 67th Annual
Geophysical Corporation for providing survey design and AVO International Conference and Exhibition, EAGE, Extended Abstracts,
software packages. Z-99.

Downloaded from http://pubs.geoscienceworld.org/geophysics/article-pdf/77/5/WC123/3230220/geo2011-0514.pdf

by CNOOC user