You are on page 1of 14

Ore Geology Reviews 126 (2020) 103776

Contents lists available at ScienceDirect

Ore Geology Reviews


journal homepage: www.elsevier.com/locate/oregeorev

Understanding seismic reflectivity across hydrothermal alteration zones T


associated with gold: Example of the Karari gold deposit, Western Australia
Andre Eduardo Calazans Matos de Souza , Stephanie Vialle

Curtin University, Exploration Geophysics, 26 Dick Perry Ave, Perth, Kensington, WA 6151, Australia

ARTICLE INFO ABSTRACT

Keywords: In this study, we examined whole core samples from a gold deposit in Western Australia, the Karari mine, located
Rock acoustic properties south of the Carosue Dam processing plant. By analyzing the mineralogy and performing petrophysical and
Seismic reflectivity acoustic measurements on 21 core samples representing different stages of hydrothermal alteration, our goal was
Gold exploration to document the values of acoustic impedance as a function of the degree of alteration and identify which
Karari gold deposit
parameters (e.g., mineralogy, porosity, foliation) affect the seismic reflectivity across the alteration zones. To do
Hydrothermal alteration zones
so, we measured the ultrasonic velocities and bulk density at ambient conditions for all samples as well as at
confining pressures for two samples. The acoustic measurements were conducted in the directions perpendicular
and parallel to the foliation of the samples. A subset of samples was selected to identify the mineral phases,
perform automated mineral mapping, measure grain density, and compute porosity using the bulk density–grain
density relation and backscattered electron (BSE) images. Our results show that, regardless of the direction of
acquisition, the acoustic impedance in the background samples is consistently smaller than that in the altered
samples. This difference in acoustic impedance is attributed to a porosity difference of around 2% between the
background and altered samples and is not owing to the presence of alteration minerals. Seismic reflectivity at
the background/hydrothermal alteration interface decreases with increasing pressure owing to the closure of
microcracks. For the Karari site, with a targeted exploration depth between 200 m and 800 m, the altered zones
can be differentiated from the host rock by their acoustic impedance, as the reflection coefficients are expected
to vary between 0.05 and 0.13, depending on the direction of seismic propagation and depth.

1. Introduction several reasons. First, the physical and chemical properties of these
altered zones provide an indirect indication of mineralization. Second,
Direct identification of gold is challenging because the ore is dis- the mineralogical and chemical distributions across the altered zones
persed within the host rock and represents only a minor component, provide key information regarding the physical and chemical condi-
typically occurring at only a few grams of ore per ton of rock (Eilu et al., tions during ore formation. Finally, the extension and thickness of the
1999). However, lode gold deposits are usually associated with hy- hydrothermal alteration zones vary widely from deposit to deposit, but
drothermal alteration zones and are controlled by geological dis- are wider and thicker than the hosted ore deposit, making their de-
continuities such as shear zones, which act as gold-bearing fluid con- tection easier (McCuaig and Kerrich, 1994). However, it is often diffi-
duits (Colvine et al., 1988; Groves et al., 1998). The interaction of fluid cult to assign specific geophysical signatures to the wide diversity of
with the host rock alters its mineralogy, chemistry, and texture, re- geological models of gold deposits, as both host rocks and alteration
sulting in hydrothermally altered zones surrounding the gold bodies minerals are relatively diverse, even within a single class of gold de-
(Rose and Burt, 1979). Since the 1980s, many geophysical airborne and posits (e.g., Groves and Foster, 1991).
ground based surveys have been conducted to identify the structures Furthermore, with the decrease in gold deposits readily available at
and environments favorable for gold (e.g., Paterson and Hallof, 1991; the Earth’s near-surface, it has become necessary to prospect and ex-
Doyle, 1986). These surveys mainly employ magnetic, electromagnetic, tract gold from greater depths. Although shallow deposits represent
and electrical methods. more than half of all global gold discoveries, analytical reports show
The identification and recognition of hydrothermally altered zones that the average depth of recent worldwide gold discoveries has in-
are often used as an exploration tool to target gold mineralization, for creased (Schodde, 2019). Considering this situation, the seismic


Corresponding author.
E-mail addresses: a.calazans@postgrad.curtin.edu.au (A.E. Calazans Matos de Souza), stephanie.vialle@curtin.edu.au (S. Vialle).

https://doi.org/10.1016/j.oregeorev.2020.103776
Received 22 April 2020; Received in revised form 5 September 2020; Accepted 11 September 2020
Available online 13 September 2020
0169-1368/ © 2020 Elsevier B.V. All rights reserved.
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

reflection method, which is the only surface geophysical method that hydrothermally altered zones associated with gold deposition. To do
combines the depth of penetration and high resolution, has thus be- this, we use a combination of laboratory characterization and geolo-
come an increasingly used technique in mineral exploration (Malehmir gical information. We examined 21 samples displaying a wide range of
et al., 2012 and references therein). The seismic reflection method has hydrothermal alteration intensities from the Karari gold deposit in
been used for the imaging of structures hosting mineral deposits and Western Australia. Rock physics laboratory measurements were con-
crustal mega-structures that aid in the interpretation of the entire mi- ducted, including (i) ultrasonic P-wave and S-wave velocities (Vp and
neral system, from the source to the migration and trapping of metal- Vs, respectively) and density measurements at ambient conditions and
rich fluids (Goleby et al., 2004). However, important challenges remain under confining pressure, (ii) porosity estimation using measured grain
for this method to become widely accepted in the mining industry and and bulk densities, and (iii) quantification and mapping of mineral
used to its full potential. One challenge is that the geological settings of phases.
these environments are extremely complex, rendering a quantitative The large-scale hydrothermal alteration system at the Karari gold
interpretation of the recorded seismic signatures difficult. In particular, deposit, which displays a characteristic zonal arrangement (including
little is known about the effects of various parameters, such as rock proximal, transitional, and distal zones) (Witt et al., 2009), makes this
texture and rock mineralogy, on the seismic attributes. mine an appropriate site to assess how acoustic properties are affected
The seismic detection of an ore body in the subsurface is governed by hydrothermal alteration. The recent high-grade drilling results from
by two factors: its geometry and its seismic reflectivity (Salisbury and the Karari gold deposit have enhanced the prospectivity of the mine and
Snyder, 2007). The geometry of the body, especially the size, affects the encouraged the extension of drilling coverage at depth to delineate
seismic resolution (i.e., the capacity to discriminate two close features). additional mineralization. To date, the boreholes of the mine have
Owing to the limits of seismic resolution, which is controlled by the targeted gold bodies at 535 m below the surface, but the reserve is
dominant wavelength of the data and depth of the body, the minimum expected to extend up to 835 m in depth (Saracen Website, 2019). As
size of a resolvable body is a few tens of meters (Dentith and Mudge, part of the extension plan for the mine, a detailed 50 km2 3-D seismic
2014; Salisbury and Snyder, 2007). Seismic reflectivity is controlled by survey was deployed in the area, covering the Karari and Dervish mines
the acoustic impedance contrast between ore bodies and the sur- down to a depth of 5 km (Saracen Website, 2019).
rounding country rocks (Telford et al., 1990). Acoustic impedance is the The results of this study may help to interpret existing seismic data
product of the seismic wave velocity and the bulk density of the rock in the area and guide future seismic acquisition, processing, and for-
(Telford et al., 1990), and acoustic impedance contrast is usually low in ward modeling. In summary, the results will assist in understanding
hard-rock environments (Salisbury et al., 2003). This, coupled with the seismic reflections across mineral deposits that are genetically asso-
low signal-to-noise ratio of the seismic data caused by ambient noise, ciated with hydrothermal alteration processes.
ground roll, and airwaves (Wu et al., 2020), makes seismic detection in
this environment challenging. Eaton et al. (2003), Salisbury and Snyder 2. Geologic setting of the Karari gold deposit
(2007), and Malehmir et al. (2012) reported that knowledge of the
acoustic properties is one of several aspects that must be considered The Karari gold deposit is located approximately 110 km northeast
when planning and interpreting seismic data in a hard-rock environ- of Kalgoorlie city, Western Australia, in the Pig Well-Yilgangi sedi-
ment. However, data on the acoustic properties of crystalline rocks mentary basin. Locally, this basin is known as the Carosue basin and
associated with mineral deposits are scarce (Dentith et al., 2020; mainly comprises metamorphosed sandstones that are intruded by
Schetselaar et al., 2019; Chopping, 2008; Souza et al., 2019; Miah et al., lamprophyre and monzonite porphyries (Fig. 1) and cut by north–south
2015; Duff et al., 2012; Malehmir et al., 2012). This scarcity may be faults that are present in all geological sections (Witt & Hammond,
attributed to the difficulty and cost of acquiring geophysical logs in 2008). The mineralization is enveloped by these metasediments and is
narrow boreholes in mineral deposits. The limited knowledge of structurally controlled by faults that coincide with intrusions of mon-
acoustic properties makes seismic interpretation more ambiguous in zonite porphyry (Gray et al., 2005; Witt et al., 2009).
complex geologic environments, such as hard-rock environments. The deposit presents several hydrothermal alteration facies (Fig. 1);
Seismic velocities and density measurements conducted in the labora- however, in terms of exploration, the potassic and sodic zones are the
tory on core samples acquired from boreholes may contribute to re- most relevant altered zones. The other altered zones, which include
ducing this gap in knowledge. chloritic alteration, muscovite alteration, and hematitization, present
In contrast to sedimentary environments in which porosity and little appeal in terms of exploratory data because they occur at a dis-
fluids are the primary control on seismic velocity, in hard-rock en- tance from the mineralized zone, have a limited distribution in the
vironments, the nature of rock-forming minerals is typically a first- mine, and postdate gold deposition.
order control (Dentith and Mudge, 2014). However, some studies have The potassic alteration zone is interpreted as a result of high-tem-
reported that pore shape also strongly affects the seismic velocity of perature and saline magmatic–hydrothermal fluid flow that occurred
low-porosity rocks such as hard rocks (Johansen et al., 2002; Wilkens prior to the gold deposition (Witt et al., 2009). The mineralogy of the
et al., 1991). In terms of seismic response, little importance has been volcaniclastic sediments was overprinted by this hydrothermal activity,
given to the texture of rocks in these environments, mainly because resulting in the potassic alteration assemblage in these rocks, mainly
they are difficult to compact, massive, and should have zero or close to composed of plagioclase–biotite–quartz (Witt et al., 2009). Regional
zero porosity. However, these rocks may develop foliation and sec- shortening of the basin deformed the metasedimentary rocks (Blewett
ondary porosity (e.g., micropores, cracks, and fractures at various et al., 2004), which resulted in a well-developed foliation formed by the
scales) if they are deformed and altered. In addition, some gold mi- preferred orientation of biotite (Witt et al., 2009). The potassic zone is
neralization is hosted in volcaniclastic sedimentary sequences that can extensive and surrounds the mineralized zone (Fig. 1). In this paper, we
exhibit relict porosity and foliation (Witt et al., 2009), and seismic wave denote the potassic altered zone as the “background” rocks because this
velocities are profoundly affected by the porosity and foliation of the represents a regional alteration and is not genetically associated with
rocks. Increasing porosity reduces the seismic velocities (Schon, 2015), the gold deposition.
and foliated rocks tend to be seismically anisotropic, with velocities Gold is associated with the sodic hydrothermal alteration zone and
typically lower when propagating perpendicular to the foliation com- is spatially influenced by monzonite porphyry intrusions (Fig. 1). This
pared with the parallel direction (Kern et al., 2009; Christensen and alteration zone progressively overprints the existing foliation and mi-
Mooney, 1995). neralogy of the potassic assemblage, suggesting that the development of
The main goal of this study is to understand the role of porosity, the sodic zone and the gold deposition both postdate the potassic zone
foliation, and mineralogy on the seismic signature across (Witt & Hammond, 2008). The rocks of the sodic zone are cross-cut by

2
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Fig. 1. Structural map of the Karari pit showing its several alteration facies and its mineralized zones: CL, FWL, HWL, NWL, K1L. Monzonite porphyry and lam-
prophyre intrusions are shown as a reference. Note the spatial association of the sodic alteration (mineralized zone) with the monzonite porphyry and lamprophyre
intrusions. Inset shows the location of the Karari gold deposit (CD) within the Archean Yilgarn Craton of Western Australia. The picture is modified from Witt et al.
(2009).

quartz veins, which, unlike the typical orogenic gold deposits of the transitional zone between background and sodic alteration, and the
Yilgarn Craton (Groves et al., 1998), are barren (Witt et al., 2009). Gold altered mineralized zone (sodic zone), at different stages of hydro-
occurs associated with pyrite and as free gold in the altered groundmass thermal alteration corresponding to moderate intensity, high intensity,
within the sodic zone. Gold lodes are tens to hundreds of meters in and the monzonite porphyry intrusion. The background samples are
strike length and several meters wide (Witt et al., 2009). cylindrical with a diameter of 5 cm, whereas samples from the altered
zone and intrusion zones are half cylinders, cut lengthwise.

3. Sample description and preparation


3.1. Sample description
We selected 21 representative rock samples from a 100 m-long drill
core provided by Saracen Mineral Holding Limited, which is the com- The background samples consist of volcaniclastic sediments that
pany operating the mine. Based on visual inspection as well as the display a relic sedimentary texture. Alternating bands of platy minerals
description and analyses performed by Mihayo (2018) and Witt et al. (biotite) and coarse-grained minerals (quartz/feldspar) at the milli-
(2009), we chose these samples to be representative of the different meter scale define a well-developed foliation (Fig. 2). There is a very
zones of hydrothermal alteration (Fig. 2): the background zone, the small angle between the plane perpendicular to the foliation and the

3
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Fig. 2. Schematic 100 m long drill core showing different zones of hydrothermal alteration. On the right, representative samples from each altered zone. The sodic
zone was qualitatively divided in moderately and highly altered intensity. Note the well-developed foliation in the background sample and the increasing of veins
towards the intrusion.

main axis of the rock specimens (approximately 0–10°), and we assume the main axis of the core.
in this study that the foliation is orthogonal to the main axis of the For the automated mineral identification and mapping using scan-
samples. We selected seven samples to represent this group, labeled T7 ning electron microscopy (SEM), the ends of eight selected samples
I, T7 III, T9 I, T9 II, T9 III, T9 IV, and T9 V. (two background, two from the transitional zone, three from the sodic
The transitional zone between the background and altered zones is zone, and one from the intrusion) were trimmed to obtain a 2 cm-thick
banded, presenting characteristics from both zones, including some slice, which was then cut into a 2.5 cm-long and 1 cm-thick block. Each
pink sodic alteration and dark green bands of potassic alteration. The sample was then polished to achieve a flat and smooth finish using a
banding is controlled by the albitization of the potassic assemblage standard grinder-polisher, and set into a 30 mm diameter epoxy resin
(background) (Witt et al., 2009). Some sedimentary textures are still block. The remaining offcuts of these samples were finely milled for a
preserved. This altered zone is 6 m thick in the drill hole (Fig. 2). We duration of 4 s using a rock milling machine. One portion of the powder
collected two samples from this group, labeled T13 I and T13 IV. was used for grain density measurement, and the other portion was
The intensity of alteration in the sodic zone was qualitatively de- packed into a small circular holder (1 cm diameter and 1 mm thickness)
scribed as moderately and highly altered, based on the proximity to for X-ray diffraction (XRD) analysis. The mounts were randomly
monzonite intrusions and on the number of quartz and calcite veins that packaged using the side-loading method, which consists in loading the
cross-cut the samples. The density of cross-cutting veins increases with powdered samples from one side of the window in the holder, in order
proximity to the intrusion, with up to 3 veins per cm2 (Fig. 2). The veins to minimize the preferred orientation of certain minerals, as a preferred
are randomly oriented, and their sizes range from a few millimeters to orientation can result in inaccurate peak intensities of XRD reflections
2 cm. The samples are texturally massive, lacking the sedimentary (Środoń et al., 2001).
texture, and the pinkish-reddish color is a result of the presence of fine-
grained hematite (‘dusty hematite’) (Witt et al., 2009). We selected six 4. Methods
samples to represent the moderately altered sodic zone (T13 II–III and
T14 I–IV), three samples to represent the highly altered sodic zone (T15 4.1. Mineral identification
I–III), and three samples to represent the monzonite intrusion zone (T18
I–III). The three representative samples of the monzonite intrusion zone Minerals were identified using XRD (Dinnebier and Billinge, 2008;
look visually similar to the samples of the sodic zone, except for the Jenkins and Snyder, 1996) and the TESCAN Integrated Mineral Ana-
large size of albite grains in this zone (see Fig. 4). lyzer (TIMA) (Hrstka et al., 2018), which is a SEM-based automated
mineralogy system. XRD was initially performed to qualitatively iden-
3.2. Preparation tify the mineral phases, which were then input into a databank (library)
of reference spectra that is used in the TIMA system. The minerals were
To measure the ultrasonic wave velocities under ambient condi- then automatically quantified based on this predefined library using
tions, the rock samples were cut into pieces 5.6 to 7.3 cm in length. For TIMA. Compared to XRD, TIMA provides spatial (2-D) information on
measurements under confining pressure, we prepared two cylindrical mineralogy, which allows quantitative parameters such as the grain size
background samples of 7.6 cm in length and 3.8 cm in diameter to fit distribution and textural relationships to be extracted. Both XRD and
the pressure chamber. Because the cores of the hydrothermally altered TIMA were performed at the John de Laeter Research Centre at Curtin
samples had been previously cut lengthwise, the dimensions of these University, Western Australia.
half cylinders were too small to extract cylindrical sub-samples with an
appropriate length and diameter to perform measurements under con- 4.1.1. XRD analysis
fining pressure (requiring a length greater than 7.6 cm and a diameter The samples were tested using the powder diffractometer D8
equal to 3.8 cm). The end surfaces were cut parallel at a precision Advance (Bruker AXS, Germany), with a copper K alpha radiation
of ± 0.1 cm to ensure a good propagation of the acoustic waves along source (at 40 kV and 40 mA) and a LynxEye detector. To construct the

4
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Fig. 3. Porosity estimation from image segmentation (sample T9III). a) BSE image highlighting the pore space in blue and biotite in a lighther grey. b) Total porosity
with BSE intensity ranging between 0 and 38. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this
article.)

diffractometer, the X-ray source was fixed, and the sample holder ro- 4.2. Bulk and grain density measurements and porosity computations
tated around angle θ, while the X-ray detector, which was attached onto
a geared arm, turned around angle 2θ. X-rays were detected and re- Because the rock samples used to obtain the bulk density had an
corded as 2θ values. The X-ray peaks were recorded every time XRD irregular shape, we could not use the caliper method as in Souza et al.
took place at a particular angle θ. Measurements were performed in the (2019), and instead opted for the water displacement method, em-
2θ range of 7.5–90°, with a spatial step of 0.015°, a time step of 0.7 s, ploying Archimedes’ principle (Abzalov, 2013). The samples were oven-
and an overall measurement time of 60 min per sample. dried at 70 °C for 24 h and their mass was then measured using a high-
Mineral phases were identified by comparing the position and in- precision balance in air and in water. To measure the mass in water, we
tensity of X-ray peaks obtained from the samples with a databank of placed the sample in a container hanging from a balance and immersed
mineral X-ray data known as a powder diffraction file. We used the it in water (as described in Abzalov, 2013). Note that when measuring
search/match algorithm DIFFRAC.EVA 3.2 (Bruker AXS, Germany) to the mass in water, it is assumed that no water enters any pores or
search the powder diffraction file (PDF4 + 2019 edition). fractures, which is ensured here by i) the fact that the rock samples
have a very low porosity (0%–2%) associated with very small pores or
4.1.2. TIMA analysis cracks (see Section 6.1.2), and ii) the very short period of time during
The TIMA GM model (Tescan, Czech Republic) was used to auto- which the sample is actually immersed in water (around 30 s).
matically identify and quantify the minerals present in the samples. The grain density was obtained by employing Boyle’s volume–-
TIMA consists of a scanning electron microscope equipped with four pressure relationship, using an automated helium gas pycnometer, the
fully integrated energy dispersive spectroscopy detectors (Hrstka et al., Accupyc II Pycnometer (manufactured by Micromeritics, 1995).
2018). Data processing and mineral classification were carried out Measurements were performed at CSIRO, Western Australia. The
using the software package TIMA 1.5. The data were acquired using the sample powders were also dried, similarly to the bulk density mea-
dot mapping acquisition mode at a spot size of 69.97 nm, at 3 μm re- surements. The total porosity, in percent, was then computed from the
solution, with a beam current of 6.91 nA in a high vacuum mode and an bulk density and grain density measurements (Manger, 1963).
accelerating voltage of 25 kV. We also estimated the porosity based on the BSE images for com-
The mineral classification combines backscattered electron (BSE) parison with the porosity obtained from the bulk density–grain density
imaging with X-ray mapping from the electron beam interaction relation. We defined the type of pores (soft or stiff) from these BSE
(Hrstka et al., 2018). Initially, as the BSE intensity is proportional to the images by image analysis, as seismic velocities are more sensitive to soft
average atomic number of the mineral, the boundaries of the grains are pores, such as crack-like pores, than to rounded pores (stiff). The TIMA
easily defined by the gradient in BSE values (Sylvester, 2012). The X- instrument’s software provides the volume fraction of each phase as an
ray energy inside each grain is then summed to enhance the quality of output, including that of the pores (i.e., the total porosity); however, we
the spectra (Hrstka et al., 2018). The X-rays generated are characteristic performed image analysis on these BSE images using the software
of each element present in the sample. AVIZO (http://www.fei.com/software/avizo3d/). Using this approach,
Minerals were automatically identified by comparing the BSE in- we were able to separate the cracks from the stiff pores and quantify the
tensity and the combined X-ray spectra of each grain with a predefined sensitivity of the computed porosity to the chosen grayscale threshold
library of reference spectra. A basic library including most common that was used to separate the pores (in black) from the minerals (in
minerals provided with the software was complemented with spectra shades of gray). For each BSE image, we first applied the median filter
for minerals previously identified through the XRD analysis. As a result, to remove noise then segmented the pore space by manual thresh-
the software created 2-D mineral maps from the rock specimens, and olding. Based on our visual inspection and the porosity results deduced
the volumetric fraction of each mineral was inferred from its surface from the experimental measurements, we determined that values of BSE
area. intensity ranging from 0 to 38 are most representative of the pore space.

5
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

We computed the sample porosity using this range for all the BSE associated with grain sizes smaller than 4 µm. We note that the un-
images, and the results are reported in Table 2. To define the pore type, classified values decrease in the direction of the intrusion (Table 1) due
we calculated the width and length of each selected pore and assumed to the porphyritic texture, with large grains of albite (up to 5 mm wide)
that pores with a width/length ratio lower than 0.5 are soft pores and embedded in a fine-grained matrix (Fig. 4).
all other pores are considered to be stiff pores. This value was defined
based on the visual identification of cracks (Fig. 3). 5.1.1. Background rocks
The major diffraction peaks observed for the background samples
4.3. Laboratory measurements of ultrasonic wave velocities are related to biotite and albite, which match the most abundant mi-
nerals identified in the TIMA data (Table 1). In the XRD data, biotite
Bench-top ultrasonic velocity measurements of P- and S-waves were and phlogopite display similar diffraction patterns, but the TIMA clas-
performed on all samples at ambient pressure and temperature condi- sification indicates that biotite is the most dominant phyllosilicate mi-
tions using the pulse transmission method (Birch, 1960), which in- neral, in agreement with Witt et al. (2009). The diffraction peak of
volves measuring the first arrival travel time of the ultrasonic wave chlorite (a clay mineral) was present at around a 2θ angle of 12°.
propagating through the sample. Souza et al. (2019) describe details of However, TIMA could not identify this mineral due to its small grain
ultrasonic wave velocity measurements conducted on samples from the size. A portion of the unclassified values of the background samples in
Karari gold deposit. The samples were placed between a single pair of the TIMA analysis (Table 1) can therefore be attributed to chlorite. The
piezoelectric transducers that transmitted and received P- and S-waves main constituents of the two background samples are albite, anortho-
simultaneously at a central frequency of 1 MHz (Souza et al., 2019). clase, biotite, and quartz (Table 1). Anorthoclase is the primary alkali
The travel time of the P-wave was manually picked as the first onset of feldspar, and albite represents more than 90% of plagioclase feldspar
energy, whereas the S-wave travel time was manually picked as the abundance. The foliated texture observed in the samples is attributed to
earliest arrival of high amplitude energy (Souza et al., 2019) (see the abundance of biotite, at 15%–22% in volume. The grain size of
Supplementary table). We placed the transducers face-to-face and ac- biotite is variable, but some grains measure up to 2 mm in size (brown
quired the system delay time by measuring the transit time between the grains in Fig. 4). Although of low abundance, calcite (1%–6% in vo-
transducers, then subtracted the system delay time from the total travel lume) is the unique carbonate mineral observed in these samples.
time. We measured the length and diameter of the sample with a high- Variable amounts of quartz, ranging from 9% to 22% in volume, occur
precision caliper, then computed P- and S-wave velocities, Vp and Vs, in this zone and the pyrite abundance is less than 0.2% of the rock
respectively, by dividing the sample length by the calculated travel volume.
times (Souza et al., 2019). Velocities were measured in three directions
for background samples (cylindrical), that is perpendicular to the fo- 5.1.2. Transitional zone
liation (i.e., along the main axis of the samples) and in two directions As in the background samples, biotite, quartz, and albite are the
parallel to the foliation (i.e., perpendicular to the main axis of the dominant minerals in the transitional zone, with albite representing
samples). For transitional and altered mineralized samples, velocities almost 41% of the volume in one sample. Compared with the back-
were measured in only two directions, along the main axis of the ground zone, the TIMA classification (Table 1) indicates a progressive
samples and perpendicular to the main axis, because the samples were decrease in biotite, down to less than 10% of the rock volume, and
half cylinders. In the following sections, the axial measurements are calcite is replaced by dolomite.
referred to as Vp0 and Vs0, and the radial measurements as Vp90 and
Vs90 (taken as an average of the two radial measurements in the case of 5.1.3. Sodic zone and monzonite porphyry intrusion zone
background samples). The error in velocity measurements is around 1% The sodic alteration zone and monzonite porphyry intrusion zone
for P-waves and around 2%–4% for S-waves. This error is attributed to are dominated by sodic plagioclase, with albite as the most abundant
ambiguities in picking the arrival times and the uncertainty in the mineral. In some samples, albite represents more than 40% of the rock
sample length and diameter measurements (approximately 0.1 mm). volume. The amount of alkali feldspar increases in the direction of the
We also selected two background samples to investigate the pres- sodic alteration zone due to the increase in orthoclase abundance (up to
sure dependence of the P-wave velocity (we do not address the pressure 27.2% in volume). This zone is intensely cut by quartz–dolomite veins
dependency of the S-waves as their identification in these samples was and pyrite-rich veinlets (Fig. 4). Dolomite is the only carbonate present
challenging). As all altered samples consisted of half cylinders, we were in this zone, making up to approximately 8% of the rock volume. Minor
unable to measure their velocity under a confining pressure. The two pyrite is present and represents up to 2% of the rock volume, although
background samples were inserted into an insulating rubber jacket and Witt et al. (2009) reported that this zone hosts up to 10% pyrite. With
placed inside a pressure vessel, which was sealed and filled with oil. the exception of one sample, the biotite volume is less than 1%, which
The specimens were monitored by three pairs of transducers (sour- explains the absence of foliation in this altered zone. The X-ray data
ce–receiver) to measure the P-wave velocity at a frequency of 1 MHz confirm the disappearance of biotite.
along three directions of wave propagation, as measured at ambient
conditions. Pore pressure was maintained at zero and the sample was 5.2. Bulk and grain density measurements and porosity computations
pressurized in increments of 2.5 MPa, up to 65 MPa.
Bulk and grain density measurements are reported along with por-
5. Results osity in Table 2. The image segmentation and the experimental mea-
surements based on the grain–bulk density relation both show that the
5.1. Mineral identification total porosity of the background samples is higher than that of the al-
tered samples. The total porosity ranges from 1.3% to 2.8% for the
Table 1 shows the modal mineralogy of the samples from different background samples, and is lower than 0.5% for the transitional, sodic,
alteration zones in the deposit characterized by TIMA. Some grains and intrusion zones, with the exception of one sample.
could not be classified (unclassified values), probably because of i) the The average bulk density is 2.72 g/cm3 for background samples,
quality of the polishing in some parts of the specimen, ii) the presence 2.76 g/cm3 in the transitional zone, 2.73 g/cm3 in the moderately al-
of minerals that were not identified during the XRD analysis and were tered zone, 2.70 g/cm3 in the highly altered zone, and 2.68 g/cm3 in
thus not considered in the input library, and/or iii) spectral inter- the monzonite intrusion zone. These values are in agreement with the
ferences caused by the small size of some grains that resulted in several felsic composition, which ranges from 2.5 to 2.8 g/cm3 (Schon, 2015).
overlapping spectra. More than half of the unclassified minerals are The grain density tends to decrease slightly from the background zone

6
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Table 1
Summary of volumetric fractions of mineral (in %) using TIMA analysis.
Group of Minerals Mineral Background Transitional zone Sodic zone Monzonite Intrusion

Sample T7I Sample T9III Sample T13I Sample T13IV Sample T13II Sample T14IV Sample T15II Sample T18III

Plagioclase feldspar Albite 29.8 26.4 40.8 28.4 41.9 27.5 27.7 45.0
Anorthite 0.7 1.5 1.2 1.8 3.3 4.3 3.2 3.0
Oligoclase 0.6 0.9 0.7 0.8 0.7 0.5 0.6 0.5
Total 31.0 28.7 42.7 31.0 45.8 32.3 31.4 48.5
Alkali feldspar Orthoclase 0.9 3.5 10.9 3.1 12.4 27.2 25.8 17.1
Anorthoclase 10.3 12.9 6.8 17.0 8.9 3.0 2.6 3.0
Total 11.2 16.4 17.7 20.1 21.2 30.2 28.4 20.1
Quartz 22.1 9.2 15.3 14.5 9.7 9.1 22.0 12.2
Calcite 1.0 6.1 0.0 0.0 0.0 0.3 0.0 0.1
Dolomite 0.0 0.4 4.0 7.3 8.2 7.6 7.1 6.7
Biotite 14.8 22.5 4.8 9.9 0.9 3.5 0.0 0.1
Pyrite 0.2 0.0 0.6 0.1 0.1 2.6 1.8 0.8
Others 2.6 2.9 2.8 1.6 2.0 1.5 0.8 1.3
Unclassified 15.9 12.8 11.7 15.0 11.8 12.5 8.1 9.8
Holes 1.0 0.9 0.5 0.5 0.2 0.2 0.3 0.2
Total 100 100 100 100 100 100 100 100

Fig. 4. Example of mineral identification and mapping using TIMA for 4 different stages of alteration. Only the more abundant minerals have been displayed (more
than 2% in vol.).

7
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Table 2
Summary of measured bulk density and grain density (in g/cm3) along with porosity.
Zone-Intensity of Samples Bulk Grain Total porosity from density Total porosity from BSE Stiff porosity from BSE Soft porosity from BSE
alteration density density relation image image image

Background T7I 2.69 2.746 2.1 ± 0.2% 1.6 ± 0.6% 1.4 ± 0.5% 0.2 ± 0.1%
Background T9III 2.73 2.800 2.3 ± 0.2% 2.0 ± 0.4% 1.6 ± 0.3% 0.4 ± 0.1%
Background T9IV 2.75 2.786 1.3 ± 0.2% – – –
Background T9V 2.78 2.857 2.8 ± 0.2% – – –
Transitional T13I 2.74 2.738 0.1 ± 0.2% 0.5 ± 0.2% 0.4 ± 0.2% 0.1 ± 0.2%
Transitional T13IV 2.79 2.787 0.0 ± 0.2% 0.5 ± 0.2% 0.5 ± 0.2% 0 ± 0.2%
Sodic-Moderate T13II 2.73 2.726 0.0 ± 0.2% 0.3 ± 0.1% 0.3 ± 0.1% 0 ± 0.1%
Sodic-Moderate T14IV 2.72 2.710 0.0 ± 0.3% 0.4 ± 0.1% 0.4 ± 0.1% 0 ± 0.1%
Sodic-High T15I 2.68 2.667 0.0 ± 0.3% – – –
Sodic-High T15II 2.74 2.767 0.9 ± 0.2% 0.4 ± 0.2% 0.3 ± 0.1% 0.1 ± 0.1%
Sodic-High T15III 2.68 2.686 0.3 ± 0.2% – – –
Monzonite intrusion T18III 2.70 2.705 0.2 ± 0.2% 0.3 ± 0.1% 0.3 ± 0.1% 0 ± 0.1%

to the monzonite intrusion zone. The background sample T9V exhibited background samples are consistently lower than in the other altered
the highest grain density, at 2.85 g/cm3. samples. The most significant difference in Vp and Vs values between
the background and altered samples occurs in the direction perpendi-
cular to the foliation, with an average difference of 1990 m/s and
5.3. Ultrasonic velocity measurements
940 m/s in Vp and Vs, respectively. When measured parallel to the
foliation, the difference between the background and the altered sam-
5.3.1. Ambient conditions
ples is on average 670 m/s and 330 m/s for Vp and Vs, respectively.
Fig. 5 shows the values of Vp and Vs under ambient conditions,
Although background and altered samples may be easily dis-
measured parallel (open dots) and perpendicular (black dots) to the
criminated using the seismic velocity values, it remains challenging to
foliation on the 21 samples from the different alteration intensity zones.
distinguish between samples characterized by distinct intensities of
As the values of velocities are close (with a difference of less than
alteration due to their similar velocity values. For example, the average
100 m/s) along the foliation of the samples in both parallel directions,
Vp variation between the transitional zone and the monzonite intrusion
we considered the average value between these directions (open dots).
zone, considering both directions of acquisition, is 180 m/s. However, it
It was challenging to pick the first S-wave arrival time in the back-
is interesting to highlight that the highest values of Vs are observed
ground samples when measuring perpendicular to the foliation due to a
when the number of quartz veins increases, as in the highly altered
low signal-to-noise ratio and the overlap of reflected P-waves. It was
sodic zone (Fig. 5b).
therefore only possible to measure Vs in this direction in three samples
from this zone (Fig. 5b). In contrast, a rapid shear wave signal was
easily identified in all the altered rocks in both acquisition directions. 5.3.2. Pressure dependence
For all background samples, the velocities show a significant dif- Fig. 6 shows the Vp values for two background samples (T9-IV in
ference depending on the direction of the measurements, with slower white and T9-V in black) measured along three directions of propaga-
velocities recorded in the direction perpendicular to the foliation. In tion as a function of confining pressure. At low pressure, the slow axis
these samples, the ε parameter, corresponding to one of Thomsen’s of wave propagation is perpendicular to the foliation plane in both
anisotropic parameters related to the anisotropy of P-waves (Thomsen, samples (circle symbols). The velocity along this direction of propaga-
1986), varies from 10% to 22%. In contrast, the Vp and Vs values tion increases rapidly with increasing pressure, which is interpreted as
measured in both directions for the samples from the transitional, sodic, resulting from the closure of microcracks. For differential pressures
and monzonite intrusion zones are close and display low values of an- greater than ~10 MPa for sample T9-V, and greater than ~30 MPa for
isotropy compared to the background samples (Souza et al., 2019). sample T9-IV, velocities measured perpendicular and parallel (square
Regardless of the direction of acquisition, the Vp and Vs values for the and triangle symbols) to the foliation become very close in magnitude.

Fig. 5. Acoustic velocities measured in the axial (black dots) and radial (open dots) directions of the cores from different altered zones, at ambient conditions. In the
case of the background samples, the black dots and open dots coincide with the directions perpendicular and parallel to the foliation, respectively. a) P-wave velocity
and b) S-wave velocity. The error in velocity measurement is around 1% to P-wave and approximately 2–4% to S-wave.

8
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

depending on the orientation of the foliation with respect to the di-


rection of wave propagation. This corresponds to a value of ε of ap-
proximately 10% at 200 m and 3% at 800 m.
We could not evaluate the pressure dependence of the acoustic ve-
locities for altered samples as the cores were half cylinders, and thus too
small to create suitable cylindrical sub-samples. However, as these
samples do not exhibit a foliated texture, we would expect a smaller
pressure sensitivity than for the background samples. Based on the
pressure dependence obtained for the velocities measured parallel to
the foliation for the background samples, we hypothesize that P-wave
velocities will increase by a maximum of 2% at 200 m and 5% at 800 m
for the altered samples. We further investigate this hypothesis using
pore space analysis in Section 6.1.2.

6.1.2. Porosity
The porosity values computed using the two different approaches
(density measurements and SEM-BSE image analysis) are in good
Fig. 6. P-wave velocity of two background samples along three directions of agreement (Table 2). The small difference in porosity values obtained
propagation as a function of confining pressure. The black dashed polygon in- using these two techniques can be attributed to the heterogeneity of the
dicates the range of expected pressure in the mine. samples, the use of 2-D data for image analysis versus 3-D data for
experimental measurements, and the choice of the grayscale thresh-
The difference in velocity gradient between the two samples could be olding value to separate pores from minerals. This threshold was mainly
associated with different aspect ratios of the cracks (i.e., the ratio be- based on a visual inspection of the output result. For example, if a
tween the width and the length of the pores). The in situ principal stress threshold of 40 was chosen instead of 38, the porosity estimated for
orientation and magnitude were estimated by Villaescusa and Hogan sample T9-III would be 2.8% instead of 2.0%, demonstrating that the
(2020) using the Acoustic Emission technique on oriented cores ex- obtained value is relatively sensitive to the chosen threshold value.
tracted from the Karari mine. In the following discussion, we use the However, a simple visual inspection of the resulting pore space showed
average of the three principal stress components, which gives a stress of that a portion of the solid phase was identified as pore space for a
9 MPa at 200 m and 30 MPa at 800 m. threshold value of 40. We estimated the error in the porosity compu-
tation using image analysis as resulting from an uncertainty of ± 1 in
grayscale intensity value.
6. Discussion
In order to assess how porosity values vary with pressure, we first
performed tests using the same pycnometer as that employed for grain
The results of this study demonstrate that background and altered
densities, but with cylindrical sub-cores and at increasing values of
samples present mineralogical, textural, and acoustic property differ-
confining pressure. However, at a low confining pressure (2 MPa), only
ences. To help us understand and quantify the effect of various para-
0.1% of the pore volume of the tested samples could be filled with
meters (i. e., confining pressure, mineralogy, porosity, and foliation) on
helium gas, indicating that most of the porosity was unconnected and
the magnitude of the acoustic impedance contrast, and consequently,
thus inaccessible using this technique. We then performed image ana-
on the seismic detectability between background and altered (miner-
lysis on the pore space in the BSE images, and classified pores as soft or
alized) rocks, we analyzed our data by correlating Vp with geological
stiff pores based on their width-to-length ratios. The assumption is that
parameters. We first discuss how the confining pressure affects Vp,
stiff pores will not close under in situ stresses, but soft pores will par-
porosity, and the acoustic impedance of the studied samples. We then
tially or totally close depending on the magnitude of in situ stresses and
use these data as inputs in a rock physics model to assess the effects of
the orientation of the cracks with respect to these stresses. The obtained
the mineralogy on acoustic properties.
values for the soft and stiff porosities are reported in Table 2. For
background samples, soft porosity was identified in all analyzed sam-
6.1. Pressure sensitivity ples and varied between 0.2% and 0.4% (representing up to 20% of the
pore space). For most altered samples, no soft porosity was detected at
6.1.1. P-wave velocity the resolution of the images (3 μm), and represented up to 0.1% only for
Laboratory measurements conducted on background samples at two samples. From these results, we can infer that (i) porosity values
ambient conditions showed a significant velocity anisotropy. As ex- under the in situ stress conditions existing at the Karari mine are ex-
pected, the lowest velocities were measured perpendicular to the fo- pected to be similar to those obtained under ambient conditions for the
liation in these samples (Fig. 5). The P-wave velocity anisotropy at altered samples and at most 25% smaller for the background samples,
ambient conditions is high, with ε = 22%, but decreases significantly and (ii) the pressure sensitivity of acoustic velocities in altered samples
for pressures greater than 20 MPa with ε approaching 0% for pressures is expected to be smaller than that observed for background samples
greater than 45 MPa (Fig. 6). Perpendicular to the foliation, velocities due to a lack of or a very small proportion of soft porosity (cracks).
increase by 24% from 0 MPa to 45 MPa, whereas velocities increase by Thus, the hypothesis that velocities will increase by at most 5% under
only approximately 10% parallel to the foliation for sample T9 IV. This the maximum in situ pressure is legitimate.
suggests that the microcracks are preferentially distributed parallel to Fig. 7 shows the evolution of porosity, Vp 90, Vp 0, and mineralogy
the foliation, in agreement with the morphology of platy phyllosilicate as a function of increasing hydrothermal alteration. For porosity, Vp 90
minerals such as biotite, which tends to display cracks parallel to the and Vp 0, the range of expected values under in situ stress conditions
rock fabric (McCuaig and Kerrich, 1994). Cracks can be easily identified are displayed in Fig. 7a, as described above. We observed a good cor-
and associated with biotite grains in the BSE images (Fig. 3). In the relation between P-wave velocities and porosity and noted that a dif-
context of the Karari mine, where effective stresses are expected to be ference in porosity of approximately 2% between the background and
between 9 MPa and 30 MPa, the P-wave velocities for background altered samples generates a difference in Vp 90 between 400 m/s and
samples are expected to be between 4300 m/s and 5200 m/s at a depth 1500 m/s and a difference in Vp 0 between 900 m/s and 2600 m/s. This
of 200 m and between 5100 m/s and 5400 m/s at a depth of 800 m, indicates that the velocity values are very sensitive to porosity.

9
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Fig. 7. The porosity-Vp relation for 12 samples (a) and the mineralogy-Vp relation for 8 samples (b) across the zones of alteration.

Fig. 8. P-wave velocity versus density acquired along, Vp0, (a) and transverse, Vp90, (b) to the main axis of the samples at ambient conditions. The colors represent
different alteration zones and symbols represent different pressure conditions. Dashed lines represent constant acoustic impedance, and the perpendicular separation
between them is the minimum acoustic impedance contrast required to generate a reflection coefficient (R) of 0.06.

6.1.3. Seismic reflectivity reflection may not be detected.


Fig. 8 displays Vp 0 (a) and Vp 90 (b) versus bulk density at ambient Considering this scenario, the seismic reflectivity between the
conditions, along with their estimated increase at the in situ stress background and altered samples is expected to decrease with increasing
conditions. The perpendicular distance between the constant acoustic depth (increasing confining pressure). Depending on the orientation of
impedance contours (black dashed lines) represents the minimum re- the foliation relative to the seismic wave propagation direction, the
flection coefficient required for a planar reflector to be detected by seismic reflectivity is expected to be between 0.07 and 0.13 at a depth
seismic reflection data, corresponding to 0.06 (Salisbury et al., 1996). 200 m and between 0.05 and 0.09 at a depth of 800 m.
To estimate the reflection coefficient, R , we used the Zoeppritz equation
for the case of normal incidence, 6.2. Effect of mineralogy on acoustic impedance contrast between
I I1 background and altered samples
R= 2 ,
I1 + I2 (1)
Based on the mineralogy data (Table 1), no clear trends in the
where I1 and I2 are the acoustic impedances of the upper and lower amount of quartz and feldspar (in % volume) were observed from the
media, respectively. Considering the presence of a seismic reflector background samples to the monzonite intrusion zone. However, we
with a favorable geometry and size, for the Vp0 case, the background/ observed that calcite was replaced by dolomite and the amount of
hydrothermal alteration interface is easily detected by the seismic re- biotite decreased significantly with alteration, from 22% to 0% in vo-
flection method for pressures lower than 10 MPa. However, no contrast lume (Fig. 7b). Dolomite exhibits a P-wave velocity (7.4 km/s) higher
was detected among the altered samples. We note that the difference in than calcite (6.5 km/s) and biotite (6.1 km/s) (Mavko et al., 2011), and
acoustic impedance is entirely attributed to the difference in velocity, its density (2.87 g/cm3) is lower than that of biotite (3.05 g/cm3) but
with no contribution from density. This is still true for the Vp90 case, higher than that of calcite (2.71 g/cm3). As acoustic impedance is the
but the seismic reflectivity is lower, close to the minimum required for product of these two quantities, the resulting effect of this mineralogy
seismic detection. Bearing in mind that the signal-to-noise ratio of the change on seismic reflectivity is not straightforward, and likely to be
seismic data is commonly low in hard-rock environments, this weak small. To quantify the effect of the mineral replacement only on the

10
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Table 3
Parameters used for the rock physics modeling. Density, bulk modulus (K), and shear modulus (µ) are in g/cm3, GPa, and GPa, respectively. Except for albite that was
taken from Bass (1995), the values of elastic moduli for each mineral were taken from the data compiled by Mavko et al. (2011). The ranges of the volume fraction of
each mineral in each altered zone were based on TIMA analysis and Witt et al. (2009).
Elastic moduli from literature Range of the volume fraction of each mineral in each zone

Mineral Density K µ Background zone Transitional zone Sodic zone

Quartz 2.65 36.6 45.0 10–40% 15–30% 10–32%


Calcite 2.71 75.0 31.0 0–12% 0–7% 0%
Dolomite 2.87 94.9 45.7 0–1% 2–12% 5–14%
Biotite 3.05 55.2 42.4 13–45% 5–15% 0–4%
Albite 2.61 56.9 28.6 15–45% 25–50% 31–50%
Pyrite 4.93 147.4 132.5 0–1% 0–1% 0–4%
Magnetite 5.20 161.4 91.4 0% 0–4% 0–1%

seismic reflectivity, we assumed zero porosity and used the Voig- clustering of the data based on their intensity of alteration (Fig. 10).
t–Reuss–Hill average rock physics model (Mavko et al., 2011). As input However, two samples from the highly altered zone (T15-I and T15-III)
in this model, we use the volume fraction of the most abundant mineral displayed a much lower value of Vp/Vs, regardless of the acquisition
phases only, and their elastic moduli and density taken from the lit- direction, and also visually present a higher number of quartz veins. As
erature (listed in Table 3). To better represent the acoustic properties in most of the typical orogenic gold deposits of the Yilgarn Craton in
each alteration zone, we first fixed a range of volume fractions for each Australia are associated with quartz veins (Groves et al., 1998), a cross-
mineral according to the expected values for each zone, based on the plot of Vp as a function of Vp/Vs could be used to identify quartz vein-
results obtained from TIMA analysis (see mineral identification section) rich zones.
along with reported values from Witt et al. (2009) (Table 3).
For the Voigt-Reuss-Hill average, we first estimated the effective
bulk modulus of a mixture of minerals (solid) using the expressions for 7. Conclusions
the Voigt upper bound and the Reuss lower bound
N This study focused on understanding the effects of geological
MVoigt = fi Mi , parameters such as mineralogy, porosity, and foliation on seismic re-
i=1 (2) flectivity in hydrothermal alteration zones associated with the Karari
gold deposit. To do so, we measured the bulk density and P- and S-wave
N
1 fi velocities at ambient conditions on 21 whole core samples representing
=
MReuss i=1
Mi (3) different intensities of hydrothermal alteration, and under confining
pressure for a subset of these. The velocity measurements were per-
respectively, where N, fi, and Mi represent the number of components, formed in the axial and radial directions of the cores. In the case of
the fraction of ith mineral constituent, and bulk (K) or shear modulus (µ) background samples, measurements along the main axis of the cores
of the ith mineral constituent, respectively. We then calculated the ar- were made perpendicular to the foliation and measurements transverse
ithmetic average of both. Once the elastic moduli were computed, we to the main axis were made parallel to the foliation. In addition, we
then calculated the density (ρ) and Vp using the following expressions, selected a subset of eight samples for careful characterization of the
respectively, mineralogy and porosity. Regardless of the direction of acquisition, Vp
N and Vs in the background samples were consistently slower than in the
= fi i , altered samples. Our data indicate that porosity is the main causative
i=1 (4) factor for lower velocity values in the background samples, and con-
sequently, for the lower acoustic impedance values. The difference in
K+

acoustic impedance is entirely attributed to the difference in velocity
3
vp = . with no contribution from density. The small contrast in porosity (ap-
(5)
proximately 2%) between the background and altered samples dom-
The results of these models show that Vp tends to increase and inates the velocity response, especially in the acquisition direction
density tends to decrease with increasing alteration (“x” symbols in perpendicular to the foliation. Although the background/alteration in-
Fig. 9). The lack of acoustic contrast between the different zones is a terface displays seismic reflectivity values that are sufficient to be de-
result of the compensation of the increase in Vp with alteration by the tected by seismic methods at pressures below 10 MPa, the seismic re-
reduction in density. Mineralogy is therefore not the primary geological flectivity of this interface tends to decrease rapidly with increasing
parameter controlling the seismic reflectivity of the background/altered pressure due to the closure of microcracks. At pressures between
rock interface. Hence, any acoustic contrast must be the result of 10 MPa and 30 MPa, seismic reflectivity is around the minimum value
parameters relating to the pore space (such as porosity, and cracks not required for a planar reflector to be detected using the seismic reflec-
closed or partially closed at in situ stresses, etc.) that will lower the tion method and is seismically undetectably at pressures above 30 MPa.
values of acoustic velocities. To investigate the effects of mineralogical replacement caused by
As previously mentioned, discriminating between samples with hydrothermal alteration on seismic reflectivity, we created hypothetical
different alteration intensities (transitional, moderate, high, and intru- zero porosity rocks with a range of mineral volume fractions for each
sion) using only Vp or Vs is challenging (Fig. 5). In the oil industry, the dominant mineral according to the expected values for each zone of
combination of seismic velocities Vp and Vs is usually used to dis- alteration. We then used these data as input in the Voigt-Reuss-Hill
criminate mineralogy and fluids (Avseth et al., 2010). Among the mi- average rock physics model. These results showed that the miner-
nerals in this study, quartz has one of the lowest Vp/Vs ratios because alogical replacement does not affect the seismic reflectivity of the de-
its shear modulus is higher than its bulk modulus (Table 3). As there is posit because the increase in Vp with increasing alteration is compen-
no clear trend in quartz content between the different alteration zones sated by a reduction in density. Although the mineralogy does not
in this study, a cross-plot of Vp as a function of Vp/Vs does not result in directly affect the seismic reflectivity, the porosity observed in the

11
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Fig. 9. The effects of mineralogical replacement (x) and porosity on Vp and density across the zones of alteration. The colors represent different alteration zones and
symbols represent different pressure conditions. To simulate the mineralogical replacement with alteration, we used the Voigt-Reuss-Hill average. Dashed lines
represent constant acoustic impedance.

Fig. 10. Vp as a function of Vp/Vs at ambient conditions.

background samples is in part attributed to the presence of biotite in Declaration of Competing Interest
these samples.
This study enhances our understanding of the acoustic contrast The authors declare that they have no known competing financial
between the gold-mineralized rocks and their host rocks, which serves interests or personal relationships that could have appeared to influ-
as a basis for seismic interpretation of the existing seismic data across ence the work reported in this paper.
the deposit, as well as for delineating additional mineralization in the
Karari gold deposit and in mineral deposits with similar geological
characteristics.

12
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Acknowledgments Goleby, B.R., Blewett, R.S., Korsch, R.J., Champion, D.C., Cassidy, K.F., Jones, L.E.A.,
Groenewald, P.B., Henson, P., 2004. Deep seismic reflection profiling in the Archaean
northeastern Yilgarn Craton, Western Australia: implications for crustal architecture
Andre Souza is grateful to Curtin University (through Curtin and mineral potential. Tectonophysics 388 (1–4), 119–133. https://doi.org/10.1016/
Strategic Stipend Scholarship) and CSIRO (through the Deep Earth j.tecto.2004.04.032.
Imaging Program) for the financial support and the physical facilities Gray, D.J., Robertson, I.D.M., Cornelius, M., Sergeev, N.B., Porto, C.G., 2005. Karari and
Whirling Dervish gold deposits, Western Australia. Cooperative Research Centre for
for this research. We thank Saracen Mineral Holding Limited for pro- Landscape Environments and Mineral Exploration (CRC LEME). CSIRO Exploration
viding the drill core. We thank Lionel Esteban from CSIRO for assisting and Mining, Bentley, Western Australia.
in the grain density measurements. We are grateful to Mustafa Sari, Joel Groves, D.I, Foster, R.P, 1991. Archaean lode gold deposits. Gold metallogeny and ex-
ploration. Springer, Boston, Ma, pp. 63–103. https://doi.org/10.1007/978-1-4613-
Sarout, Ludwig Monmusson, and Shane Kager from CSIRO's 0497-5_3.
Geomechanics and Geophysics Laboratory for assisting in the velocity Groves, D., Goldfarb, R.J., Gebre-Mariam, M., Hagemann, S.G., Robert, F., 1998. Orogenic
measurements under confining pressure. Dr. Mehrooz Aspandiar, gold deposits: a proposed classification in the context of their crustal distribution, and
relationship to other gold deposit types. Ore Geol. Rev. 13, 7–27. https://doi.org/10.
Veronica Avery, and Andrew Wieczorek, from Applied Geology at
1016/S0169-1368(97)00012-7.
Curtin University, are thanked for their useful help during the selection Hrstka, T., Gottlieb, P., Skala, R., Breiter, K., Motl, D., 2018. Automated mineralogy and
and preparation of the samples. Part of this research was undertaken petrology – applications of TESCAN Integrated Mineral Analyzer (TIMA). J. Geosci.
using the XRD and TIMA instrumentation (ARC LE LE0775551 and 63, 47–63. https://doi.org/10.3190/jgeosci.250.
Jenkins, R., Snyder, R.L., 1996. Introduction to X-Ray powder diffractometry. In:
LE140100150] at the John de Laeter Centre, Curtin University. We are Winefordner, J.D. (Ed.), Chemical Analysis. John Wiley & Sons, New York.
grateful to Geological Society of Australia for permission to reproduce Johansen, T.A., Drottning, A., Lecomte, I., Gjøystdal, H., 2002. An approach to combined
copyright material of Fig. 1 after W. K. Witt, D. R. Mason & D. P. rock physics and seismic modeling of fluid substitution effects. Geophys. Prospect.
50, 119–137. https://doi.org/10.1046/j.1365-2478.2002.00304.x.
Hammond (2009) Archean Karari gold deposit, Eastern Goldfields Kern, H., Mengel, K., Strauss, K., Ivankina, T., Nikitin, A., Kukkonen, I., 2009. Elastic
Province, Western Australia: a monzonite-associated disseminated gold wave velocities, chemistry and modal mineralogy of crustal rocks sampled by the
deposit, Australian Journal of Earth Sciences, 56:8, 1061-1086, rep- Outokumpu scientific drill hole: evidence from lab measurements and modeling.
Phys. Earth Planet. Inter. 175 (3–4), 151–166.
rinted by permission of Taylor & Francis Ltda. Part of this work was Malehmir, A., Durrheim, R., Bellefleur, G., Urosevic, M., Juhlin, C., White, D.J., Milkereit,
supported by resources provided by Pawsey supercomputing Centre B., Campbell, G., 2012. Seismic methods in mineral exploration and mine planning: a
with funding from Australian Government and the Government of general overview of past and present case histories and a look into the future.
Geophysics 77 (5), WC173–WC190. https://doi.org/10.1190/geo2012-0028.1.
Western Australia. Manger, G.E., 1963. Porosity and bulk density of sediments. Bullentin. https://doi.org/
10.3133/b1144E.
Appendix A. Supplementary data Mavko, G., Mukerji, T., Dvorkin, J., 2011. Rock Physics Handbook. Tools for Seismic
Analysis of Porous Media, second ed. Cambridge University Press, New York.
McCuaig, T.C., Kerrich, R., 1994. P-T-t-deformation-fluid characteristics of gold deposits:
Supplementary data to this article can be found online at https:// evidence from alteration systems. In: Lentz, D.R. (Ed.), Alteration and Alteration
doi.org/10.1016/j.oregeorev.2020.103776. Processes Associated with Ore-forming Systems. Geological Association of Canada,
Short Course Notes, pp. 339–379.
Miah, K.H., Bellefleur, G., Schetselaar, E., Potter, D.K., 2015. Seismic properties and ef-
References fects of hydrothermal alteration onVolcanogenic Massive Sulfide (VMS) deposits at
the Lalor Lake in Manitoba, Canada. J. Appl. Geophys. 123 (2015), 141–152.
Abzalov, M.Z., 2013. Measuring and modelling of dry bulk rock density for mineral re- Mihayo, F., 2018. Geology of the Karari Deposit, Leonora, WA: Investigation of a
source estimation. Appl. Earth Sci. 122 (1), 16–29. https://doi.org/10.1179/ Mineralised Core. M.S. thesis. Curtin University.
1743275813Y.0000000027. Paterson, N.R., Hallof, P.G., 1991. Geophysical exploration for gold. In: Gold Metallogeny
Avseth, P., Mukerji, T., Mavko, G., 2010. Quantitative Seismic Interpretation: Applying and Exploration. Springer, Boston, MA. https://doi.org/10.1007/978-1-4613-0497-
Rock Physics Tools to Reduce Interpretation Risk. ISBN 0-521-81601-7. Cambridge 5_12.
University Press, New York. Rose, A.W., Burt, B.M., 1979. Hydrothermal alteration. In: Barnes, H.L. (Ed.),
Bass, J.D., 1995. Elasticity of minerals, glasses, and melts. In: Ahrens, T.J. (Ed.), Mineral Geochemistry of Hydrothermal Ore Deposits, second ed. Wiley Interscience Chap.,
Physics & Crystallography: A Handbook of Physical Constants. American Geophysical pp. 173–235.
Union, Washington, DC, pp. 45–63. Salisbury, M.H., Milkereit, B., Bleeker, W., 1996. Seismic imaging of sulfide deposits: Part
Birch, F., 1960. The velocity of compressional waves in rocks to 10 Kilobars, part 1. J. I. Rock properties. Econ. Geol. 91, 821–828. https://doi.org/10.2113/gsecongeo.91.
Geophys. Res. 65, 1083–1102. https://doi.org/10.1029/JZ065i004p01083. 5.821.
Blewett, R.F., Cassidy, K.F., Champion, D.C., Henson, P.A., Goleby, B.S., Jones, L., Salisbury, M.H., Harvey, C.W., Matthews, L., 2003. The acoustic properties of ores and
Groenewald, P.B., 2004. The Wangkathaa Orogeny: an example of episodic regional host rocks in hardrock terranes. In: Eaton, D.W., Milkereit, B., Salisbury, M.H. (Eds.),
D2 in the late Archean Eastern Goldfields Province, Western Australia. Precambr. Res. Hard Rock Seismic Exploration. SEG, pp. 9–19.
130, 139–159. https://doi.org/10.1016/j.precamres.2003.11.001. Salisbury, M.H., Snyder, D., 2007. Application of seismic methods to mineral exploration.
Chopping, R., 2008. Geophysical signatures of alteration. Project A3 Final Report: In: Goodfellow, W.D. (Ed.), Mineral Deposits of Canada: A Synthesis of Major Deposit
Predictive Mineral Discovery Cooperative Research Centre, Geoscience Australia. Types, District Metallogeny, the Evolution of Geological Provinces, and Exploration
Christensen, N.I., Mooney, W.D., 1995. Seismic velocity structure and composition of the Methods. Geological Association of Canada, Mineral Deposits Division, Special
continental crust: a global view. J. Geophys. Res. 100, 9761–9788. Publication, pp. 971–982.
Colvine, A.C., Fyon, J.A., Heather, K.B., Marmont, S., Smith, P.M., Toop, D.G., 1988. Saracen Website. Saracen Mineral Holding Limited’s ASX released august 2019. Available
Archean lode gold deposits in Ontario, Canada. Ontario Geol. Surv., Miscell. Paper online: https://www.saracen.com.au/files/2315/6462/0081/1953851.pdf,
136–139. %20accessed%2016%20October%202019%20acessado%20em%2016/10/2019.
Dentith, M., Mudge, S.T., 2014. In: Geophysics for the Mineral Exploration Geoscientist. (accessed on 10 October 2019).
Cambridge University Press. https://doi.org/10.1017/CBO9781139024358. Schetselaar, E., Bellefleur, G., Hunt, P., 2019. Elucidating the effects of hydrothermal
Dentith, M., Enkin, R.J., Morris, W., Adams, C., Bourne, B., 2020. Petrophysics and mi- alteration on seismic reflectivity in the footwall of the Lalor Volcanogenic Massive
neral exploration: a workflow for data analysis and a new interpretation framework. Sulfide Deposit, Snow Lake, Manitoba, Canada. Minerals 9, 384. https://doi.org/10.
Geophys. Prospect. 68 (1), 178–199. https://doi.org/10.1111/1365-2478.12882. 3390/min9060384.
Dinnebier, R.E., Billinge, S.J.L., 2008. In: Powder Diffraction: Theory and Practice. RSC Schodde, R.C., 2019, Long term trends in gold exploration: presentation to the 2019
Publishing, Cambridge. https://doi.org/10.1039/9781847558237. NewGenGold conference. https://minexconsulting.com/long-term-trends-in-gold-
Doyle, H.A., 1986. Geophysical exploration for gold – a review. Explor. Geophys. 17 (4), exploration-2/, accessed on July 2020.
169–180. https://doi.org/10.1071/EG986169. Schon, J., 2015. Physical Properties of Rocks: Fundamentals and Principles of
Duff, D., Hurich, C., Deemer, S., 2012. Seismic properties of the Voisey's Bay massive Petrophysics. Elsevier, Oxford.
sulphide deposit: insights into approaches to seismic imaging. Geophysics 77, Souza, A.E.C.M., Vialle, S., Bona, A., 2019, Ultrasonic wave velocities measurements and
WC59–WC68. https://doi.org/10.1190/geo2011-04831. seismic anisotropy at Karari gold deposit: Implications for gold exploration. in: 2nd
Eaton, D., Milkereit, B., Salisbury, M., 2003. Hardrock seismic exploration: mature Australasian Exploration Geoscience Conference, AEGC 2019, Expanded Abstracts,
technologies adapted to new exploration targets. In: Foreword to Hardrock Seismic Paper 132. DOI:10.1080/22020586.2019.12073010.
Exploration. SEG. https://doi.org/10.1190/1.9781560802396.fm. Środoń, J., Drits, V.A., McCarty, D.K.J., Hsieh, J.C., Eberl, D.D., 2001. Quantitative X-ray
Eilu, P.K., Mathison, C.I., Groves, D.I., Allardyce, W.J., 1999. In: Atlas of Assemblages, diffraction analysis of clay-bearing rocks from random preparations. Clay Miner. 49,
Styles and Zoning in Orogenic Lode-Gold Deposits in a Variety of Host Rock and 514–528. https://doi.org/10.1346/CCMN.2001.0490604.
Metamorphic Setting. Geology and Geophysics Department (Centre for Strategic Sylvester, P., 2012. Use of the mineral liberation analyzer (MLA) for mineralogical studies
Mineral Deposits) & UWA Extension, The University of Western Australia, of sediments and sedimentary rocks. In: Sylvester, P. (Ed.), Short-Course.
Publication, pp. 30–50. Mineralogical Association of Canada (MAC), St. John’s, NL, Canada, pp. 1–16.
Telford, W.M., Geldart, L.P., Sheriff, R.E., 1990. In: Applied Geophysics, second ed.

13
A.E. Calazans Matos de Souza and S. Vialle Ore Geology Reviews 126 (2020) 103776

Cambridge University Press. https://doi.org/10.1017/CBO9781139167932. Australia. Econ. Geol. 103, 445–454. https://doi.org/10.2113/gsecongeo.103.2.445.
Thomsen, L., 1986. Weak elastic anisotropy. Geophysics 51 (10), 1954–1966. https://doi. Witt, W.K., Mason, D.R., Hammond, D.P., 2009. Archean Karari gold deposit, Eastern
org/10.1190/1.1442051. Goldfields Province, Western Australia: a monzonite-associated disseminated gold
Villaescusa, E., Hogan, P., 2020, Stress measurements from oriented core using the deposit. Aust. J. Earth Sci. 56, 1061–1086. https://doi.org/10.1080/
Acoustic Emission method at Karari mine. Saracen’s internal report (unpublished). 08120090903246188.
Wilkens, R.H., Fryer, G.F., Karsten, J., 1991. Evolution of porosity and seismic structure Wu, G., Chen, G., Wang, D., Cheng, Q., Zhang, Z., Yang, J., Xie, S., 2020. Identifying
of upper oceanic crust: importance of aspect ratios. J. Geophys. Res. 96, mineral prospectivity using seismic and potential field data in the Hongniangyu
17981–17995. https://doi.org/10.1029/91JB01454. district, Inner Mongolia, China. Ore Geol. Rev. 119, 103317. https://doi.org/10.
Witt, W.K., Hammond, D.P., 2008. Archean Gold Mineralization in an intrusion-related, 1016/j.oregeorev.2020.103317.
geochemically zoned district-scale alteration system in the Carosue Basin, Western

14

You might also like