Rock physics modeling constrained by sequence stratigraphy

TANIMA DUTTA, TAPAN MUKERJI, and GARY MAVKO, Stanford University, California, USA

We present a methodology to constrain rock physics mod-

els and their input parameters from principles of sequence
stratigraphy. The purpose is to demonstrate how we can
obtain critical sedimentological parameters and relative trends
of their spatial variation from sequence stratigraphic inter-
pretation. We present an example of obtaining relative trends
of sorting or variation of grain size in progradational lobes
of a submarine fan system from the Campos Basin, Brazil.
Sorting, a critical sedimentological parameter, can serve as a
constraint in rock physics modeling, thereby reducing uncer-
tainty in predicting reservoir properties from seismic veloc-
ities.
Sequence stratigraphy is the correct geologic interpreta-
tion of process/response events. It can predict the likely
occurrence of reservoir facies, source rocks, and seals.
Conventional stratigraphic interpretation from seismic data
has been predominantly qualitative and based on visual
inspection of geometric patterns in poststack seismic reflec- Figure 1. The workflow describing our approach to generate seismic
tion data. However, quantitative interpretation of seismic attributes by integrating sequence stratigraphy and rock physics.
attributes is possible if we can extract information about com-
positional maturity (mineralogy, clay content) and textural
maturity (sorting, grain angularity, sphericity, and rounded-
ness) using principles of sedimentology.
Quantitative seismic interpretation uses rock physics to
link seismic amplitude with reservoir properties like poros-
ity, clay-content, sorting, and diagenetic cements. In quanti-
tative seismic interpretation model parameters are calibrated
at well locations. However, one major source of uncertainty
in rock physics modeling arises due to our lack of knowledge
about trends of input parameters away from well data. This
can be reduced by constraining input parameters (for exam-
ple, compositional maturity and textural maturity) as guided
by the sequence stratigraphic framework. We believe that
future developments of reservoir property prediction from
seismic amplitudes could benefit from a close linkage between
sequence stratigraphy and rock physics.
In this article, we demonstrate how we can obtain criti-
cal sedimentological parameters and their spatial variations Figure 2. Vertical profile of sedimentological parameters from sequence
from sequence stratigraphic interpretation. We will then apply stratigraphy. Transgression and regression exhibit opposite trends
this technique to obtain sorting trends in progradation lobes (modified from Wagoner et al.).
deposited in a submarine fan, and use the trends to constrain
rock physics modeling and compute AVO attributes. The key
steps are summarized in Figure 1.

Step 1: Identify sedimentological properties and predictable

trends of their variations. Wagoner et al. (1990) showed that
the following sedimentological properties change predictably
during transgression and regression: sand-shale ratio, bed
thickness, grain size, sorting, and bioturbation.
Interestingly, the changes in these sediment properties
have opposite trends for transgression and regression (Figure
2). During regression, depositional energy tends to increase
upward resulting in increased bed thickness, higher net-to-
gross, better sorting, and decrease in bioturbation. On the
other hand, a marine transgression signifies a decrease in Figure 3. T1, T2, T3, and T4 are chronostratigraphic surfaces and
seismic reflectors usually follow chronostratigraphic surfaces. The
depositional energy and exhibits an opposite trend of the gradual change in color within each layer indicates gradual change in
above sediment parameters. We identify sand-shale ratio and sedimentological properties (adapted from Emery and Myers).
sorting as indicators of compositional maturity and textural
maturity, respectively. If the rock undergoes longer trans- compositionally more mature. In addition, the grains tend to
portation from the source area, usually the quartz/feldspar be better sorted, more rounded, and less angular. As a result,
ratio increases, clay content decreases, and the rock becomes textural maturity also increases with longer transportation his-

870 THE LEADING EDGE JULY 2007

Downloaded 18 Jan 2011 to 203.255.66.225. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
Figure 4. Seismic section showing geometry and truncation pattern of reflectors. The arrow indicates a submarine fan system.

tory of the sedimentary grains. The compositional and tex-

tural maturity constitute important sedimentological prop-
erties that affect the elastic stiffness of the rocks and in turn
affect seismic amplitudes.
These trends of sediment properties are more robust in
shallow marine or marginal marine environments since they
are deposited at or near sea level and are sensitive to changes
in relative sea level. The sediment delivery process and tec-
tonic setting in a deepwater system can be quite different from
the marginal marine environment. However, we can still ben-
efit by applying similar trends of sediment properties in deep-
water exploration. Core descriptions and thin sections, if
available, are useful to validate or modify the trends of sed-
imentological parameters.
Figure 5. Well logs with three parasequences (PS) within depositional
Step 2: Identify spatial gradients of sedimentological prop- lobes. The upper, middle, and lower PS are highlighted in red, blue, and
erties. Gradients of sediment parameters are not the same green, respectively. An increasing trend of porosity is observed from
lower-PS to upper-PS.
across the seismic reflector and along the reflector. Most seis-
mic reflectors and their amplitudes correspond to chrono-
stratigraphic surfaces. Multiple reflections, fluid contact, and Step 3: Rock physics analysis. Rock physics establishes the
diagenetic boundaries may generate seismic reflectors, but relationship between sedimentological properties and elas-
they lack chronostratigraphic significance. Chronostrati- tic moduli. After we determine the spatial trends of sediment
graphic surfaces represent depositional hiatus. Hiatus is parameters in a stratigraphic package, appropriate rock
defined as a gap in sedimentation record due to erosion or physics models are selected. The input parameters are guided
nondeposition. Changes in sedimentological properties are by our results from steps 1 and 2. The rock models are cali-
abrupt across the hiatus and gradual along the hiatus (Figure brated to well-log data. As output we obtain effective bulk
3). modulus, shear modulus, and density, and VP and VS as a
The vertical gradients of sediment parameters are cali- function of porosity.
brated from well data. The estimation of lateral gradients
requires multiple wells or horizontal wells. In the absence of Step 4: AVO modeling. AVO forward modeling (Shuey’s
such data, we assume that the lateral gradient of sediment approximation) is used to obtain intercept and gradient at key
parameters is linear. However, these trends are specific to stratigraphic interfaces using effective moduli predicted from
parasequence and parasequence sets, and parasequences are rock physics analysis in step 3. Trends in textural maturity
identified from sequence stratigraphic interpretation. Thus, are carried through from sequence stratigraphy to the AVO
using sequence stratigraphy, we obtain relative trends of vari- plane, via rock physics. Finally, the modeling results can be
ation in sediment properties within a depositional sequence. used to interpret amplitudes in terms of sedimentological
These trends then constrain the input parameters in rock properties and reservoir quality.
physics modeling.

JULY 2007 THE LEADING EDGE 871

Downloaded 18 Jan 2011 to 203.255.66.225. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
 Figure 6. Three parasequences (PS) interpreted within progradational depositional lobes. The changes in sorting are abrupt across the parasequence
boundaries (Trend A) and gradual along the boundaries (Trend B).

Example. We applied this concept in a deepwater turbidite

system from Campos Basin, offshore Brazil. The facies pack-
age was deposited during a lowstand system tract. Although
submarine fans can be deposited at any time, they are most
likely to be deposited during lowstand. Figure 4 shows the
geometry and truncation patterns of seismic reflectors in the
submarine fan. The stacking patterns in well logs are shown
in Figure 5. In this submarine fan system, unchannelized
depositional lobes show a porosity profile that increases
upward.
We consider three parasequences (PS) within the pro-
grading depositional lobes (Figure 6) and use sequence strati-
graphic principles to obtain the relative sorting trend, an
important parameter that controls porosity. Poor sorting
decreases porosity and increases the elastic moduli, and hence
seismic velocities.
We create the sorting trends as follows: A sequence strati-
graphic model from deepwater prograding depositional lobes Figure 7. Porosity decreases with a reduction in sorting. Each colored
(Walker, 1984) indicates that depositional energy increases ver- curve represents different grain sizes in sandstone (Beard and Weyl,
1973; Jorden and Campbell, 1984).
tically upward. We expect better sorting associated with
higher depositional energy. Therefore, we select lower PS to
be very poorly sorted, middle PS moderately sorted, and the
upper one well sorted (Trend A, Figure 6). The Trend A of
porosity at the proximal location is calibrated with well data.
The mean porosity computed from well logs in the upper,
middle, and lower PS is 26%, 23%, and 21% respectively. The
porosity at the proximal location in each PS is assigned by
this mean well-log porosity. The Piper (1978) model suggests
a slow process of segregation of silt from clay flocs during
transportation of turbidity current and thereby an improve-
ment in sorting in the direction of transport (Stow and Bowen,
1980). However, the exact functional form of the trend (i.e.,
linear, quadratic, cubic, or something else) is uncertain. In this
paper, we assume sorting improves linearly basinward within
each parasequence (Trend B, Figure 6). Additional data (such Figure 8. Porosity section in depositional lobes. At the proximal loca-
as multiple wells, horizontal wells or cores available at any tion porosities are obtained from well logs. The lateral variations of
parasequence) could have verified or modified these linear porosity away from the well are predicted by sequence stratigraphy.
872 THE LEADING EDGE JULY 2007

Downloaded 18 Jan 2011 to 203.255.66.225. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/

Figure 9. The scatter points show measurements of porosity and veloc-
ity from well logs in three parasequences of prograding lobes. The rock-
physics model explains the scatter in terms of sedimentological
parameters quantitatively. The friable sand model can be used to repre-
sent the trend of sorting or textural maturity. The solid circles repre-
sent mean porosities at proximal locations for each parasequence and
corresponding velocities obtained using the rock model.
sorting trends.
Relationship between sorting and porosity. We use an exper-
imental sorting-porosity relationship for artificially mixed
sand. Figure 7 shows the range of porosities for different sort-
ing and various sandstone grain sizes. Sorting is expressed
in terms of the standard deviation of the distribution of the
logarithm of the grain size. Figure 8 shows a plausible poros-
ity section of progradational lobes. At the proximal location,
the mean of porosities at each parasequence is calibrated
from well logs. The lateral variation of sorting away from the
well data is obtained from the trend predicted by sequence
stratigraphy. The mean porosity at distal locations is selected
using the mean porosity of medium sand-sized grains for dif-
ferent sorting. The mean porosity at the distal location in upper
PS (well sorted), middle PS (medium sorted), and lower PS
(very poorly sorted) is 40%, 34%, and 26%, respectively. These
values can be calibrated by porosities computed from devi-
ated wells or additional wells at distal location.
Selection of rock physics model and input parameters. The
trends of the variations in sorting, obtained from sequence
stratigraphy, guide us to select a rock model that accounts
for sorting. The friable-sand model (Dvorkin and Nur, 1996;
Mavko et al., 1998) can be used heuristically to compute elas-
Downloaded 18 Jan 2011 to 203.255.66.225. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
Figure 10. AVO modeling results. Intercept (R0) and Gradient (G)
crossplot color-coded by porosity. The data indicate a distinct trend of
variation in textural maturity from landward to basinward. The
“stars” indicate the position of the mean of R0 and G computed from
neighboring well logs at a proximal location. The ellipses around the
mean value emphasizes that there will be scatter when utilizing real
data.
tic moduli of both well sorted and poorly sorted sands. The
effective moduli at a well-sorted, high-porosity end member
(~40%) are computed using Hertz-Mindlin theory.
The Hertz-Mindlin contact theory provides the following
expressions for the bulk (KHM) and shear (GHM) moduli of a
dense random pack of identical spherical grains subject to an
effective pressure P:
(1)
(2)
where φ0 is the critical porosity (~40% for sandstone), C is the
coordination number (average number of grain contacts); G
and ν are the mineral shear modulus and Poisson’s ratio. The
elastic moduli at poorly sorted, zero porosity end member
are given by mineral moduli.
The effective elastic moduli of sand with porosities
between 0 to critical porosity (~40%) are interpolated using
lower Hashin-Shtrikman bound. The heuristic argument is
that adding small grains in the pore spaces adds minerals in
a way that keeps the effective elastic moduli low, very close
to the lower bound. The lower Hashin-Shtrikman bound is
theoretically the lowest elastic stiffness possible for a mix of
multiple phases.
The sandstones in prograding lobes of a submarine fan
in Campos Basin are compositionally immature. A relatively
lower quartz/feldspar ratio indicates compositional imma-
turity. Therefore, we include feldspar and clay along with
quartz in rock physics model.
Figure 9 shows porosities and velocities obtained from
three parasequences in a well log as shown in Figure 5. The
curves in Figure 9 represent predicted velocities using the fri-
able sand model. The effective pressure is obtained from the
well-log measurements and the coordination number is
assumed to be 8 for well-sorted end members. The trends are
in good agreement with the friable-sand model predictions.
The parallel lines in Figure 9 represent model predictions for
different compositional maturity or different quartz/feldspar
content. Higher compositional maturity increases the veloci-
JULY 2007 THE LEADING EDGE 873
 ties. The stiff-sand rock model is suitable for cemented sands
and does not capture the trends of these data. The friable sand
model with 70% quartz, 20% feldspar, and 10% clay along with
mean well-log porosity in each parasequence are used to com-
pute the effective moduli at the proximal location. To model
spatial variation of effective moduli, the porosity section from
the sequence stratigraphic interpretation (Figure 8) is used as
input to the calibrated rock model.
The effective VP, VS, and density are used to compute
intercept (R0) and gradient (G) at the interfaces of the three
parasequences using Shuey’s two-term approximation. They
indicate distinct trends in R0-G plane (Figure 10). These trends
can serve as a template to guide the interpretation of observed
intercept and gradient away from the well data. Since these
trends incorporate the information from sequence stratigra-
phy, they can be used to predict the spatial variation in reser-
voir properties. We can now make quantitative interpretations
about porosity and sorting based on the calibrated rock model
in addition to the qualitative trend interpretation.
Discussion and conclusions. We have developed a method
of using spatial trends of sedimentological parameters from
sequence stratigraphic interpretations as constraints in rock
physics modeling away from well data. These trends are rel-
ative trends within a stratigraphic sequence. We have applied
this method to determine trends of sorting or grain-size vari-
ation in prograding lobes of a submarine fan in the Campos
Basin, Brazil. However, there is uncertainty in our linear sort-
ing trend since lobe-switching in the submarine fan will com-
plicate the predictive power of the sequence stratigraphic
model. Further developments to this workflow will include
(1) consideration of multiple sequence stratigraphic models
and (2) calibrating the trends of sediment parameters from mul-
tiple well logs, core descriptions, thin sections, and grain-size
analysis data, and (3) including uncertainty in the trends.
Rock physics modeling was constrained using trends of
grain sorting as guided by sequence stratigraphic interpreta-
tions. In addition, the sands in submarine fan in Campos Basin
are compositionally less mature. Therefore feldspar and clay
were included in the rock model to compute the effective elas-
tic moduli as a function of sorting and porosity.
The effective moduli computed from rock physics mod-
eling were used to generate AVO attributes at two parase-
quence boundaries. In the R0-G plane, we obtain a linear trend
of variation in sediment properties from landward to basin-
ward. The trend approximates the mean of probability distri-
bution of sedimentological properties. Geologic variability
will add scatter around this trend. Textural maturity pro-
gressively increases along this trend basinward. Furthermore,
since these trends are calibrated with the rock physics model
it is possible to make quantitative interpretations about poros-
ity and sorting. For example mean porosity progressively
changes from ~24% to ~34% and ~22% to ~27% along the
upper and lower interfaces, respectively.
The workflow developed in this paper allows us to con-
strain rock physics modeling, away from well data, using spa-
tial trends of sedimentological parameters from sequence
stratigraphic interpretations. The direct advantage of inte-
grating sequence stratigraphy and rock physics is that seis-
mic attributes (e.g., intercept, gradient) can be quantitatively
interpreted in terms of underlying sedimentological proper-
ties.
Suggested reading. Quantitative Seismic Interpretation: Applying
874 THE LEADING EDGE JULY 2007
Downloaded 18 Jan 2011 to 203.255.66.225. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
Rock Physics Tools to Reduce Interpretation Risk by Avseth et al.
(Cambridge, 2005). “Influence of texture on porosity and perme-
ability of unconsolidated sand” by Beard and Weyl (AAPG
Bulletin, 1973). “Recent advances in seismic lithologic analysis”
by Castagna (GEOPHYSICS, 2001). “Elasticity of high-porosity sand-
stones: Theory for two North Sea data sets” by Dvorkin and Nur
(GEOPHYSICS, 1996). “Pressure-solution and the rock physics dia-
genetic trend in quartzose sandstones” by Florez-Niño and Mavko
(SEG 2004 Expanded Abstracts). Sequence Stratigraphy by Emery and
Myers (Blackwell, 1996). “Stratigraphy-guided rock physics” by
Gutiérrez et al. (TLE, 2002). The Rock Physics Handbook by Mavko
et al. (Cambridge, 1998). “Compliance of elastic bodies in con-
tact” by Mindlin (Journal of Applied Mechanics, 1949). “Sequence
stratigraphy: Basic elements, concepts and terminology” by
Mulholland (TLE, 1998). “Turbidite muds and silts on deepsea
fans and abyssal plains” by Piper (in Sedimentation in Submarine
Canyons, Fans and Trenches, 1978). “Sequence stratigraphy—a local
theory with global success” by Neal and Vail (Oilfield Review, 1993).
“Seismic stratigraphy—applications to hydrocarbon explorations”
by Playton (in AAPG Memoir 26, 1977). “Seismic-stratigraphic
study of the Oligocene-Miocene shelf-fed turbidite systems of the
Campos Basin, Brazil” by Peres (PhD thesis, the University of Texas
at Austin, 1990). “A physical model for the transport and sorting
of fine-grained sediment by turbidity currents” by Stow and
Bowen (Sedimentology, 1980). “Do seismic reflections necessarily
have chronostratigraphic significance?” by Tipper (Geological
Magazine, 1993). Siliciclastic Sequence Stratigraphy in Well Logs,
Cores, and Outcrops by Wagoner et al. (AAPG Methods in
Exploration Series, 1990). “Turbidites and associated coarse clas-
tic deposits” by Walker (in Facies models, Geoscience Canada, Reprint
Series-1, 1984). “Facies mapping from three dimensional seismic
data: potential and guidelines from a tertiary sandstone-shale
sequence model, Powderhorn field, Calhoun Country, Texas” by
Zeng et al. (AAPG Bulletin, 1996). Well Logging in Rock Properties,
Borehole Environment, Mud and Temperature Logging by Jorden and
Campbell (SPE and AIME, 1984). TLE
Acknowledgments: This work was supported by the Stanford Rock Physics
and Borehole Geophysics (SRB) project and by DOE awards DE-FC26-
04NT15506 and DE-FG02-03ER15423. We acknowledge Norsk Hydro and
Fugro for permission to use the data.
Corresponding author: tanima@pangea.stanford.edu