You are on page 1of 18

Geophysical Prospecting, 2015, 63, 1015–1032 doi: 10.1111/1365-2478.

12255

Characterization of seismic reflections from faults in a crystalline


environment, Schneeberg, Germany
Lena Schreiter ∗ , Felix Hloušek, Olaf Hellwig and Stefan Buske
Institute of Geophysics and Geoinformatics TU Bergakademie Freiberg, Freiberg Gustav-Zeuner-Str., 12 D-09599 Freiberg, Germany

Received October 2014, revision accepted January 2015

ABSTRACT
We present an approach for analysing seismic reflections from faults in a crystalline
hard rock environment. We analysed 3D seismic reflection data for geothermal reser-
voir characterization acquired in the Erzgebirge Region, Germany. The seismic image
derived from this data set revealed two main features: a less pronounced reflector
corresponding to a steeply dipping major fault zone Roter Kamm and a group of
pronounced reflectors attributed to the existence of conjugate mineralized faults. We
analysed these reflections in the pre-stack data to characterize the nature and origin of
reflectivity. This was done by extracting the corresponding waveforms from the raw
data and carefully pre-processing them, including amplitude correction for geometri-
cal spreading and signal-to-noise enhancement. Reflection coefficients were derived
from the pre-processed shot gathers by comparing the amplitudes of the reflected and
direct waves. Synthetic waveform modelling using the reflectivity method has been
performed for several model families consisting of one-dimensional velocity–depth
functions with varying velocities, densities, and thicknesses of the layers. A compari-
son of the modelled and observed waveforms revealed that a reflection coefficient of
0.18 for the conjugate mineralized faults can be explained by single layers with high
impedance contrast and a thickness between 30 m and 40 m, whereas the reflection
from the Roter Kamm fault zone with a reflection coefficient of −0.23 requires a
model consisting of several low-velocity layers with a total thickness of up to 100
m embedded in a high-velocity background model. These results are in accordance
with the geological interpretation of these reflectors. However, the characteristics of
these reflections vary significantly within the investigation area, both in terms of the
reflection coefficient and the waveform, which is also in agreement with the general
lateral variation of fault zone characteristics known from tectonic investigations such
as geological mapping of outcrops and fabric analysis.

Key words: Modelling, Interpretation, Reservoir geophysics, Rock physics, Seismics.

INTRODUCTION structural elements in crystalline rock units (Eaton, Milkereit,


and Salisbury 2003). However, sharply defined and highly re-
Seismic investigations of the Earth’s crust often provide diffuse
flective structures can be observed in several cases. Initially,
and, in some cases, even incoherent images of structures in a
their cause seems unclear and is often associated with tectonic
crystalline hard rock environment. This illustrates the charac-
fault systems or ore mineralization (Malehmir et al. 2014). In
teristic heterogeneity and small-scale distribution of reflective
some cases, the influence of trapped fluids is proposed as a
cause of highly pronounced reflectors (bright spots) at mid-
∗ E-mail: lena.schreiter@student.tu-freiberg.de crustal levels (Klemperer and BIRPS Group 1987).


C 2015 European Association of Geoscientists & Engineers 1015
1016 L. Schreiter et al.

Studies on the origin of seismic reflections in crystalline This approach is similar to studies using conventional
rocks require seismic data sets with high signal-to-noise ra- amplitude-versus-offset (AVO) analysis; however, we are not
tios and efforts to preserve the true amplitudes. The task is performing actual AVO analysis since it is not appropriate
to identify reflectors by relating physical parameters to ge- for characterizing complex geology and low signal-to-noise-
ological settings. Usually this is accomplished by correlating ratio events. We designed model families with variations in
borehole seismic measurements or laboratory measurements geometrical and petrophysical parameters of the fault sys-
of the elastic rock properties in core samples (e.g., White, tem under consideration. The modelling parameters contain
Secord, and Malinowski 2012) with images from surface seis- characteristics of the fault system, including the width of the
mic surveys. For example, Juhlin (1990) interpreted several different layers, the velocity variations within the layers, and
reflectors in the Siljan impact structure (Sweden) with an up density variations due to mineralization or possible rupture
to 60-m-thick dolerite intrusion in contact with granite. zones. Synthetic seismograms act as a reference for the ob-
In another case study, Zillmer, Müller, and Stiller (2002) served waveforms and provide potential interpretations for
analysed a 3D seismic data set from the German Continen- the reflection strength and other waveform characteristics. As
tal Deep Drilling site (KTB) in order to relate petrophysical a result of this study, these parameters are indirectly derived
parameters to the observed reflectors. By using the true am- from the structure image, although the actual frequency con-
plitudes of the seismic data set, we followed the methodology tent of the source does not allow us to determine a unique
of Zillmer et al. (2002) to derive the approximate reflection model that explains the observations.
coefficients and an internal fault zone structure. Laminated In this study, we describe the geological setting and the
models with strong velocity contrast and maximum fault zone dominant tectonic features within our hard rock environment,
widths of 300 m produced the highest reflection coefficients followed by structural characterization based on a 3D seismic
of 0.1 – 0.2. For the same study area, Harjes et al. (1997) image cube. After that, the approach for deriving reflection
suggested that high reflection coefficients of 0.1 – 0.15 origi- coefficients is presented, which includes the analysis of the
nate from fluid-filled fractures and cataclastic fracture zones. direct wave and the fault reflections in terms of their reflection
In another seismic study of the crystalline crust, Trela (2003) strength and waveform characteristics. Finally, this method is
performed seismic modelling to evaluate strong reflections of applied to the real data set, resulting in quantitative structural
the PS-converted wave using zero-offset vertical seismic pro- models of the two different types of faults under consideration
filing seismograms. By performing reflectivity modelling and and includes a geological interpretation.
adjusting the frequency dependence, the best approximation
was found using a 20-m-thick low-velocity homogeneous pla-
nar layer with a Vp/Vs ratio of 2 and a P-wave velocity reduc- GEOLOGICAL SETTING AND SEISMIC
tion from 6500 m/s to 4800 m/s. IMAGING RESULT
Based on the aforementioned methodology, we were mo-
Geological setting
tivated to conduct a similar study on a high-quality 3D seis-
mic data set acquired in the Erzgebirge region within the The investigation area is located in the Western Ore Moun-
project Seismik im Kristallin Sachsen (SIKS). The aim of this tains (see Fig. 1), situated between the cities Aue and Schnee-
project was to study the feasibility of seismic exploration berg at the intersection of two large-scale tectonic structures
methods to characterize a major fault system in crystalline (Hiller and Schuppan 2008; Wallner et al. 2009). These are the
rock, which could potentially be used as a deep geothermal SW–NE striking Loessnitz–Zwoenitzer syncline and the NW–
reservoir. SE trending Gera–Jachymov fault system. The geology is char-
The purpose of our work was to understand the origin of acterized by various schists (phyllites and mica schist), which
significant reflective structures observed in the seismic image were deformed by folding during the Variscan orogeny. Later,
in a systematic way with the help of different reflector models magmatic activities penetrated the formation with granite in-
and seismic waveform simulation. Specifically, the main trusions in the S and SW. The older metamorphic succession
focus lies on the characterization of structures whose thickn- covers the younger granite body in the central survey area.
ess is much smaller than the wavelength, such as tectonic The phyllites and mica schists are in turn covered by volcanic
faults and vein systems, resulting in piecewise coherent and sedimentary of Rotliegendes in the northern part.
reflections and poorly imaged features. Based on this goal we The study area represents the most exposed part of the
designed a 1D acoustic forward modelling workflow. Gera–Jachymov fault zone in Saxon territory, whose activity


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1017

Figure 1 Geographical location and geological setting of the investigation area. SGs and receiver lines, used for our reflectivity study, are shown
in the left map, and the strike of the structure Roter Kamm is visible in the right map. The line A-A’ corresponds to the geological cross section
shown in Fig. 2.

had the greatest influence on the actual geological setting. As 3D seismic survey
a result of faulting, a latticed system of fractures is observed. The project SIKS aimed to image the complex and partly
Multiple activations of the fault zone during the post-orogenic steeply dipping fault system using a high-resolution 3D Vi-
granite intrusion lead to a complex system of dykes. During broseis seismic survey acquired in 2012 and covering an area
these processes the system also experienced multi-stage hy- of about 11.5 km × 14.0 km (black rectangle in Fig. 1). The
drothermal mineralization (Hiller and Schuppan 2008). This source signal was a 12 Hz–96 Hz upsweep with a length of
is observed at many NW–SE trending discontinuities. These 12 seconds. Up to 6000 receivers recorded simultaneously for
hydrothermal filled faults can be distinguished from contin- each of the 5349 source positions. Further details of the survey
ually occurring large structures as formations with low ex- parameters and the results of applying advanced seismic imag-
tension to micro fractures. They can be classified by extent, ing techniques to this data set can be found in Hloušek, Hell-
thickness, and composition (Wallner et al. 2009). wig, and Buske (2015). The main features found in the result-
The dominant faults have been mainly observed and ing 3D seismic image cube are a group of parallel SW-dipping
analysed in the upper 700 m–1000 m due to former mining high-amplitude reflections. According to the geological data
activities (Fig. 2). The main structure of the study area is (see Figs. 1 and 2), these reflectors presumably represent the
the Roter Kamm fault zone, which has the character of a conjugate faults embedded in the granite body as described
normal fault and documented thicknesses of up to 100 m above. In contrast, the Roter Kamm fault zone is in some parts
(Wallner et al. 2009; Hiller and Schuppan 2008). In contrast, not directly visible in the 3D seismic image cube. However,
several other structures can be described as conjugate faults, an internal offset in the image of some of the conjugate faults
which are tectonically overprinted and exhibit hydrothermal can be associated with a normal fault character and indirectly
multi-stage mineralization. According to the geological sec- with the existence of the Roter Kamm fault (Hloušek et al.
tion, their dip (50◦ –70◦ SW) is opposite to that of the Roter 2015). We carefully selected the subsets, the so-called shot
Kamm fault (65◦ –70◦ NE). The thickness of the conjugate groups (SGs), from the unmigrated 3D seismic data set that
faults is determined to be at least 10 m from mining activities. illuminate the Roter Kamm fault in the NW of the survey area
Near the granite body, a general increase is observed with (Fig. 1). Consequently a direct and strongly pronounced im-
fault thicknesses varying between 40 m and 60 m (Hiller age of the Roter Kamm was obtained. The appearance of both
and Schuppan 2008). In general, a network of large-scale fault zones is illustrated in Fig. 3 for SG 58 in the SW and SG
structures with a complex vein system dominated by two 34 in the NE of the survey area (see Fig. 1). The Roter Kamm
oblique structures, the tectonic rupture zone (Roter Kamm), fault appears as a single highly reflective structure down to a
and conjugate faults with mineralization complexes consti- depth of approximately 3 km (Fig. 3a). Its dip (65◦ –70◦ ) fits
tute the geological and tectonic framework within the study to the extrapolated surface trace and to the documented in-
area. mine observations (Hiller and Schuppan 2008). The conjugate


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1018 L. Schreiter et al.

Figure 2 Geological cross section from SW to NE corresponding to line A-A’ in Fig. 1, modified after Wallner et al. (2009). The cross-section
shows the location of the granite body (red) covered by various schists (blue, green). The steeply dipping normal fault Roter Kamm represents
the main structure. Also, several conjugate faults (Bad Elster, Schwerin, etc.) are shown (mapped locations in the mine indicated in green and
extrapolated to greater depth in black).

Figure 3 Result of Kirchhoff pre-stack depth migration for subsets of the 3D data set (stacked absolute values, see Hloušek et al. (2015) for
details). (a) Single NE-dipping structure imaged by SG 34 representing the Roter Kamm fault. (b) SW dipping reflectors imaged by SG 58 (see
Fig. 1 for SG locations) associated with the conjugate faults.

faults are similarly imaged in the migration result, where the Kamm and continue into its footwall. In general, it is observed
shallower conjugate faults appear in SG 58 between depths that the reflectivity of the conjugate structures weakens with
of 1 km and 3 km (Fig. 3b). Therefore, the conjugate faults depth, which might be related to the limited coverage and
extend deeper in the structural image than what is shown in aperture of the survey area or simply due to the decreasing
the geological cross section, indicating they cross-cut the Roter reflectivity with depth. Compared with the 3D seismic image


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1019

cube by Hloušek et al. (2015), the image slice shown here The resulting fit for selected shots is presented as a solid line
(Fig. 3b) displays only two single representative structures in Fig. 4. Despite some local effects, the fitted curve explains
with a highly reflective character, which are part of the group the general amplitude decay of the first arrival with offset
of parallel NW–SE-dipping faults. These results allow us to quite well. In general, the first arrivals are affected by local
directly compare the highly reflective structures to the domi- near-surface inhomogeneities and ground coupling effects of
nant coherent onsets of reflections observed in the pre-stack the geophones, as well as energy losses when the wave travels
shot gathers. through the subsurface. To compensate for these effects, two
corrections were applied. First, in order to reduce the local
near-surface effects, the amplitudes were aligned to match the
WORKFLOW FOR ESTIMATING
theoretical amplitude decay curve with offset. Then, a scaling
REFLECTION COEFFICIENTS
factor was applied to compensate for the general amplitude
Analysis of the direct wave decay with offset. In a homogeneous medium, the energy de-
crease due to geometrical spreading of the wavefront as it
The first step in our characterization workflow is to derive
propagates away from the source can be described by the
the reflection coefficients. For that a comparison between the
reciprocal of the squared travel distance (Sheriff and Geldart
arrival of direct and reflected P-waves is performed according
1995). In an inhomogeneous medium, energy decrease may be
to Zillmer et al. (2002). In this case we call the direct wave
faster than expected and can be approximately compensated
the reference wave.
by the scaling term rx, which was applied to the amplitudes
The analysis of first arrivals has been performed for 10
of the direct wave using the exponent determined by the fit-
shots in SG 58 and 21 shots in SG 34 (green and blue boxes
ting procedure described above. Differences in the x values are
in Fig. 1), with each having up to 6000 traces along 22 and
observed within all receiver lines and are caused by local near-
23 receiver lines per shot, respectively. The distance between
surface inhomogeneities. The c-value refers to the first-arrival
geophone lines and geophone groups was about 400 m and 30
amplitude for zero offset and varies on average between 12
m, respectively. Due to local conditions (property access) the
and 16.
distribution of geophone groups was irregular in some parts.
The maximum offset was about 10 km.
The first-arrival waveforms are similar to the autocor- Analysis of the reflections from the conjugate faults
relation function of the used Vibroseis sweep signal, i.e., the and the Roter Kamm fault zone
Klauder wavelet. We determined the first-arrival amplitudes
In order to analyse the reflections from the Roter Kamm fault
by selecting the absolute amplitude maximum within a time
and the conjugate faults in more detail, we conducted a closer
window of 50 ms around the waveform. These absolute am-
inspection of the pre-stack single shot gathers within the se-
plitudes are shown in Fig. 4 for some representative SGs as a
lected SGs. No gain correction or other pre-processing steps
function of the offset r. For small offsets (r < 1.0 km), we ob-
were applied to the gathers, meaning the analysis was per-
serve a strong decay of the first-arrival amplitudes as expected.
formed on the raw data.
For intermediate offsets (r = 1.5 – 3 km) some variations are
We used the source and receiver coordinates to calculate
present, which are probably caused by local inhomogeneities
theoretical two-way travel times for all shot-receiver config-
in the weathering layer, and result in strong velocity and den-
urations relating to the potential reflectors visible in the mi-
sity changes.
gration slices (Fig. 3). For these calculations, the background
We use a power law, i.e.,
velocity was set to 6000 m/s in accordance to the velocity
A = cr−x , (1) model derived by Hloušek et al. (2015). This simple constant
velocity model is justified because of the rapid transition from
to correct for the characteristic amplitude decay with off-
the near-surface overburden to the approximately constant
set, where A is the maximum absolute amplitude of the
velocity of 6000 m/s for the granite body. The calculated two-
first-arrival waveform.
way travel times allowed easy identification of the reflections
The coefficients c and x are determined by linear poly-
from the two fault zones under consideration within the single
nomial curve fitting of the calculated amplitudes to the real
single-shot gather.
amplitudes according to
Reflections from the conjugate faults can be clearly
log(A) = c − x log(r ). (2) identified in all shots within SG 58, predominantly on six


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1020 L. Schreiter et al.

Figure 4 Reference wave amplitudes for indicated shot locations of group no. 58 (shot no. 1430, 1431, 1433, and 1437) and 34 (shot no. 5410,
5411, 5412, and 5418) and some receiver lines (Fig. 1). Amplitudes (dots) are shown for the offsets corresponding to the observed reflections
in the seismograms. Factor x reproduces the fitted amplitude decay (solid line) by applying the used power law.

receiver lines, between two-way travel times of 1 seconds and After this first visual inspection, reflection coefficients for
2 seconds. Figure 5 shows an example from shot number 1433 both fault reflectors were determined. In each case we consid-
and receiver line number 5160, which runs in the SW–NE di- ered only a limited number of traces that showed sufficiently
rection almost perpendicular to the strike of the conjugate good signal quality. These traces were taken from different
faults. In this seismogram section several pronounced reflec- offsets, mainly between 500 m and 2000 m, with the assump-
tions can be traced as coherent signals. One can clearly iden- tion that they contain reflections resulting from near-normal
tify similar first-arrival and reflected-arrival polarities. Also, incidence onto the corresponding reflector. For these traces,
reflections from the Roter Kamm fault can be easily identi- varying time shifts was applied so the reflection waveforms
fied on several shots of SG 34 between two-way travel times all have the same travel time. This time shift was obtained
of 1.4 seconds and 1.8 seconds. Figure 6 shows an example by cross-correlating the waveforms. Subsequently, the aligned
from shot number 5412 and receiver line number 5253. Al- waveforms were stacked to obtain a representative waveform
though the waveforms of the direct and reflected arrivals are with an increased signal-to-noise ratio.
not as clear as in Fig. 5, it already appears that the direct Co-located traces to be used for stacking were selected
arrival and the reflected arrival have opposite polarity in the considering the first Fresnel zone on the reflector. The maxi-
case of the Roter Kamm reflection. mum spatial spread of the reflection points can be calculated,


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1021

Figure 5 Data example for shot 1433 and receiver line 5160. The seismogram section shows a clear first arrival as well as coherent reflections
(marked by green arrows) from the conjugate faults. An automatic gain control (AGC) has been applied for display purposes.

taking the first Fresnel zone into account. For a target depth reflection coefficients vary between 0.1 and 0.3, with an aver-
of 2000 m and a wavelength of 125 m, this yields a diameter age of around 0.18. This clearly represents a high value in ac-
of 700 m. In most cases, the spatial spread of the reflector cordance with the pronounced appearance of these conjugate
area was on average 300 m. This also represents the maxi- reflectors in the seismic image and with the a priori geological
mum offset difference within one selected seismogram section interpretation as mineralized fault zones. Nevertheless, pos-
(for stacking and extracting waveform), i.e., the distance from sible tuning effects must be taken into account, which must
the first to the last trace within one section. In order to obtain be discussed in detail also with regard to the waveform of
representative waveforms, we stacked on average ten neigh- the reflection. Applying the same procedure to the reflection
bouring traces. from the Roter Kamm fault yields an average reflection coef-
Next, the amplitude of this representative waveform was ficient of 0.23 (see Fig. 8). However, in this case, we found
corrected for geometrical spreading in the same way as de- a wider spread of amplitude ratios, which together with the
scribed above for the direct wave by using the reciprocal of more complex reflection waveform, suggests that interference
the corresponding travel distance. In this case, the travel dis- effects for certain angles and offsets are present. This might
tance was calculated as the product of velocity and travel time. be caused by a more complex internal structure of the Roter
The resulting representative waveform can be considered the Kamm fault.
near-normal-incidence reflection, which can be compared di-
rectly with the direct wave in terms of its amplitude ratio.
ANALYSIS OF THE REFLECTIONS AND
For that we used the stacked maximum amplitudes of the di-
EXPLANATION OF WAVEFORMS BY
rect wave within the corresponding section. This yields the
FORWARD MODELLING
corresponding near-normal-incidence reflection coefficient.
Applying this procedure for selected conjugate reflections Forward modelling studies are mandatory for understand-
lead to the amplitude ratios as a function of offset shown ing not only the values of the reflection coefficients but also
in Fig. 7 for different shots and receiver lines. The resulting the waveforms of the reflections. This includes an assessment


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1022 L. Schreiter et al.

Figure 6 Data example for shot 5412 and receiver line 5253. Although the first-arrival waveform appears more complicated compared with
Fig. 5, the reflection from the Roter Kamm fault (marked by the green arrow) is pronounced and well visible. An AGC has been applied for
display purposes.

of the influence of geometrical and petrophysical parameter the duration of the autocorrelation function of the Vibroseis
variations with regard to the internal structure of the faults. source signal (12 seconds, upsweep 12 Hz–96 Hz) in our data
In this case, the geological situation is relatively simple be- set.
cause a reflecting fault is located within a nearly homoge-
neous granitic environment that has no strong velocity con-
Resolution test using wedge model
trasts. Moreover, the ray paths between source and reflector
are almost straight; therefore, we can assume that acoustic for- The vertical resolution depends on the wavelength of the
ward modelling based on a 1D model is sufficient to explain source signal propagating in the subsurface with a certain
the first-order effects on the waveforms and amplitudes. The velocity. For the Ricker wavelet with 48-Hz peak frequency
study is carried out for a certain wavelet with fixed wave- and an increased velocity for an embedded fault of 6500 m/s,
form (frequency content) assuming normal incidence onto the wavelength is approximately 135 m. In order to separate
the reflector. We explicitly exclude the elastic case, includ- two reflection events within a single seismogram, their dif-
ing conversions, because the acquisition was performed with ference in travel time has to be larger than the length of the
vertical-component receivers only, and we also exclude in- signal itself so that these two events do not interfere. In our
trinsic and scattering attenuation. Our applied 1D modelling case this means that the layer must have a thickness of at least
approach is based on the reflectivity method introduced by 68 m. If the layer thickness is larger, then the two separate
Müller (1985). It comprises a matrix formalism that is capable reflections can be easily distinguished. If the layer thickness is
of computing the full wavefield response of an incident plane smaller, then the two reflections interfere, resulting in a more
wave with chosen source signal onto a horizontally layered complex waveform. The amplitudes of this resulting wave-
medium, including transmission losses and internal multiple form are influenced by this interference, causing the so-called
reflections. tuning effects.
For the first resolution tests we have chosen a Ricker We have tested the behaviour of seismic waves in a wedge
wavelet with a peak frequency of 48 Hz as a source signal. This model that represents our fault within hard rock environment,
wavelet is well understood and resembles well the shape and i.e., a layer of varying thickness (10 m–100 m) with high


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1023

Figure 7 Amplitude ratios between selected conjugate fault reflections and the reference wave as a function of source–receiver offset. The
assigned reflection coefficient is the calculated average for this shot/receiver line. Crosses represent the raw amplitude ratio (calculated for
the uncorrected data) and green circles correspond to the amplitude ratio of the stacked and corrected waveforms. The dashed lines indicate the
general trend within the respective shot/receiver line.

velocity (6500 m/s) and density (3300 kg/m3 ) embedded in a From the seismic response in Fig. 9 we can conclude the
homogeneous background with velocity of 6000 m/s and den- following.
sity of 2600 kg/m3 . Figure 9 shows the resulting waveforms
including an analysis of their estimated peak-to-peak ampli- – The layer thickness of about 27 m produces the maximum
tudes. The superposition and amplification of the individual peak-to-peak amplitude. For this layer thickness (light blue
wavelets can produce reflections of considerable strength. trace in Fig. 9, left) constructive interference leads to an


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1024 L. Schreiter et al.

Figure 8 Amplitude ratios between the Roter Kamm reflection and the reference wave as a function of source–receiver offset. The reflection
coefficient is the calculated average for this shot/receiver line. Crosses represent the raw amplitude ratio (calculated for the uncorrected data)
and red circles correspond to the amplitude ratio of the stacked and corrected waveforms. The dashed lines indicate the general trend within
the respective shot/receiver line.

amplification (tuning), which is approximately 1.43 times and multiple reflections are also considered in the waveform
larger than the amplitude for a single interface reflection simulation, so that an exact doubling of the maximum tuning
(Amax ). This thickness is slightly smaller than the often amplitude cannot be expected.
assumed quarter-wavelength “tuning thickness” (Kallweit – For a layer thickness larger than the tuning thickness, the
and Wood 1982). Constructive interference occurs for a har- amplitude of the first reflection converges to the reflection
monic wave with frequency ω if the reflected waves from the from a single interface (green wavelet in Fig. 9 left, green
top and the bottom of a layer with thickness d are shifted by level in right window). The thickness of the layer can be
half a period T, so that directly determined from the time difference of the wavelets.
The minimum resolvable layer thickness is given by the clear
vT λ
d= = , (3) separation of two reflected wavelets (approximately 68 m).
4 4
– For a layer thickness smaller than the tuning thickness, the
with the layer velocity v and the wavelength λ. In our case two reflections from the top and the bottom of the layer
the Ricker wavelet has a broad frequency band around its interfere and cannot be distinguished from each other. The
peak frequency of 48 Hz, so that this formulation can only layer itself can be detected, but its boundaries cannot be
be considered an approximation. Furthermore, transmission resolved.


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1025

Figure 9 Wedge modelling result showing the interference effects of the resulting waveform (left) and the related differences in the peak-to-peak
amplitude (blue) and two-way travel time (red) for two reflected zero-phase wavelets with opposite polarity (right). The peak-to-peak amplitude
is normalized to the amplitude for a first-order discontinuity (single interface) with the same parameter contrast (green).

These observations show that we must analyse the ob- First-order discontinuity
served waveforms of the reflections in terms of their ampli-
The first model considered is a first-order discontinuity, which
tudes and travel-time differences between their constituting
is described by an interface between two homogeneous half-
wavelets.
spaces each with a constant velocity and density. The dimen-
sion of the model space is therefore reduced to two parame-
Model space and parameters ters: velocity and density contrast.
For the conjugate faults we assumed an impedance in-
For a simple model of a single layer we can vary three vari-
crease based on the previous observations (polarity), so that
ables: the velocity contrast, the density contrast, and the thick-
the reflection coefficient is expected to be positive. Therefore,
ness of the layer. These three variables define the model space
to reproduce a reflection coefficient of 0.18, as inferred from
that has to be searched for the parameter combinations that
the data analysis, a first-order discontinuity requires one of the
explain the observed amplitude and waveform of the reflec-
following:
tion. Different combinations of these three parameters enable
the classification of model families. We first distinguish be- – velocity increase of 43% to 8600 m/s with no density
tween models with a first-order discontinuity and a single contrast;
layer within a homogeneous medium, which is the so-called – density increase of 44% to 3740 kg/m3 with no velocity
lamella. Multi-layer models are considered later for a more de- contrast;
tailed analysis of the waveform. In all cases, the surrounding – velocity and density increase each of 20% to 7200 m/s and
medium represents a granitic environment with a P-wave ve- 3120 kg/m3 , respectively.
locity of 6000 m/s and a density of 2600 kg/m3 (Schön 1983).
In contrast, the description of the Roter Kamm fault as
According to the geological setting, these parameters represent
a rupture zone implies an impedance reduction; therefore, the
a good approximation of the rock mass around the illuminated
modelled reflection coefficient is expected to be negative. A
parts of the Roter Kamm fault and the conjugate faults.
reflection coefficient of −0.23 can be produced by any of the
The modelling results yield a certain waveform with peak-
following parameter variations:
to-peak amplitude and corresponding reflection coefficient,
which has to be assessed by comparing it with the representa- – velocity decrease of 37% to 3760 m/s with no density
tive waveform obtained from the real data. contrast;


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1026 L. Schreiter et al.

Figure 10 Model space for a single layer embedded in a hard-rock environment with a range of values for the P-wave velocity (v2, p ), density
(ρ2 ), and layer thickness (z2 ). The model signal was a Ricker wavelet with 48 Hz. The embedded layer (index 2) has an increased impedance
I2 , compared with the homogeneous hard-rock medium (I2 > I1 = I3 , indexes 1 and 3 for the upper and lower half-spaces, respectively). Red
colours show the region of reflection coefficients (R = 0.20), which are close to the averaged observed reflection coefficient of the conjugate
faults in the real data.

– density decrease of 37% to 1640 kg/m3 with no velocity For this model we have performed the corresponding
contrast; modelling calculations as described above and computed ab-
– velocity and density decrease each of 21% to 4750 m/s and solute values of the normal incident reflection coefficients
2060 kg/m3 , respectively. for a range of layer velocities (6000 m/s–8000 m/s), densities
(2600 kg/m3 –3600 kg/m3 ), and thicknesses (10 m–65 m). The
In general, high impedance contrasts are required to gen-
largest thickness of 65 m has been chosen to ensure the su-
erate reflections of the observed intensity. These estimates
perposition of the reflected wavelets from the top and the
only give a general impression of the required contrasts for a
bottom of the layer. Any thickness greater than that would
single interface; however, the geological situation shows that
allow the distinction between the two reflected wavelets (see
we have to deal with a relatively thin fault layer (lamella)
Fig. 9) and is not considered here for comparison with the
embedded in a mostly homogeneous granitic environment.
observed waveform in the real data. The results are shown in
Fig. 10, together with representative examples of the resulting
Lamella
waveforms for a velocity of 6500 m/s for the layer, as well
The second model includes lamella and allows for the discus- as in Fig. 11 for a specific layer thickness of 30 m. The red
sion of constructive interference in order to generate observed region in both figures represents absolute values of the re-
high reflection amplitudes for smaller impedance contrasts flection coefficient around 0.2, which is close to the averaged
compared with a first-order discontinuity. A lamella is a ho- reflection coefficients that we estimated from the real data for
mogeneous layer with a certain thickness and a velocity and the conjugate faults. It is clear from both figures that certain
density contrast to the surrounding two half-spaces. parameter combinations within the model space can be used


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1027

Figure 11 Part of the model space shown in Fig. 10 (green box), i.e., for a single layer embedded in a hard-rock environment with a layer
thickness of 30 m. The model signal was a Ricker wavelet with 48 Hz. The seismic traces illustrate the effects on the waveforms for a variation
of a single parameter of the embedded layer (v2, p or ρ2 ). The parameter is increased by 10% (reflection coefficient R = 0.07), and the variation
is determined for which a reflection coefficient of R = 0.2 is reached (v2, p +32% and ρ2 + 35%).

to explain waveforms with similar reflection coefficients, e.g., range. Note again the appearance of the reflection between
the ones observed for the conjugate faults. two-way travel time of 1.5 seconds and 1.6 seconds in Fig. 5
Assuming that the velocity and density change simulta- and amplitude ratios in Fig. 7.
neously, a reflection coefficient of 0.2 can be explained by In general, the characteristic composition for all wave-
different rates of increase for various layer thicknesses. For forms related to the conjugate faults in the small to interme-
layer thicknesses smaller than the tuning thickness, the den- diate offset range (see left blue trace in Fig. 12) is an initial
sity and velocity increase would have to be approximately strong negative peak from the top interface followed by a
30% for a 10-m layer (7800 m/s, 3350 kg/m3 ) and 20% for strong positive peak from the bottom interface. The real data
a 20-m layer (7100 m/s, 3100 kg/m3 ) to explain a reflection waveform (left blue trace in Fig. 12, from shot 1433, line
coefficient of 0.2. Near the tuning thickness (≈ 30 m), the 5160, stack of seven traces 3686–3692) matches well with
same reflection coefficient can be explained by a small in- the modelled waveform from the lamella layer model with a
crease of approximately 13% (6800 m/s, 3000 kg/m3 ). If the thickness of 43 m, a velocity increase to 7400 m/s, and a den-
layer thickness exceeds the tuning thickness, this rate increases sity increase to 2800 kg/m3 . The reflection coefficients range
to approximately 23% (7380 m/s, 3200 kg/m3 ). from 0.1 to 0.18 and up to 0.23 for increasing offsets. This
can be related to discontinuous thicknesses along the lateral
extent of the reflector as well as the internal structure. The
Multi-layer models
models that best describe the reflection strength are plotted
The previous analysis has focused on the values of the reflec- next to the extracted waveforms in Fig. 12 and are charac-
tion coefficient. In the following section, the corresponding terized by a mean velocity of 7100 m/s (+/− 400 m/s), mean
variation of the waveform is discussed in more detail because density of 2830 kg/m3 (+/− 170 kg/m3 ), and mean thickness
a good approximation of the reflection coefficient does not of 38 m (+/− 5 m). These parameter values are applicable to
necessarily imply that a good approximation of the reflected most of the other examples. When comparing the shape of the
waveform has been obtained. waveform in each section, they sufficiently match the smallest
To continue the analysis from the previous section, we and middle offset ranges.
first focus on the waveforms of the conjugate faults. Repre- The modelled waveforms can explain most of the ob-
sentative extracted waveforms from the real data are shown served traces, but as expected, not all of them, as shown in
as blue traces in Fig. 12 and correspond to coherent signals Fig. 12 (right). Taking multi-layer models into account, we
from shot 1433 on receiver line 5160 for a sequential offset can assume that the highly reflective part of the interface


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1028 L. Schreiter et al.

Figure 12 Modelling result for the conjugate faults for three reflector models with a 1D velocity–depth function and the corresponding synthetic
seismogram. The model signal was a Ricker wavelet with 48 Hz. Reflection coefficients are given on the top of the boxes and are compared
with real data. The first-order discontinuity shows the parameter contrast for the mean reflection coefficient (0.18) with a velocity increase to
7200 m/s (density increase to 3120 kg/m3 ). For the lamella and the multi-layer model, the comparison was carried out with reference to the
waveform (real data, blue). The real data waveform of shot 1433, line 5160, and stack 3686 – 3692 (seven traces) matches the composite
waveform by the lamella layer model of 43 m with a velocity increase to 7400 m/s and density of 2800 kg/m3 . A multi-layer model best matches
the waveforms of shot 1433, receiver line 5160, stacked along traces 3670 – 3685 (16 traces), which is characterized by a embedded dominant
layer of 38 m thickness, a velocity increase to 6800 m/s, and density of 2700 kg/m3 as well as surrounding smaller low-velocity zones according
to a typical fracture model.

refer to a high impedance contrast caused by observed re- an additional positive peak and extend with some more side
flection strength. However, this reflective part could be sur- lobes, clearly distinguishable from the noise level. An analysis
rounded by small low-impedance fractured zones, which of the multi-layer model yields a sufficient explanation of the
slightly change the waveform phases. To reproduce the ob- more complex waveforms, as well as a better fit of the peak-to-
served waveforms, it is assumed that reflected waveforms from peak amplitude ratios within the waveforms. This is why we
the interfaces within the multi-layer model are superposed. can relate the interior of the fault zone with several layers of
Since this effect was encountered several times in the data set, different parameter variations and thus different impedance,
we can infer the geological model. The highly reflective part which generates a similar reflection pattern. Of course, one
can be associated with granite fractures of different vertical cannot expect that the waveforms can be reproduced in every
extent and discontinuous hydrothermal mineralization, such detail due to small-scale scattering effects and the underlying
as quartz dikes or carbonate-rich mineralization as implied simplified assumptions of the modelling approach. Finally, the
by geological prospecting. The adjacent zone on the top and following geological model is proposed for the Roter Kamm
bottom interfaces could be characterized by brecciated struc- fault: it is represented by a stepwise 1D velocity–depth model,
tures and dislocations of the surrounding rock, followed by a where the high amplitudes originate from a low-velocity zone
transition to the unfractured granite. and then return stepwise to the background velocity. Such a
Modelling tests for the Roter Kamm fault are exemplified medium is commonly used to model fault zones (Eaton et al.
in Fig. 13 and show that a single-layer model can produce the 2003). Another possible model is described by alternating
reflection strength of most signals by an average parameter layers within the interior of the fault zone. This seems
combination of 5300 m/s (+/− 400 m/s) for the velocity, plausible given what is known from the previous geological
2000 kg/m3 (+/− 300 kg/m3 ) for the density, and 33 m (+/− prospecting of the near-surface area, where dislocated rocks
6 m) for the thickness. Since the assumption of a single-layer alternate with quartz dykes, rock veins, and clay beds.
model relates mainly to the maximum amplitudes of the
signals, the above parameters can be used only for the highly
DISCUSSION
reflective zone of the Roter Kamm fault. This main part of
the reflection is led by a strong positive first peak followed Reflection coefficients were derived for the conjugate faults
by a trough. In some waveforms these phases are followed by and the Roter Kamm fault by initially comparing the


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1029

Figure 13 Modelling result for the Roter Kamm fault showing three reflector models with a 1D velocity–depth model (solid black step function),
the resulting synthetic waveform (black, solid), and the observed waveform (blue) of two different shots (5411 and 5417) on two different lines
(5253 and 5106), stacked along 11 traces (4103 – 4113) and 20 traces (1124 – 1143), respectively. The model signal was a Ricker wavelet with
48 Hz. The composite waveforms (left side) can be explained by a gradient model with a velocity decrease to 5100 m/s, a density decrease to
2480 kg/m3 , and a total thickness of 59 m. A second example can be modelled by alternating low- and high-velocity zones with a maximum
velocity drop to 5500 m/s, density of 2280 kg/m3 , and total thickness of 113 m. Reflection coefficients are given on the top of the boxes and
are compared with real data observations.

reflections with the first arrivals (direct wave). The reflection by the seismic quality factor Q, which quantifies the loss of
coefficients were estimated by calculating the amplitude ratio energy per wave cycle. In our forward modelling approach,
of the first arrivals and the reflections. The incident wave on no intrinsic attenuation was taken into account, which means
the reflector was approximated by the direct wave since we that the quality factor was set implicitly to Q = ∞. In or-
assumed almost straight ray paths in a nearly homogeneous der to consider attenuation, we could assume a representative
granitic environment with no significantly different rock units. quality factor of Q = 500 that corresponds to a mainly ho-
Therefore, we also assumed that the correction term for com- mogeneous granitic environment (Zillmer et al. 2002). The
pensating energy losses due to spherical divergence is valid for amplitude of the signal A(x) can then be written as (Mavko,
both the direct wave and the reflection. Given the actual wave- Mukerji, and Dvorkin 2009):
 
length of our source signal, intrinsic attenuation and scattering −π 2x
A(x) = A0 e Q λ
, (4)
attenuation are considered to be small in our rather consoli-
dated granitic environment. where A0 is the amplitude of the source signal, x is the length
The reflector model of a first-order discontinuity repre- of the travel path, and λ denotes the wavelength. For a travel
sents an initial assumption for the impedance contrast, causing path of about 2000 m and a dominant wavelength of 125 m,
a mean reflection coefficient observed in subsets of +0.18 for the amplitude decreases to approximately 82% of the source
the conjugate faults and −0.23 for the Roter Kamm fault. signal amplitude. In our case, with reflection coefficients for
In addition, some attenuation effects in the almost homo- the conjugated faults of +0.22 and the Roter Kamm fault of
geneous and consolidated granitic environment can affect the −0.28, by calculating the reflection coefficients for the case
derived approximate reflection coefficients and should be dis- of Q = ∞, we are underestimating the amplitudes by ap-
cussed. We observed that the reflection waveforms are similar proximately 22%. This result is similar to estimates given
to the auto-correlation function of the Vibroseis source signal by Zillmer et al. (2002), where the reflection coefficients for
and do not show a significant coda caused by multiple scatter- the SE1 reflector are increased by 35% when using a qual-
ing. For this reason, we assumed that scattering attenuation ity factor of Q = 500 compared with the case of Q = ∞. In
does not play a major role in our case and does not have such that sense, our calculated reflection coefficients are realistic
a significant impact on our reflection coefficients. However, but slightly underestimated because the intrinsic attenuation
intrinsic attenuation may have an impact. It can be described has not been considered. However, for a more quantitative


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1030 L. Schreiter et al.

Figure 14 Geological model derived by combined interpretation. The colour scale ranges from high-velocity zones (yellow) over a granitic
environment (red, approximately 6000 m/s) to low-velocity zones (blue). The conjugate faults may be represented by a mineralized fault and
surrounding low-velocity zones due to local tectonic dislocations. The Roter Kamm fault may be represented by alternating low and intermediate
velocity layers, which emphasizes a typical fault zone model with a dominant rupture zone.

analysis of this effect, the spatial distribution of attenuation because the superposition of the reflected signatures on several
above the reflectors, e.g., derived by attenuation tomography, interfaces are comparable to the observed waveforms.
would be required. This is the subject of further investigations. The reflections of the conjugate faults could originate
The comparison of synthetic waveforms with observed from fracture zones healed by mineralizations, which causes
reflections implies that the reflective structures can be ex- higher impedance with a velocity increases of 18% (7100 m/s)
plained by the reflector models consisting of a single thin and density increases of 9% (2830 kg/m3 ). Furthermore con-
layer, with thicknesses smaller than the wavelength, and em- sidering the local tectonic situation, it is possible that the sur-
bedded in a granitic background medium. Based on the ana- rounding breccia structures could impact the lower impedance
lysis of matching maximum amplitudes and the waveforms, contrast. The strongest reflective structure of the Roter Kamm
the highly reflective parts of the Roter Kamm fault and the fault could be associated with a low-velocity fault zone, where
conjugate faults have thicknesses ranging from 30 m to 40 m. different low-impedance layers might be present.
The differences between these structures are revealed in the Multi-layer models show that a total thickness of fault
polarity analysis, which is an essential tool in distinguishing zone of up to approximately 100 m is possible. The gradient
low- and high-velocity zones (Juhlin, Lindgren, and Collini models, as are typical for fault zones (Eaton et al. 2003) and
1991), and furthermore on interference effects, which provide which fit the Roter Kamm fault reflector best, also include
the potential to declare composite waveforms. alternating sequences of high and low impedance. The fault
We selected the best fitting models so that the density and is represented by a strong lateral variation of composition
velocity change with similar rates, which is a simplified but and reflection strength and an average velocity decrease of
plausible assumption. Furthermore, the tradeoff mainly be- 12% (5300 m/s) and density decrease of 23% (2000 kg/m3 ).
tween velocity and thickness indicates that a clear determina- Figure 14 represents a schematic illustration of the combined
tion of a parameter contrast is not possible, and the mentioned interpretation of these findings.
values only represent a trend. Additionally, focusing on repro- Although the survey geometry illuminates the dipping
ducing the waveforms and not only their absolute maximum faults in large areas, it is possible that the reflectivity of
amplitude, multi-layer models provide sufficient explanations the Roter Kamm fault changes locally and that most of its
of reflectivity of the Roter Kamm and conjugate fault system parts within the investigation area cannot generate such high


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
Seismic reflections from faults 1031

reflection amplitudes as exemplified by this subset. As Zillmer changes in the same order. Such high reflection strength re-
et al. (2002) discussed, the relation of reflection strength for quires constructive interference of the single reflections. How-
the low-velocity zone at the KTB, we must also assume that ever, reflection strength and waveforms vary within the inves-
less reflective parts for the Roter Kamm fault exist and that we tigation area, which is in accordance with lateral variation
have determined the impedance that represents the strongest of the fault characteristics. The evidence for our results comes
possible parameter contrast. In general, the assumption of from a 3D high-resolution Vibroseis survey and tectonic inves-
extrapolating the derived fault zone parameters to greater tigations. The results quantitatively support the idea that the
depth along the same structure is of course limited. conjugate faults represent fractures healed by hydrothermal
Our study is a first step in deriving rock-physical mineralization and have thicknesses smaller than the wave-
parameters from seismic data within the project SIKS and length used, ranging between 30 m and 40 m. The reflective
shows that fault zones and mineralized fractures in granite part of the Roter Kamm fault can reach a thickness up to 100
environments are distinguishable by high-resolution 3D m and can be associated with a rupture zone, which is possibly
surface seismic data. Describing the structures in more detail the result of a normal fault.
and characterizing more features observed in the 3D seismic
image cube indicate that further seismic investigations, such
ACKNOWLEDGEMENTS
as borehole logs and vertical seismic profiling data, are
necessary for correlating petrophysical parameters. One step The authors would like to thank the German Federal Ministry
in analysing the influence of rock veins and their filling for Economic Affairs and Energy (BMWi) for funding the
could be performed using shear wave analysis and anisotropy project SIKS (FKZ 0325363B) and Catherine Alexandrakis
investigations. Finally, the relationship between reflectivity for language editing.
and the observed seismicity in the mine should be considered.
Since velocity reduction is not the only indicator for increased
micro fracturing, determination of Poisson ratios may indicate REFERENCES
if there are fluids within micro fractures (Trela 2003). The Eaton D.W., Milkereit B. and Salisbury M.H. 2003. Hardrock
derivation of rock-physical parameters should be supported Seismic Exploration. Society of Exploration Geophysicists. ISBN
by borehole logs and laboratory studies on cores. This is nec- 9781560801146.
essary to evaluate the inferred impedance values and to study Harjes H.-P., Bram K., Dürbaum H.-J., Gebrande H., Hirschmann G.,
the influences of lithological, structural, and micro fracture Janik H. et al. 1997. Origin and nature of crustal reflections: Re-
sults from integrated seismic measurements at the KTB superdeep
variations. Finally, an additional near-surface seismic survey
drilling site. Journal of Geophysical Research 102(18), 267–
may give an answer in classifying parameters due to their 288.
lateral variation in comparison with the geological section Hiller A. and Schuppan W. 2008. Bergbau in Sachsen - Geologie
and makes extrapolations for deeper regions more applicable. und Uranbergbau im Revier Schlema-Alberoda, Bd. 14. Sächsiches
Landesamt für Umwelt Landwirtschaft und Geologie (LfUG). ISBN
3981142136.
CONCLUSIONS Hloušek F., Hellwig O. and Buske S. 2015. Three-dimensional fo-
cused seismic imaging for geothermal exploration in crystalline rock
Reflection coefficients of the strongest continuous reflections (Schneeberg, Germany), Geophysical Prospecting 63(4), 999–1014.
within an almost homogeneous granitic environment are on Juhlin C. 1990. Interpretation of the reflections in the Siljan Ring area
average +0.18 for the conjugated faults and -0.23 for the based on results from the Gravberg-1 borehole. In: Seismic Probing
of Continents and their Margins (eds J.H. Leven, D.M. Finlayson,
Roter Kamm fault. The analysed reflections satisfied our crite-
C. Wright, J.C. Dooley and B.L.N. Kennett). Tectonophysics 173,
ria relating to near-normal incident, coherence, signal-to-noise
345–360.
ratio, and spatial spread of the reflection point. Nevertheless, Juhlin C., Lindgren J. and Collini B. 1991. Interpretation of seis-
many weaker reflections are present, and it is possible that mic reflections and borehole data from Precambrian rocks in
some reflection amplitudes are below the noise level. Litho- the Dala Sandstone area, central Sweden, First Break 9, 24–
logical explanations for the analysed faults comprise compo- 36.
Kallweit R.S. and Wood L.C. 1982. The limits of resolution of zero –
sitional layering with velocity contrasts of about 6000 m/s to
phase wavelets. Geophysics 47(7), 1035–1046.
7100 m/s for the conjugate fault scenario and a velocity de- Klemperer S.L. and the BIRPS Group 1987. Reflectivity of the crys-
crease from 6000 m/s to 5300 m/s for the Roter Kamm fault. talline crust: hypotheses and tests. Geophysical Journal of the Royal
In both cases, the velocity contrast is combined with density Astronomical Society 89, 217–222.


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032
1032 L. Schreiter et al.

Malehmir A., Koivisto E., Manzi M., Cheraghi S., Durrheim Trela C. 2003. Seismische Untersuchungen der kristallinen Kruste
R. J., Bellefleur G. et al. 2014. A review of reflection seismic inves- an der KTB-Lokation. PhD thesis, Christian-Albrechts-University,
tigations in three major metallogenic regions: The Kevitsa Ni–Cu– Kiel, Germany.
PGE district (Finland), Witwatersrand goldfields (South Africa), Wallner O., Hiller A., Frenzel M. and Frommhold R. 2009. De-
and the Bathurst Mining Camp (Canada). Ore Geology Reviews taillierte geowissenschaftliche Datenaufbereitung im Vorzugsge-
56, 423–441. biet Aue-Schwarzenberg im Rahmen des Tiefengeothermieprojektes
Mavko G., Mukerji T. and Dvorkin J. 2009. The Rock Physics Hand- Sachsen. Internal Report.
book: Tools for Seismic Analysis in Porous Media. Cambridge Uni- White D.J., Secord D. and Malinowski M. 2012. 3D seismic imaging
versity Press. ISBN 9780521861366. of volcanogenic massive sulfide deposits in the Flin Flon mining
Müller G. 1985. The reflectivity method: a tutorial. Journal of Geo- camp, Canada: Part 1–Seismic results. Geophysics 77(5), WC47–
physics 58, 153–174. WC58.
Zillmer M., Müller G. and Stiller M. 2002. Seismic reflections from
Schön J. 1983. Petrophysik. Akademie-Verlag, Berlin, Germany.
crystalline crust below the Continental Deep Drilling Site KTB:
Sheriff R.E. and Geldart L.P. 1995. Exploration Seismology. Cam- Modeling and inference on the reflector properties. Journal of Geo-
bridge University Press. ISBN 0521462827. physical Research 107(B9), ESE 2-1–ESE 2-15.


C 2015 European Association of Geoscientists & Engineers, Geophysical Prospecting, 63, 1015–1032

You might also like