You are on page 1of 9

See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/249868008

Shale rock physics and implications for AVO


analysis: A North Sea demonstration
ARTICLE in THE LEADING EDGE JUNE 2008
DOI: 10.1190/1.2944164

CITATIONS

DOWNLOADS

VIEWS

128

218

4 AUTHORS, INCLUDING:
Per Avseth
Norwegian University of Science and Techno
58 PUBLICATIONS 354 CITATIONS
SEE PROFILE

Available from: Per Avseth


Retrieved on: 15 August 2015

Shale rock physics and implications for AVO analysis:


A North Sea demonstration
PER AVSETH, Rock Physics Technology, Bergen, Norway.
ANDERS DRGE and AART-JAN VAN WIJNGAARDEN, StatoilHydro Global Exploration Technology, Bergen, Norway
TOR ARNE JOHANSEN, University of Bergen, Norway
ARILD JRSTAD, Lundin-Norway, Oslo, Norway

Shales normally constitute more than

80% of sediments and sedimentary


rocks in siliciclastic environments.
Shales are important both in controlling the overburden seismic wave
propagation as well as the reflectivity
contrast between cap rocks and reservoir rocks in prestack seismic data.
Therefore, during AVO analysis it is
crucial to understand the seismic
properties of shales as a function of
mineralogy and compaction. In this
study, we derive the local shale trend
for a North Sea gas-and-oil field by
integrating rock physics modeling
with well-log and seismic data analysis.
Based on the shale trend model- Figure 1. Schematic illustration of sand and shale compaction. At about 70C it is common to
ing, we focus on the following key observe a change from mechanical compaction to predominantly chemical compaction in siliciclastic
systems. In deep-marine depositional systems, smectite-rich shales will experience illitization and
issues in this study: 1) What is the seis- release of bound water, causing both a porosity reduction and mineralogy change with depth. For
mic significance of clay mineral trans- quartz-rich sands, initial cementation tends to start at the same depth level. One possible external
formations occurring at various depths source of cement is in fact derived from the smectite to illite transition in embedding shales.
(only smectite-to-illite transition is considered in this paper)? 2) How is the
AVO background trend controlled by shale compaction? 3) shales are often associated with distance to the coastline and
What are the optimal AVO attributes to be used at various are characterized by variation in silt content and certain types
depths given the interplay between shale and sand com- of clay minerals. However, in this study, we will focus on
paction? 4) What is the effect of shale intrinsic anisotropy the diagenetic trends of marine shales encountered in deepas a function of depth, and how will this affect the AVO sig- water sequences of the Tertiary age in the North Sea and
how the chemical compaction of these shales is associated
natures of sandstone reservoirs capped by shale?
We demonstrate that improved understanding of the with the quartz cementation in embedded turbidite sandphysical properties of shales as a function of burial depth, stones.
Figure 1 shows a schematic representation of shale and
in conjunction with a good understanding of how compaction affects rock properties of sandstones, will improve sand compaction curves and a sequence of interbedded turour ability to characterize and predict hydrocarbons in sand- bidite sands and marine shales, typical for the North Sea
deep-marine environment of the Tertiary age. The deposistone reservoirs embedded in shales.
tional porosity in shales is normally much higher (6080%)
Mechanical versus chemical compaction in shales and than in sands (about 40%), but we expect a shallow crossover
sandstones. Until recently, shales have often been regarded with depth due to the mechanical collapse of the shales. The
among geophysicists as a unique type of lithology, and platy clay fabric in the shales is more prone to compaction
minor attention has been given to the great variance in min- than the assemblage of spherically shaped grains in sands;
eralogy, texture, and porosity of shales during seismic data hence, the more rapid mechanical porosity reduction in
analysis. This is partly because the rock properties of clay shales than sands. During burial to ~2 km depth, both sands
minerals are difficult to measure in the laboratory, but also and shales are exposed mainly to mechanical compaction
because the oil industry has given little priority to the acqui- (e.g., Storvoll et al., 2005). It is found that the marine shales
sition of detailed log data and core samples in shale in the Tertiary North Sea are very smectite-rich, and these
sequences. Geologists, however, have documented the com- have a very low permeability. In thick smectite-rich shale
plexity of shales, and there exists a vast amount of published masses, it is therefore normal to observe undercompaction
literature on the geochemistry and sedimentology of shales and associated overpressure even at several hundred meters
(e.g., Bjrlykke, 1998; Peltonen et al., 2008). With increased burial depth. However, at about 70C, chemical alteration
focus on cross-disciplinary integration, geophysicists are of smectite can initiate, and we expect a mineral transforstarting to incorporate this geological knowledge into the mation to illite and possibly chlorite. This is a typical minmodeling and analysis of geophysical data (e.g., Drge et eral transformation seen in marine shales all over the world
(Bjrlykke, 1998). Bound water in the smectite layers is
al., 2006; Brevik et al., 2007; Mondol et al., 2007).
Similar to sandstones, shales can exhibit both deposi- released when the temperature reaches this critical temtional and diagenetic trends. Depositional trends in marine perature, and this results in a porosity decrease. Moreover,

788

THE LEADING EDGE

JUNE 2008

Figure 2. Rock physics depth trends for shales (blue) and sandstones (cyan), juxtaposed on North
Sea well-log data penetrating a Tertiary sequence of siliciclastic sediments and rocks. A gas zone is
indicated in yellow and an oil zone in red. The remaining interval of the Heimdal Formation is brinefilled. The Heimdal Formation is embedded in the Lista Formation shale.

Amorphous silica can dissolve at temperatures lower than crystalline


quartz. There are also potential internal sources of quartz cement in the
sandstones, either from authigenic or
detrital clays in the sandstone matrix,
or from the quartz grains themselves.
Moreover, it has been confirmed that
clay minerals can both inhibit the precipitation and catalyze the dissolution
of quartz in sandstones. The presence
of oil has also been shown to inhibit
quartz cementation. There are controversies among geologists over
whether effective pressure at quartz
grain contacts in sandstones can cause
quartz dissolution and reprecipitation
around the grain contacts. More likely,
time and temperature control the
quartz cementation. Regardless of the
complexity of the clay and quartz diagenesis, there is empirical evidence
that both smectite-to-illite transformation in shales and quartz cementation of sandstones are geochemical
processes that tend to happen concurrently at around 70-80C. In the
North Sea, this corresponds to a burial depth of about 2 km, and this is
around the target depth of the prolific
Paleocene and Eocene reservoir sands
that represents major prospects for the
oil industry. As geophysicists, we
therefore need to pay attention to these
geological factors during AVO analysis of reservoir sands, because dramatic changes in the seismic
signatures may reflect not only pore
fluid changes, but diagenetic alterations in the cap rock shales and/or
reservoir sandstones.

Rock physics and AVO depth trend


modeling. The primary objective of
Figure 3. Modeled rock physics depth trends of shales, showing the effect of illitization of marine
this study is to investigate the rock
smectite-rich shales. The anisotropy parameters derived from the modeling, that are important for
physics depth trends of shales, and
AVO studies, include in green and in red.
how they affect the AVO signatures.
However, in order to predict depththe presence of potassium cations (for instance in feldspar dependent AVO signatures in siliciclastic environments, we
or mica) causes quartz to be produced as a by-product. This also need to consider the depth trends of sands.
quartz can precipitate as microcrystalline quartz within the
Figure 2 shows log data from a North Sea well peneshale matrix, or, if connectivity allows, the quartz may pre- trating siliciclastic sediments and rocks of Tertiary age, juxcipitate as cement in adjacent sandstones. The chemical reac- taposed with rock physics depth trends for shales and
tion can be summarized as follows (Peltonen et al., 2008): sandstones. The sandstone rock physics models as a function of depth are modeled by combining Hertz-Mindlin
(1) contact theory for unconsolidated sands with the DvorkinSmectite + K+ Illite + SiO2 + H2O
Nur contact cement model for cemented sandstones
This chemical reaction shows that there can be a link between (Dvorkin and Nur, 1996; Avseth et al., 2003; Avseth et al.,
the quartz cementation of sandstones and illitization of 2005). The input porosity-depth trends are calibrated with
smectite-rich shales. It is important to note that the source local compaction trends according to empirical relations. The
of quartz cement in sandstones is not unique, and can result light blue model trend curves in Figure 2 show how the
from a variety of geological processes. Figure 1 also includes velocities increase drastically for sands as we go from the
a volcanic tuff layer that is typically encountered in the unconsolidated regime with only mechanical compaction to
Tertiary of the North Sea (i.e., within the Balder Formation). the cemented regime with predominantly chemical comSmectite is often generated from the alteration of volcanic paction. The onset of quartz cement is known to happen at
tuff, but the tuff also includes amorphous silica that can pre- about 70C, corresponding to about 2 km burial depth in
cipitate as quartz cement, even at temperatures below 70C. the North Sea (Bjrlykke, 1998). The Heimdal Formation
JUNE 2008

THE LEADING EDGE

789

Figure 4. Rock physics crossplots of well-log data juxtaposed on depth


trend models for shale (blue), brine sands (cyan), oil sands (red), and gas
sands (yellow). The data are colored in terms of shale volume (upper) and
hydrocarbon saturation (lower) and include both a gas zone and oil zone
(see Figure 2).

sandstones in the well shown in Figure 2 appear to be


cemented as we see a drastic jump in velocities relative to
the overburden, and this is in agreement with earlier regional
studies of this stratigraphic unit (Avseth et al., 2000).
Shales have different composition and texture than sandstones. Therefore, the rock physics models applicable for
sandstones do not necessary apply for shales. This paper
uses the shale compaction model (Drge et al., 2006; Ruud
et al., 2003) to estimate the anisotropic effective properties
in mechanically compacted shales. The first seismically
important mineral reaction in shales is commonly the smectite-to-illite reaction. The reaction has several implications
for the shale; the soft smectite is replaced by stiffer illite,
which might be distributed differently in the rock, the reaction produces water, the amount of solids is decreased (i.e.
illite has a denser mineral structure than smectite), quartz
is generated as a by-product, and porosity is reduced by
chemical compaction. When the shales are moving into the
chemical compaction regime, a new set of rock physics models is applied to estimate the seismic properties. The
anisotropic versions of a differential effective medium (DEM)
model and self consistent approximation (SCA) are used to
approximate the elongated pores and grains in shales
(Hornby et al., 1994).
In this paper, pores in chemically compacted shales are
considered isolated, while the pores in the mechanical compaction regime are connected. We define a transition zone,
where the properties change from the mechanical to chemical regime. In Figure 2, the initial shale (<1500 m) is considered to be smectite-rich, while the deeper (>2200 m)
illite-rich shale is somewhat stiffer. The depth trends in
Figures 2 and 3 suggest that the background shale properties are neither static nor linear. There are two counteracting effects on anisotropy. The initial alignment of grains leads
to more aligned pores and increasing anisotropy. But decreasing porosity leads to decreasing anisotropy, culminating in the

Figure 5. AVO depth trends, including acoustic impedance, VP/VS, intercept (R(0)) and AVO gradient (G). AVO reflectivity modeling (see Figure 6)
is performed at three different depth levels: (1) at 1600 m burial depth where sands are unconsolidated and shales are smectite-rich; (2) at 2000 m burial
depth where initial cementation and associated illitization have started; (3) at 2400 m burial depth where the sandstones are well consolidated. (The
color scales of the two right panels are the same as the numerical scales along the x-axes, representing R(0) and G, respectively).
790

THE LEADING EDGE

JUNE 2008

Figure 6. AVO reflectivity modeling at the three different depth levels


indicated in Figure 5: (1) At 1600 m burial depth we obtain AVO class
IV for brine sands and class III for hydrocarbons. Note that the anisotropy
effect (dashed lines) is actually changing the sign of the gradient for brine
sands from slightly positive to slightly negative. (2) At 2000 m burial
depth, we observe a weaker fluid sensitivity, and the brine sands have
almost zero near-stack response, but there are significant negative gradients for all fluid scenarios. (3) At 2400 m burial depth, the brine sandstones show a class I AVO signature, whereas oil- and gas-filled
sandstones show a more class II AVO response. For all cases, the
anisotropy starts becoming significant beyond about 30 angle of
incidence.

anisotropic properties of the solid material at zero porosity.


The pores introduce higher anisotropy than the solid, since
pores commonly are weaker orthogonal to the longest axis,
while the solids are less dependent on direction of wave propagation. The Thomsen parameters of anisotropy, and , are
important for the AVO modeling below, because the transverse
isotropy of shales can alter the AVO gradient.
Figure 4 shows rock physics crossplots (acoustic imped792

THE LEADING EDGE

JUNE 2008

ance versus VP/VS including the well-log data in Figure 2, as


well as the rock physics depth trends for sandstone and shale.
In this crossplot, we also include the depth trends for oil and
gas sands; the various fluid scenarios are estimated using
Gassmann theory. We see that the modeled shale trend fits
nicely with the shale data from the well logs. The sandstone
data in the target zone are diagnosed as cemented based on
the rock physics models. As expected, the gas and oil filled
sandstones, indicated by low water saturation values in Figure
4, show low VP/VS values relative to the brine-sand values, in
accordance with the modeled depth trends. The separation
and detectability of hydrocarbon-filled sandstones would
have been even larger in terms of VP/VS if the sands were shallower and not cemented. It is expected that cementation stiffens the rock frame and reduces the fluid sensitivity. Acoustic
impedance is not a good discriminator of either lithology or
fluids for most of the depth range considered in this case. These
rock physics observations indicate that AVO analysis is
required in order to seismically discriminate hydrocarbons
from brine for this North Sea case. By conducting AVO depthtrend modeling, we can verify the feasibility of AVO analysis
as a function of burial depth and diagenesis and thereby
extrapolate to other burial depths. This can be a useful task
in exploration or appraisal where we want to extend our
search to slightly deeper or shallower prospects surrounding
existing discoveries.
Figure 5 shows the depth trends of acoustic impedance
and VP/VS superimposed onto the well-log data, together with
depth trends of intercept (R(0)) and AVO gradient (G). The
intercept and gradient curves are estimated on a sample-bysample basis where the cap rock properties are given by the
modeled shale trends, whereas the reservoir properties are
given by the modeled sandstone trends, for different pore fluid
scenarios (brine, oil and gas). Note how the expected seismic
reflectivity changes drastically as we go from the mechanical
compaction domain to the chemical compaction (cementation)
domain. Also note how the modeled fluid sensitivity drastically decreases over a range of only a few hundred meters, as
sandstones become cemented.
Figure 6 includes AVO reflectivity modeling of half-space
boundaries between cap rock shale and reservoir sandstone
performed at three different depth levels: (1) at 1600 m where
only mechanical compaction is occurring (i.e., in the unconsolidated regime); (2) at 2000 m in a transition zone where initial chemical compaction has taken place (i.e., in the poorly
consolidated regime); and (3) at 2400 m where the sandstones
are moderately to well cemented (i.e, in the consolidated
regime). In this reflectivity modeling, we include both isotropic
and anisotropic AVO modeling. In the latter case we assume
the cap rock shale to be transversely isotropic, with the
Thomsen parameters estimated from the shale modeling
described above. The offset-dependent reflectivity curves for
three-term AVO, including both isotropic and anisotropic
terms, are estimated using the following formula (Kim et al.,
1993):
RPP()=RIPP()+RAPP()

(2)

where the isotropic reflectivity term can be expressed as follows:


RIPP ()=R(0) + G . sin2 () + F .[tan2()sin2 ()]

(3)

Here, R(0) is the intercept or zero-offset reflectivity, G is the


AVO gradient, and F is the third-order curvature term affecting far offsets. is the average of the angles of incidence
and transmission (normally approximated to be the angle

Figure 7. AVO crossplots for well-log data in well 1, based on half-space


modeling between cap rock and target rock. The depth trends of brine
(light blue), oil (red) and gas (yellow) sands are superimposed. Brine
saturation is shown as color scale. The hydrocarbon anomaly matches the
models fairly well. The consolidated nature of the reservoir sandstones
results in a class II or weak class III AVO signature for the hydrocarbon
zone.

of incidence).
Moreover, the anisotropic reflectivity term for the case
where the cap rock is transversely isotropic and the reservoir is isotropic, is given by:

Figure 8. Seismic depth trends and expected product stack response. Note
that both oil and gas sands show large positive product stack values down
to about 22002300 m burial depth, whereas brine sands show weak or
negative product stack values for all depth ranges. The product stack is
known to be a good class III indicator. However, the gas and oil zone in
the well-log data have weaker product stack values than expected from the
clean sandstone trends. This is likely because the reservoir sands are
somewhat shaly.

(4)
As we see from this expression, one of the Thomsen parameters, , is affecting the same offset range as the contrast
in VP/VS ratio (i.e., the AVO gradient), whereas is affecting larger offsets. In the reflectivity modeling in Figure 6,
we consider angles of incidence up to 45. The isotropic
reflectivity curves are plotted as continuous lines, whereas
the anisotropic reflectivity curves are plotted as dashed
lines. We observe that the anisotropic curves start to deviate from the isotropic curves at around 30 for most of the
cases. However, for the uppermost case, the brine response
shows a change in gradient from weakly positive to weakly
negative due to the anisotropy. The AVO reflectivity modeling at the three different depths and compaction regimes
furthermore show that we go from a AVO class III to class
II for hydrocarbon-saturated sandstones within a few hundred meters depth.
Using the modeled shale trend as a cap rock analog, the
well-log data in Figure 2 can be crossplotted as AVO intercept versus AVO gradient (Figure 7). For each depth sample, we assume a two-layer model, where layer 1 (i.e., the
cap rock) is represented by the modeled shale trend, and
the properties of layer 2 are given by the well-log data.
Hence, this is only showing the expected top layer 2
responses for simple half-spaces. The plot does not include
the expected base layer 2 responses, neither does it include
any scale or tuning effects, which is expected in similar
crossplots derived from seismic data. Certainly, this crossplot representation of the well-log data is simplified, but it
gives a qualitative picture of the expected location of the
well-log data at a given depth. We have superimposed the
projections of the modeled sandstone trends for different fluids, where we also have used the modeled shale trend as
associated cap rock. These lines show the expected location
of brine, oil and gas sands as a function of the studied depth
range. For the well-log data, the hydrocarbon zone appears
as a class II AVO response, and this matches nicely with the

Figure 9. Seismic depth trends and expected fluid factor. Note that both
oil and gas sands show large negative fluid factor values down to about
2300-2400 m burial depth, whereas brine sands show weak or positive
fluid factor values for all depth ranges. Gas sands seem to give negative
fluid factor even deeper than the modeled window, beyond 2600 m. The
fluid factor is known to better discriminate both class II and IV AVO
anomalies in addition to class III and for the cemented section the fluid
factor should be a better attribute than the product stack. The gas and oil
zone in the well-log data also shows a relatively strong fluid-factor
anomaly as expected from the depth trends. Hence, the shaliness of the
reservoir seems to have smaller impact on the fluid factor attribute than
on the product stack.

modeled sandstone trends for oil and gas. However, most


of the reservoir is gas saturated, and the data plots between
the oil and the brine trend. It should be mentioned that the
modeled sandstone trends represent clean sandstone,
whereas the reservoir zone in the well-log data has a varying amount of clay, both pore-filling and laminated. The shaliness of the sands is expected to bring the hydrocarbon
anomaly closer to the brine saturated background trend.
Next, we use the modeled shale and sandstone trends
JUNE 2008

THE LEADING EDGE

793

Figure 10. Seismic sections intersecting the well studied in this case (well 1), including near stack, far stack, and the estimated far-near stack. The blue
square indicates the window where a background trend is estimated. The yellow ellipse highlights the gradient anomaly of the gas-and-oil discovery of
well 1. The red ellipse highlights an adjacent oil discovery.

to estimate AVO attributes as a function of depth. These


depth trends can be compared with the AVO attributes estimated from the well-log data. We do this for two wellknown AVO attributes estimated from combinations of
intercept and AVO gradient, the product-stack (i.e, R(0)
 G) and the fluid factor, respectively. The fluid factor,
which is defined as the distance from the background trend
in an AVO crossplot, can be expressed in terms of intercept
and gradient as (Fatti et al., 1994):
(5)
where  is a constant at a given depth, depending on the
background VP/VS ratio, , and the slope of the Vp-Vs relationship of the modeled shale trend, m:
 = m

(6)

Commonly, m is selected to be the slope of the mudrock line


(Castagna et al., 1985), where m = 1.16. Similarly, the background VP/VS ratio () is often set equal to 2, giving a  value
of .58. However, with our modeled shale trends, we can estimate more realistic depth dependent values of  and get a
better control on the expected background trend to be used
in the fluid factor.
Note that the product stack of both oil and gas sands is
expected to give large positive anomalies for depths down
to approximately 2200 m, beneath which it gets close to
zero, as we go from AVO class III to AVO class II (Figure 8).
In contrary to the product stack, the fluid factor is a much
more robust AVO attribute with respect to hydrocarbon
detection (Figure 9). For this attribute, we observe both the
794

THE LEADING EDGE

JUNE 2008

Figure 11. Intercept (near) versus gradient (i.e. far-near) for the seismic
stack section in Figure 10. Only data from the selected polygons/ellipses
are included. The blue points represent the background trend right above
the target; the yellow points represent data from the gas (and oil) discovery in well 1, and the red data points represent data from the adjacent oil
discovery.

gas sand and the oil sand to cause a negative fluid factor
anomaly relative to the background brine trend. The thin
oil zone is almost on the brine trend, likely due to relatively
high clay content and poor sand quality. However, the chemical compaction has a significant impact on the absolute
value of the fluid factor, which is directly related to the fact
that the fluid sensitivity is decreasing with depth and increas-

Figure 12. Product stack (above) and fluid-factor stack (below). The studied hydrocarbon discovery shown in the well-log data above is appearing
between CDP 450650 (indicated by yellow ellipse), with a relatively weak but still marked product stack anomaly and a strong fluid factor anomaly.
Both the top and the base of the reservoirs show positive product stack anomalies, whereas the top shows negative fluid factor, and the base shows
positive fluid factor. An adjacent oil discovery is apparent between CDP 8501000 (indicated by red ellipse) with a relatively strong product stack and a
similar fluid factor anomaly. (A third, smaller gas discovery is intersected by this line, between CDP 150200).

ing rock stiffness. It is well known that the fluid factor


detects AVO anomalies of class II, III, and IV, as long as the
deviation from the background trend is sufficient, and, in
this case, we see that the fluid factor will detect the class II
behavior of the anomaly encountered by the investigated
well.
Application to real seismic data. Finally, we want to test
the results from the modeling of expected AVO attributes
conducted above, on real seismic data from the same area.
Figure 10 shows seismic stack sections intersecting the well
studied above, including a near stack, far stack, and the difference between far and near. The latter is representative of
the AVO gradient as the near, and the far-stack data have
been preprocessed for true offset dependent amplitude
analysis. This line intersects three hydrocarbon discoveries
in Tertiary turbidites in the North Sea; in this study, we will
concentrate on two of these. The one penetrated by the
marked well (well 1) is highlighted by a yellow ellipse, and
this is the AVO anomaly caused by the gas and oil zone
shown in the well-log data. Furthermore, the anomaly highlighted by the red ellipse is a neighboring oil discovery. It
is interesting to note that the oil anomaly to the right of the
well has a strong soft (i.e., negative amplitudes) top on the
near stack, whereas the gas anomaly at the indicated well
location has a weak, close-to-zero amplitude top on the
same stack. Both anomalies show strong negative far-stack
responses. However, the gas anomaly shows a somewhat
stronger gradient anomaly. These observations are evident
in the crossplot in Figure 11. Here we see that the gas (and
796

THE LEADING EDGE

JUNE 2008

thin oil) discovery in well 1 shows a class II AVO anomaly,


whereas the neighboring oil discovery shows a class III AVO
anomaly. First of all, it is interesting to note that the modeled AVO attributes based on the well-log data and a representative shale trend match very nicely with the
observations in the seismic data (compare Figures 7 and 11).
However, at first glance, it is counterintuitive that the adjacent oil anomaly shows a class III anomaly. This cannot be
explained from the fluid properties, but it has to do with
the sandstone quality. The previous AVO depth-trend modeling indicates that we are at a depth level very close to the
initial quartz cementation of sandstones. Going back to the
schematic illustration in Figure 1, we can get a clue of why
the oil sands show up as class III while the gas sands show
up as class II. There is a good chance that some of the sands
at this depth are not yet cemented. There can be local variations and uncertainties of the onset of chemical compaction.
This can be related to many factors: local variations in temperature gradient, local variations in clay mineralogy within
the sands (e.g., clay coating of sand grains inhibit the quartz
cementation), or local variation of connectivity for external
sources of quartz, e.g., from illitization of shales or amorphous silica within the stratigraphically overlying volcanic
tuffs. Moreover, it has been documented that early migration of oil into reservoir sands can also inhibit quartz cementation.
Finally, we produce a product stack and a fluid factor
stack for the seismic line shown in Figure 10. By simple multiplication of the near stack with the far-near stack, we obtain
the product stack shown in Figure 12. The neighboring oil

discovery shows a strong product stack, whereas the gas discovery at well 1 shows a much weaker product stack. This
is expected, as documented in the depth trend modeling (see
Figure 9), where the gas discovery has been confirmed to
produce a class II AVO anomaly. We also produce a fluid
factor stack, which is included in Figure 12. The fluid factor can be derived from near- and far-stack amplitudes using
the formula:
F = Near . (FarNear)

(7)

Here, is not the same as  in the earlier formulation of the


fluid factor, since far-near is not exactly the same as the AVO
gradient. However, the two-term AVO makes a linear relationship between the far-near and the AVO gradient, and
therefore and  can also be easily correlated. Assuming
the far stack to be around 30 and the near stack to be at 0
(normally the far stack will be representative for slightly
lower angles than 30, whereas near stack will be representative for angles slightly higher than 0), we obtain the
following approximate relationship between the gradient
and the far-near:
FarNear = R (30)R(0) = G . sin (30) = G . 0.25
(8)
Solving for the background trend, that is when F=0, we
obtain:

Near =
hence we get:

20
(Far
8 5

Near )

(9)

(10)
The slope of the background trend in the near versus farnear stack is 1/, and this shows that we can in fact estimate the expected background trend for uncalibrated (but
offset balanced) range-limited stacks using the modeled
shale trends as illustrated in this study. This can be a useful task to verify the correct amplitude balancing between
near- and far-stack data during AVO crossplot analysis.
Based on the shale model trends, we estimate a value of
to be of order 2.5. This means the background trend of the
shale in the AVO crossplot is relatively flat, equaling -0.4,
i.e., far-near = -0.4 near. We observe indeed that the shale
background trend is relatively flat (blue cloud in Figure 11).
Using this background trend, the resulting fluid-factor
attribute in Figure 12 shows that both the gas and the oil
discoveries are standing out as strong fluid factor anomalies,
whereas the background data are relatively weak. It is
expected from the well-log observations that the brine-saturated rocks beneath the hydrocarbon reservoirs will have
a steeper negative slope in the near versus far-near crossplot. This can easily be tested by inputting different values
for the background VP-VS relationships.
Conclusions. A rapid transition from predominantly
mechanical compaction to predominantly chemical compaction occurs at around 70C corresponding to about 2 km

burial depth in the North Sea. Illitization of smectite-rich


shales and associated quartz cementation of both silty shales
and sandstones cause drastic changes in the elastic properties as this threshold burial temperature is exceeded. These
diagnetic alterations are important to consider during AVO
screening and quantitative seismic interpretation, especially
when the target depth is around this depth. Reservoir sandstones will obtain a stiffened framework and associated
decrease in fluid sensitivity as cementation occurs. Moreover,
the change in shale mineralogy and porosity will determine
the AVO background trend, and an improved understanding of rock physics depth trends and anisotropy of shales
can help us to predict expected AVO classes and better constrain the estimation of optimal fluid-factor attributes.
Suggested reading. Rock physics modeling of shale diagenesis by Drge et al. (Petroleum Geoscience, 2006). Clay mineral diagenesis and quartz cementation in mudstones: The
effects of smectite to illite transformation on rock properties
by Peltonen et al. (Marine and Petroleum Geology, 2008). Clay
mineral diagenesis in sedimentary basinsA key to the prediction of rock properties; examples from the North Sea Basin,
by Bjrlykke (Clay Minerals, 1998). Experimental mechanical
compaction of clay mineral aggregatesChanges in physical
properties of mudstones during burial by Mondol et al. (Marine
and Petroleum Geology, 2007). Documentation and quantification of velocity anisotropy in shales using wireline log measurements by Brevik et al. (TLE, 2007). Velocity-depth trends
in Mesozoic and Cenozoic sediments from the Norwegian
Shelf by Storvoll et al. (AAPG Bulletin, 2005). Anisotropic effective medium modeling of the elastic properties of shales by
Hornby et al. (GEOPHYSICS, 1994). Seismic properties of shales
during compaction by Ruud et al. (SEG Expanded Abstract,
2003). Elasticity of high-porosity sandstones: Theory for two
North Sea datasets by Dvorkin et al. (GEOPHYSICS, 1996). Rock
Physics diagnostics of North Sea sands: The link between
microstructure and seismic properties by Avseth et al.
(Geophysical Research Letters, 2000). AVO classification of lithology and pore fluids constrained by rock physics depth trends
by Avseth et al. (TLE, 2003). Quantitative Seismic Interpretation
Applying Rock Physics Tools to Reduce Interpretation Risk by Avseth
et al. (Cambridge University Press, 2005). Detection of gas in
sandstone reservoirs using AVO-analysis: A 3D seismic case history using the Geostack technique by Fatti et al. (GEOPHYSICS,
1994). Effects of transverse isotropy on P-wave AVO for gas
sands by Kim et al. (GEOPHYSICS, 1993). Relationships between
compressional-wave and shear-wave velocities in clastic silicate rocks by Castagna et al. (GEOPHYSICS, 1985). TLE
Acknowledgements. We thank Torbjrn Fristad, Erik degaard, Cindy
Gosse, and Trond Seland at StatoilHydro; Ran Bachrach at
WesternGeco/Schlumberger; Vigdis Lv at Lundin-Norway; and Knut
Bjrlykke at University of Oslo for valuable discussions and input on the
topics of AVO depth trends. Thanks to Lundin-Norway for providing data
used in this study.
Corresponding author: PAVS@StatoilHydro.com

JUNE 2008

THE LEADING EDGE

797

You might also like