GEOPHYSICS, VOL. 66, NO. 2 (MARCH-APRIL 2001); P. 398–412, 9 FIGS., 5 TABLES.

In honor of the new millennium (Y2K), the SEG Research Committee is inviting a series of review articles and tutorials that
summarize the state-of-the-art in various areas of exploration geophysics. The present article by Dr. Zhijing (Zee) Wang is the
first of this series, and deals with the rapidly evolving field of rock physics. Further invited contributions will appear during the
next year or two—Sven Treitel, Chairman, SEG Research Subcommittee on Y2K Tutorials and Review Articles.

Y2K Tutorial

Fundamentals of seismic rock physics

Zhijing (Zee) Wang∗

INTRODUCTION every experimental finding in this paper. The objective of this

tutorial paper, as opposed to a review paper, is not to present
During the past 50 years or so, tremendous progress has
a comprehensive review on rock physics; instead, I attempt to
been made in studying physical properties of rocks and miner-
discuss some of the most important, yet practical, theories and
als in relation to seismic exploration and earthquake seismol-
major factors that influence seismic properties of sedimentary
ogy. During this period, many theories have been developed
rocks. As such, most of the discussions are qualitative and
and many experiments have been carried out. Some of these
descriptive.
theories and experimental results have played important roles
This paper only covers a fraction of existing knowledge in
in advancing earth sciences and exploration technologies. This
rock physics and is not intended for rock physics experts, The
tutorial paper attempts to summarize some of these results.
targeted readers are students and potential users as an intro-
In exploration seismology, seismic waves bring out subsur-
duction to seismic rock physics. For those who are interested in
face rock and fluid information in the form of travel time, re-
in-depth knowledge of rock physics, see Bourbie et al. (1987),
flection amplitude, and phase variations. During the early years
Nur and Wang (1989), Wang and Nur (1992, 2000), and Mavko
of exploration seismology, seismic data were interpreted pri-
et al. (1998). For those who are interested in other physical
marily for structures that might trap hydrocarbons. With the
properties of rocks, see Gueguen and Palciauskas (1994) and
advancement of computing power and seismic processing and
Schön (1995).
interpretation techniques, seismic data are now commonly an-
This tutorial consists of three parts: fluid substitution, factors
alyzed for determining lithology, porosity, pore fluids, and sat-
influencing seismic properties, and rock physics challenges. In
uration. Because rock physics bridges seismic data and reser-
the first part, I present the most commonly used tool for fluid
voir properties and parameters, it has been instrumental in
substitution (the Gassmann equation) and discuss its applica-
recent years in the development of technologies such as 4-D
bility and assumptions. It is naturally followed by discussions
seismic reservoir monitoring, seismic lithology discrimination,
on tools for obtaining pore fluid properties because one of the
and direct hydrocarbon detection with “bright-spot” and angle-
most important input parameters to the Gassmann equation
dependent reflectivity analyses.
is fluid properties. The second part discusses effects of various
Seismic properties are affected in complex ways by many
rock and reservoir parameters on seismic properties and tab-
factors, such as pressure, temperature, saturation, fluid type,
ulates some rules of thumb of rock physics. Finally, I briefly
porosity, pore type, etc. These factors are often interrelated
discuss several issues and challenges in applying rock physics.
or coupled in a way that many also change when one fac-
tor changes. The effect of these changes on seismic data can FLUID SUBSTITUTION: THE GASSMANN EQUATION
be either additive or subtractive. As a result, investigation of
the effect of varying a single parameter while fixing others be- The Gassmann (1951) equation has been used for calculat-
comes imperative in understanding rock physics applications to ing the effect of fluid substitution on seismic properties using
seismic interpretations. Table 1 shows some factors influencing the frame properties. It calculates the bulk modulus of a fluid-
seismic properties of rocks. These factors will be summarized saturated porous medium using the known bulk moduli of the
and elaborated in a later section. solid matrix, the frame, and the pore fluid. For a rock, the solid
Because of the vast amount of information in the literature matrix consists of the rock-forming minerals, the frame refers
on rock physics, it is impossible to summarize every theory and to the skeleton rock sample, and the pore fluid can be a gas,

Manuscript received by the Editor March 14, 2000; revised manuscript received July 24, 2000.
∗
Chevron Petroleum Technology Company, Room D1124, Building D, P.O. Box 6019, San Ramon, California 94583-0719. E-mail: zwan@
chevron.com.
°c 2001 Society of Exploration Geophysicists. All rights reserved.

398
 Fundamentals of Rock Physics 399

oil, water, or a mixture of all three: The bulk density ρ f of the fluid mixture is calculated by
(1 − K d /K m )2 ρ f = Sw ρw + So ρo + Sg ρg ,
K ∗ = Kd + , (1) (7)
φ 1−φ Kd
+ − 2 where ρw , ρo , and ρg are the bulk densities of water, oil, and
Kf Km Km
gas, respectively.
where K ∗ is the bulk modulus of a rock saturated with a fluid of The basic assumptions for the Gassmann equation are that
bulk modulus K f , K d is the frame bulk modulus, K m is the ma-
trix (grain) bulk modulus, and φ is porosity. The shear modulus 1) The rock (both the matrix and the frame) is macroscopi-
G ∗ of the rock is not affected by fluid saturation, so that cally homogeneous.
G∗ = Gd , (2) 2) All the pores are interconnected or communicating.
3) The pores are filled with a frictionless fluid (liquid, gas,
where G d is the frame shear modulus of the rock. The density or mixture).
ρ ∗ of the saturated rock is simply given by 4) The rock-fluid system under study is closed (undrained).
5) The pore fluid does not interact with the solid in a way
ρ ∗ = ρd + φρ f . (3)
that would soften or harden the frame.
where ρ ∗ and ρd are the fluid-saturated and dry densities of the
rock, respectively, and ρ f is the pore fluid’s density. Note that Assumption (1) is common to many theories of wave prop-
ρd = (1 − φ)ρm , where ρm is the matrix (grain) density. agation in porous media. It assures that the wavelength is long
The frame bulk and shear moduli are calculated using the compared to the grain and pore sizes. For most rocks, waves
measured velocities in the frame with frequencies ranging from seismic to laboratory frequen-
µ rock: ¶ cies can generally meet this assumption. Brown and Korringa
4
K d = ρd V p2 − Vs2 , (4) (1975) extended Gassmann’s equation to anisotropic rocks.
3
Assumption (2) implies that the porosity and permeability
G d = ρd Vs2 . (5) are high and there are no isolated or poorly connected pores
in the rock. The purpose of this assumption is to ensure full
It is important to point out that the frame moduli are not the equilibrium of the pore fluid flow, induced by the passing wave,
same as the dry moduli. In the correct use of the Gassmann within the time frame of half a wave period. The pore intercon-
equation, frame moduli should be measured at irreducible sat- nectivity is therefore relative to the wavelength or frequency.
uration conditions of the wetting fluid (normally water). The For the Gassmann equation, which assumes an infinite wave-
irreducible fluid is part of the rock’s frame, not the pore space. length (zero wave frequency), most rocks can meet this as-
Overdrying a rock in the laboratory will result in erroneous sumption no matter how poorly the pores are interconnected.
Gassmann results. For seismic waves, however, only unconsolidated sands can ap-
The bulk modulus K f of a fluid mixture can be calculated proximately meet this assumption because of the sands’ high
using Wood’s equation (Wood, 1941): porosity and permeability (Wang, 2000a). For high-frequency
1 Sw So Sg waves such as those used in logging and in the laboratory, most
= + + , (6)
Kf Kw Ko Kg rocks may not meet this assumption. Consequently, the log- or
lab-measured velocities are often higher than those calculated
where K w , K o , and K g are the bulk moduli of water, oil, and with the Gassmann equation.
gas, respectively; Sw , So , and Sg are the water, oil, and gas satu- Assumption (3) implies that the viscosity of the saturating
rations, respectively, expressed as volume fractions of the pore fluid is zero. The purpose of this assumption is again to ensure
space; and Sw + So + Sg = 1. Equation (6) implies that the pore full equilibrium of the pore fluid flow. This assumption is also
fluid is uniformly distributed in the pores. relative to the wavelength or frequency. If the wave frequency
is zero, fluids with any viscosity will equilibrate within the time
Table 1. Factors influencing seismic properties of sedimen- frame of half a wavelength (infinite time). If the viscosity is
tary rocks (with increasing importance from top to bottom). zero, the pore fluid will be easy to equilibrate. In reality, be-
cause all fluids have finite viscosities and all waves have finite
Rock properties Fluid properties Environment wavelengths, most calculations using the Gassmann equation
Compaction Viscosity Frequency will violate this assumption.
Consolidation Density Stress history Assumptions (2) and (3) are the key points, and constitute
history the essence of the Gassmann equation. They imply that the
Age Wettability Depositional wave frequency is zero. This is perhaps the reason why the mea-
environment
sured laboratory and logging bulk modulus or velocity are usu-
Cementation Fluid Temperature
composition ally higher than those calculated with the Gassmann equation.
Texture Phase Reservoir process At finite frequencies, relative motion between the solid matrix
Bulk density Fluid type Production history and the pore fluid will occur, so that the waves are dispersive.
Clay content Gas-oil, Layer geometry The relative motion between the pore fluid and the rock ma-
gas-water ratio trix is caused by the finite wavelength and the high contrast
Anisotropy Saturation Net reservoir pressure
Fractures in bulk and shear moduli between the pore fluid and the rock
Porosity matrix.
Lithology Assumption (4) means that for a laboratory rock sample, the
Pore shape rock-fluid system is sealed at the boundaries so that no fluid
400 Wang

can flow in or out of the rock’s surface. For a rock volume v, and M R is the Reuss (1929) average,
which is part of a much larger volume Vo (such as a forma-
1 Xn
ci
tion in a reservoir), the system v must be located within Vo at = , (11)
such a distance from the surface of Vo that the stress variations MR i=1
Mi
generated by the passing wave do not cause any appreciable
flow through the surface of v. This is the key in calculating where ci and Mi are the volume fraction and the modulus of the
the effect of pore fluid change on seismic properties with the ith component, respectively. The effective medium is macro-
Gassmann equation, because if the system is open, changes in scopically isotropic. Hashin and Shtrikman (1963) provide a
seismic properties due to pore fluid changes will only be related more tightly bounded but more complicated estimate of M.
to the fluid density change. For practical purposes, when the elastic constants of the com-
Assumption (5) eliminates any effects of chemical/physical ponents do not differ widely, the VRH and Hashin-Shtrikman
interactions between the rock matrix and the pore fluid. In re- models yield similar results.
ality, the pore fluid will inevitably interact with the rock’s solid Wang (2000a) carried out an extensive comparison be-
matrix to change the surface energy. When a rock is saturated tween Gassmann predicted results and laboratory data. He
by a fluid, the fluid may either soften or harden the matrix. For shows that for rocks with interconnected high aspect-ratio
example, when loose sand grains are mixed with a heavy oil, (≈1) pores such as unconsolidated clean sands and sandstones
the mixture will have higher bulk and shear moduli. When a at high effective pressures, little difference exists between
shaley sandstone is saturated with fresh water, the rock’s ma- Gassmann-calculated and laboratory-measured seismic veloc-
trix is often softened due to clay swelling. An extreme case is ities (The aspect ratio of a two-dimensional pore is defined
that a dried mud (clay-water mixture) has higher elastic mod- as the short axis divided by the long axis. It is always less
uli than the water-saturated clay. This is partly why one should or equal to one). For rocks with very low aspect-ratio (¿1)
never overdry shaley rocks in the lab. This also emphasizes that pores such as poor grain contacts, cracks, or fractures, veloc-
the input “dry” frame bulk modulus to the Gassmann equation ities measured at seismic frequencies may be closer to the
should be obtained at irreducible fluid saturation conditions. laboratory-measured values than to the Gassmann-calculated
As seen in equations (1)–(7), the Gassmann equation re- values. For these rocks, significant wave dispersion may occur
quires several input parameters to calculate fluid effects on at very low frequencies, especially when these rocks are satu-
seismic velocities. The dry frame bulk and shear moduli, poros- rated with high-viscosity pore fluids, so that both the seismic
ity, grain density, and the fluid bulk modulus (incompressibility) and laboratory frequencies are in the same “high” frequency
are mostly measured in the lab. If lab data are not available, band.
they can often be measured or estimated through well logs For rocks with medium to low aspect-ratio pores, veloc-
or empirical relations. For example, porosity can be derived ities measured at seismic frequencies are higher than the
from neutron or acoustic logs. Dry frame moduli can be either Gassmann-calculated values but lower than the laboratory-
estimated using the backward Gassmann equation if other in- measured values. However, the real magnitudes of the devi-
put parameters are known or derivable from acoustic log data. ation of the seismic data from either the Gassmann equation
Such information can come from gas- and oil/water-saturated predictions or from the laboratory-measured data will remain
sections in the same well, assuming that lithology does not vary. unknown until the Gassmann equation can be rigorously veri-
However, the log-derived parameters are usually inaccurate as fied in the laboratory or in the field.
they are indirect measurements and affected by hole condi- Wang (2000a) also compared the Gassmann results and lab-
tions, saturation, and lithology variations. oratory results of the effect of fluid displacement on seis-
From equation (1), the backward Gassmann equation can mic properties. The effects of fluid displacements on seismic
be derived as velocities agree well between the Gassmann-predicted and
µ ¶ laboratory-measured values, provided that the frame proper-
φ Km
∗ ties provided to the Gassmann equation (K d , G d , and ρd ) are
K + 1 − φ − Km
Kf measured at the irreducible water saturation or under moist
Kd = . (8)
φ Km K∗ conditions. Figure 1 shows the effect of fluid displacement
+ −1−φ (water and CO2 flood) on compressional velocities (V p ) in
Kf Km
sands, sandstones, and dolomites. The Gassmann-calculated V p
change as a result of fluid displacement is very close to what
The input grain (matrix) bulk and shear moduli are from the
is measured in the laboratory (Figure 1a). For shear velocities
moduli of the minerals that comprise the rock. If mineralogy is
(Figure 1b), the Gassmann equation predicts slightly greater
known for the rock, one can use the Voight-Reuss-Hill (VRH)
effects of fluid displacement compared to those measured in
average (Hill, 1952) to calculate an effective K m and G m :
the laboratory.
The results shown in Figure 1 suggest that the fluid displace-
1
M= (MV + M R ) (9) ment process, though not necessarily the rock, approximately
2 conforms with the Gassmann assumptions as only the fluids
in large, well connected pores are likely to be swept. As a re-
where M is the effective grain modulus (which can be either sult, when the input frame properties are measured at the ir-
K m or G m ), MV is the Voigt (1928) average, reducible water saturation condition, the Gassmann-predicted
X
n and the laboratory-measured effects of fluid displacements on
MV = ci Mi , (10) seismic properties might be directly applied to 4-D seismic fea-
i=1 sibility studies and interpretations.
 Fundamentals of Rock Physics 401

Biot (1956a, b, 1962) extended the Gassmann equation to the are therefore needed for modeling and interpreting seismic
full frequency range. At zero frequency, Biot’s theory reduces data. Vargaftik (1975) presents some data on CO2 properties
to the Gassmann equation. At an infinite frequency, Biot’s the- as functions of pressure and temperature.
ory can be described by a set of analytical equations (Geertsma
and Smit, 1961). However, Biot’s theory predicts very little ve- Hydrocarbon oils
locity dependence on frequency because the calculated velocity
difference between zero and infinite frequencies is usually less Both the bulk modulus and density of hydrocarbon oils in-
than 3% for most reservoir rocks (Winkler, 1985; Wang and crease with pressure, but decrease with increasing temperature.
Nur, 1990a). As a result, the Gassmann equation, rather than Batzle and Wang (1992) presented a set of empirical equations
Biot’s full-frequency theory, is often used in fluid substitution for calculating hydrocarbon oil properties as functions of pres-
analysis. sure and temperature:

ρo + (0.00277P − 1.71 × 10−7 P 3 )(ρo − 1.15)2 + 3.49 × 10−4 P

ρ= , (12)
0.972 + 3.81 × 10−4 (T + 17.78)1.175

Many other models exist in addition to Gassmann’s and µ ¶1/2

ρo
Biot’s theories. For a comprehensive review, see Wang and Nur V = 2096 − 3.7T + 4.64P
(1992) and Mavko et al. (1998). 2.6 − ρo
"µ ¶1/2 #
18.33
+ 0.0115 − 16.97 − 1 TP, (13)
PORE FLUID PROPERTIES ρo
One of the ultimate goals of exploration seismology is to
delineate and map pore fluids and their type and distribution
in petroleum reservoirs. To achieve this goal, seismic data has to
be able to resolve the compressibility contrasts among different
pore fluids in the reservoir. As a result, seismic properties of
pore fluids have to be fully understood.
Batzle and Wang (1992) summarized seismic properties of
commonly encountered pore fluids based on their own data
as well as on the literature. Their correlations are now being
widely used in the oil industry.

Hydrocarbon gases
Batzle and Wang (1992) present a set of equations for cal-
culating the bulk modulus and density of hydrocarbon gases.
Because most gases are extremely compressible under reser-
voir conditions, in many cases the bulk modulus (incompress-
ibility) of a hydrocarbon gas can be set as 0.01 to 0.2 GPa in
seismic modeling. Errors in gas bulk modulus will yield lit-
tle uncertainty in the calculated seismic properties in a fluid-
saturated rock. An excellent source for thermophysical proper-
ties of methane, ethane, and butane can be found in Younglove
and Ely (1987) where both acoustic velocity and bulk density
are given along with other physical parameters.

Carbon dioxide (CO2 )

CO2 is often injected into hydrocarbon reservoirs to dis-
place oil. The displacement process can be either miscible or
immiscible depending on pressure and temperature. Further-
more, many gas reservoirs produce substantial amount of CO2
that has to be disposed of. One disposal process is injecting it
into geological formations. Seismic technologies can be used to FIG. 1. Gassmann-calculated versus lab-measured effects of
fluid displacement (water and CO2 flood) on compressional
monitor the injected CO2 and its displacement process (Langan (a) and shear (b) velocities in sands, sandstones, and dolomites.
et al., 1997) only if the injected CO2 causes sufficient changes Solid circles are waterflooded sandstones and sands, triangles
in seismic properties of the formation rock. CO2 properties are CO2 -flooded dolomites.
402 Wang

Once the velocity and bulk density have been calculated, the
K = ρV 2 , (14) bulk modulus can be obtained from equation (13).

where ρ, V , and K are the bulk density (g/cm3 ), velocity (m/s), Gas-in-water solutions and mixtures
and bulk modulus (kPa), respectively, at pressure P(MPa)
and temperature T (◦ C); ρo is a reference density measured Dodson and Standing (1945) found that for small amounts
at 15.6◦ C and one atmosphere. of dissolved gases, the compressibility of water increases al-
Equations (12) and (13) are valid for dead oils (oils without most linearly with the gas-to-water ratio in water containing
dissolved gases) only. For live oils (oils with dissolved gases), dissolved gases:
one needs to substitute the saturation density ρG for ρo in equa-
tion (12) and the pseudo density ρ 0 for ρo in equation (13): βG = β B (1 + 0.0494RG ), (21)
ρo + 0.0012RG G where βG is the compressibility of water with dissolved gases,
ρG = , (15)
Bo β B is the compressibility of gas-free water, and RG is the gas-to-
ρo water ratio in liter/liter. Sergeev (1948) also reported substan-
ρ0 = , (16) tial compressibility increases as gases are dissolved in brine.
(1 + 0.001RG )Bo
For many years, equation (21) has been used for estimat-
where ing water and brine compressibilities as functions of dissolved
" µ ¶1/2 #1.175
G gases. However, this equation predicts too much dependence of
Bo = 0.972+0.00038 2.495RG + T + 17.8 water compressibility on the amount of dissolved gases. For ex-
ρo ample, a gas-to-water ratio of 10 liter/liter (56.2 ft3 /bbl) would
(17) increase the compressibility by 50%!
is called the formation volume factor and G is gas specific grav- Osif (1988) studied the effects of salt, gas, temperature, and
ity (relative to air). RG is gas-to-oil ratio (GOR) in liter/liter pressure on the compressibility of water. Contrary to Dodson
(l liter/liter = 5.615 ft3 /bbl). and Standing (1945), Osif’s results show little to no effect of
gas in solution on compressibilities of water and brine.
Water and brine Osif’s findings were confirmed by Liu (1998), who studied the
effect of dissolved methane, nitrogen gas, CO2 , and ammonia
Water is a fluid that possesses several abnormal properties on acoustic velocities in water. His results show that the dissolu-
due in part to the electronic polarity of the water molecule: tion of methane and nitrogen gas in water has little effect on the
(1) Under atmospheric pressure, water’s density reaches a max- acoustic velocities. The dissolution of CO2 and NH3 increases
imum at 4◦ C, (2) for its molecular weight, water has abnormally the acoustic velocities in water substantially. In CO2 -water so-
high boiling point and melting points, and (3) unlike hydrocar- lution, acoustic velocity increases by about 50 m/s (about 3%)
bon liquids in which acoustic velocities decrease monotoni- as CO2 is dissolved into water at room temperature and about
cally with increasing temperature, acoustic velocity in water 400 psi. In NH3 -water solution, acoustic velocity increases by
increases with increasing temperature, reaches a maximum at about 150 m/s (about 10%) as NH3 is dissolved into water at
around 73◦ C, and then decreases as temperature further in- room temperature and about 75 psi. Such magnitude increases
creases at atmospheric pressure. Acoustic velocities and bulk are so substantial that they cannot be neglected in practice.
densities in gas-free water and brine can be calculated using
the equations summarized in Batzle and Wang (1992):
Steam
−6
ρw = 1.0 + 10 (−80T − 3.3T + 0.00175T + 489P
2 3
Acoustic properties of steam and hot water are important for
− 2TP + 0.016T 2 P seismic modeling and interpretation of 4-D seismic monitoring
data. The American Society of Mechanical Engineers (ASME)
− 1.3 × 10−5 T 3 P − 0.333P 2 − 0.002T P 2 ), (18) published tables of specific volumes of water and steam versus
−6 pressure and temperature (Meyer et al., 1979). The specific
ρb = ρw + 0.668S + 0.44S + 10 S[300P − 2400PS
2
volume data can be converted to isothermal bulk modulus data
+ T (80 + 3T − 3300S − 13P + 47PS)], (19) through
X4 X 3 ∂P
Vw = wi j T i P j , K T = −v , (22)
∂v
i=0 j=0
where v is specific volume, P is pressure in MPa, and K T is the
Vb = Vw + S(1170 − 9.6T + 0.055T 2 − 8.5 × 10−5 T 3 isothermal bulk modulus in MPa.
Because acoustic velocities are governed by the adiabatic
+ 2.6P − 0.0029T P − 0.0476P 2 ) + S 1.5 (780 − 10P
bulk modulus and bulk density, the isothermal bulk modu-
+ 0.16P 2 ) − 1820S 2 , (20) lus K T has to be converted to the adiabatic bulk modulus K
through
where ρw and ρb are the bulk density of water and brine, re- CP
spectively; Vw and Vb are the velocities of water and brine, K = KT , (23)
respectively; S is salinity in weight fraction, T is temperature
Cv
in ◦ C, and P is pressure in MPa. The constants wi j can be found where C p and Cv are the specific heats at constant pressure
in Batzle and Wang (1992). and volume, respectively. In the laboratory, C p and Cv cannot
 Fundamentals of Rock Physics 403

be obtained easily and accurately, so we assume a constant C p pendence on pore fluid type and saturation makes 4-D seismol-
over Cv ratio of 1.5 in our calculations. ogy a powerful tool for reservoir fluid mapping and monitoring.
Figure 2a shows the adiabatic bulk modulus of water and Although good pore fluid compressibility contrasts are nec-
steam versus temperature and pressure. The bulk modulus essary for 4-D and AVO, they may not be sufficient because
drops sharply as water transforms to steam. Figure 2b shows seismic reflectivity is dependent on both fluid properties and
the bulk density of water and steam versus temperature and rock properties. Fluid effects on rocks with stiff pores (hard
pressure. Bulk density also drops sharply as water transforms rocks) are always smaller than those on rocks with compli-
to steam. The bulk modulus in water increases with tempera- ant pores (soft rocks). Figure 3 shows the P-wave impedance
ture up to about 73◦ C. It then decreases as temperature further change as a result of water flood in 44 sandstone and sand sam-
increases. ples versus porosity. Low porosity rocks, which tend to have
For properties of fluids not discussed here (e.g., drilling hard elastic frames, show smaller waterflood effect on seismic
muds), see Wang and Nur (2000). impedances. However, fluid saturation effects are controlled
by the compressibility of the pores (or pore shape), not neces-
FLUIDS AND SATURATION sarily all by porosity. Rocks with cracks or fractures, no matter
One of the key factors in 4-D seismic reservoir monitoring how low the porosity may be, always show a large effect of
and angle-dependent reflectivity [amplitude variation with off- fluid saturation on seismic velocity. This is because cracks and
set (AVO)] is the seismic response to fluid saturation and sub- fractures are very compliant.
stitution. Rocks saturated with less compressible fluids show
higher P-wave velocities and impedances. S-wave velocities POROSITY, PORE SHAPE, AND CLAY CONTENT
and impedances are much less affected by pore fluids because
Velocity and impedance of rocks decrease with increasing
fluids have no rigidity. Consequently, rocks with less compress-
porosity. Such velocity- or impedance-porosity relationships,
ible fluids have higher velocity ratios (V p /Vs ). Such V p and
however, are only statistically valid because seismic properties
V p /Vs dependence on pore fluid type is the physical basis for
of a rock are more affected by the pore shape than by porosity
use of angle-dependent reflectivity for direct hydrocarbon de-
(e.g., Kuster and Toksoz, 1974). For example, a low-porosity
tection and fluid delineation. The velocity and impedance de-
rock with flat, low-aspect ratio pores may have lower seismic
velocities than a high-porosity rock with spherical, high-aspect
ratio pores because flat pores are much more compressible
than spherical pores. As a result, the scatter in a velocity- or
impedance-porosity relationship can be partly attributed to dif-
ferences in pore shapes among rock samples. However, pore
shape varies in wide ranges in sedimentary rocks and is hard to
quantify. In practice, statistical relationships, along with stan-
dard deviation, between porosity and seismic properties need
to be established for each facies in a reservoir.
Many reservoir sands and sandstones contain clays. The ef-
fect of clays on seismic properties depends further on the posi-
tion of the clay particles in the rock and on the clay type. If the
clays are part of the rock matrix and if the clays are more com-
pressible than quartz, velocity and impedance will decrease

FIG. 3. P-wave impedance change as a result of water flood in

FIG. 2. Adiabatic bulk modulus (a) and density (b) of water and 44 sandstone and sand samples versus porosity. Low porosity
steam versus temperature and pressure. Both bulk modulus rocks, which tend to have hard elastic frames, show smaller
and density drop sharply as water transforms to steam. effect of water flood on seismic impedances.
404 Wang

as clay content increases. Except for the density effect, pore- result, knowing the reservoir pressure regime is very impor-
filling clays have little influence on seismic properties, unless tant in seismic applications such as 4-D and AVO. For exam-
the pores are completely filled. Tosaya and Nur (1982) were ple, Figure 4 shows the V p versus net overburden pressure in a
among the first to study the combined effects of porosity and sandstone. Increasing the net overburden pressure from 1650
clay content on seismic velocities. Han et al. (1986) extended to 2650 psi causes a 5.2% V p increase, whereas the same amount
this study and developed a set of empirical relations between of net overburden pressure increase from 4500 to 5500 psi only
velocity, porosity, and clay content that were based on labora- results in a 0.5% V p increase.
tory measurements of about 80 sandstone samples. The rela- During production and enhanced oil recovery (EOR) pro-
tions take the linear forms cesses, both reservoir pressure and fluid saturation change.
Pressure and saturation effects may reinforce or compete with
V p = V po − a1 φ − a2 C (24) each other. The resulting changes in seismic properties (ve-
locities and impedances) depend on the combined effects of
and
pressure and saturation changes. For example, in a water-drive
Vs = Vso − b1 φ − b2 C, (25) process, water displaces oil so that oil saturation decreases. In
the meantime, reservoir pressure may drop due to production,
where φ and C are porosity and clay content in volume fraction, resulting in an increase in the net overburden pressure. Both
respectively. V p and Vs are compressional and shear velocities decreases in oil saturation and reservoir pressure will increase
in km/s. The regression constants are functions of the net over- the P-wave velocities and impedances so that the two effects
burden pressure (Table 2). reinforce each other. In contrast, in a water-injection process,
Equations (24) and (25) clearly show that both V p and Vs the injected water displaces oil so that oil saturation decreases
statistically decrease with increasing porosity and clay content. (water saturation increases), but the reservoir pressure is typi-
These relations take no account of the position of clay particles cally increased by water injection, resulting in a decrease in the
in the rock. They are only empirical, not necessarily physical. net overburden pressure. Lower oil saturation would increase,
Eberhart-Phillips et al. (1989) further extended Han et al.’s but higher reservoir pressure would decrease the P-wave ve-
(1986) regressions to include pressure. Castagna et al. (1985) locities and impedances. As a result, saturation and pressure
reported relationships similar to equations (24) and (25), which compete with each other in this case, partially, or even totally,
they found from log data. canceling each other’s effect on seismic properties. These two
examples illustrate the importance of understanding reservoir
PRESSURE process and rock property changes in planning and timing 4-D
seismic surveys.
In a reservoir, there always exist two distinct pressures: over-
burden pressure and reservoir pressure. The overburden pres- TEMPERATURE
sure (Po ), also called confining pressure, is the pressure exerted
by the total overburden rock strata, whereas the reservoir pres- Seismic velocities and impedances decrease only slightly in
sure (Pp ), also called fluid pressure or pore pressure, is exerted rocks saturated with gas or water as temperature increases
by the fluid mass. The difference between the overburden pres- (Timur, 1977; Wang and Nur, 1990b). When a rock is satu-
sure and reservoir pressure is called the net overburden pres- rated with oil, however, seismic properties may decrease by
sure (Pd ), which is also called differential pressure or some- large amounts with increasing temperature, especially in un-
times effective pressure (Pe ). Strictly speaking, Pe 6= Pd . In consolidated sands with heavy oils. Such velocity dependence
fact, Pd = Po − Pp , whereas Pe = Po − n Pp , where n ≤ 1. It is the
net overburden pressure that controls the seismic properties of
reservoir rocks. This is because the pore fluid pressure supports
a part of the overburden pressure, thereby decreasing the load
supported by the total rock strata.
Seismic velocities and impedances, both P and S, increase
as the net overburden pressure increases. The relationships
between seismic properties and the net overburden pres-
sure, however, is nonlinear: seismic properties increase faster
(higher slope) in low net overburden pressure regions. As a

Table 2. Regression constants for equations (24) and (25).

Net pressure V po a1 a2 Vso b1 b2
Water saturated
40 MPa 5.59 6.93 2.18 3.52 4.91 1.89
30 MPa 5.55 6.96 2.18 3.47 4.84 1.87
20 MPa 5.49 6.94 2.17 3.39 4.73 1.81
10 MPa 5.39 7.08 2.13 3.29 4.73 1.74 FIG. 4. V p versus net overburden pressure in a sandstone. In-
5 MPa 5.26 7.08 2.02 3.16 4.77 1.64 creasing the net overburden pressure from 1650 to 2650 psi
Air saturated causes a 5.2% V p increase, whereas the same amount of net
40 MPa 5.41 6.35 2.87 3.57 4.57 1.83 overburden pressure increase from 4500 to 5500 psi only re-
sults in a 0.5% V p increase.
 Fundamentals of Rock Physics 405

on temperature in hydrocarbon-saturated rocks provides the (71◦ F or 22◦ C) to 350◦ F (177◦ C). In Figure 5a, V p decreases by
physical basis for seismic monitoring of thermal EOR. Tosaya about 15% as temperature increases from 22◦ to 177◦ C. Fig-
et al. (1987) first showed dramatic decreases of compressional ure 5b shows that Vs also decreases by around 15% as temper-
velocities in heavy oil sands. As temperature increases from ature increases from 22◦ to 177◦ C. Because Vs is theoretically
25◦ to 125◦ C, V p can drop by 35% to almost 90%! Such huge not affected by fluids, the Vs decrease in Figure 5b is caused
decreases are caused in part by the compressibility increases of by changes in the rock frame and in the rock-fluid interactions.
the oil. Furthermore, part of the V p decrease was caused by the Because heavy oils are viscous, there exists a strong interfacial
abnormally high pore pressure in Tosaya et al.’s experimental force between the oil and rock grains (one analogy is the asphalt
setup: the oil inside the pore space expanded very fast when the used in road paving). As temperature increases, oil viscosity
core was heated up, yet the pore pressure tube was plugged by and the interfacial force decrease, loosening up the sand grains
the cold viscous oil outside the pressure vessel. In spite of this, so that both the bulk modulus and shear rigidity decrease.
the magnitude of V p decreases was so profound that further V p shows a step drop between 120◦ and 177◦ C at low pore
investigation was needed. pressures (50 and 100 psi) as water inside the rock turns to
Wang and Nur (1990b) measured numerous heavy oil sands steam (the top two curves), adding another 10% or so to the
and hydrocarbon-saturated rocks. Their results show that both V p decrease. In contrast, Vs increases by about 5% as a result of
V p and Vs decrease with increasing temperature. They avoided steam. The additional V p decrease is caused by the higher com-
the overpressure problem in their experiments and found that pressibility of steam compared to that of hot water (Figure 2a).
V p decreased by as much as 40% as temperature increased from The Vs increase is caused by the bulk density decrease as steam
20◦ to 125◦ C. displaces liquids out of the pore space due to volume expansion.
Figures 5a and 5b show V p and Vs , respectively, versus tem-
perature in a heavy oil sand. In the experiments, V p and Vs were LITHOLOGY
measured as functions of temperature from room temperature
For a given porosity and pore aspect ratio spectrum,
dolomite has the highest V p among reservoir rocks, followed
by limestone, sandstone, and unconsolidated sand. The order
is different in terms of decreasing Vs , with a possible order
of sandstone, dolomite, limestone, and unconsolidated sand.
Because of this, limestone has the highest V p /Vs ratio, fol-
lowed by dolomite, deeply buried unconsolidated sand, and
sandstone. Shallow unconsolidated sands may have very high
V p /Vs due to poor grain contacts. But once the grains contact
each other well under pressure, unconsolidated sands normally
show lower V p /Vs ratios than carbonates. The V p /Vs ratio dif-
ference therefore provides a tool for lithology discrimination.
Of course, both the velocity values and V p /Vs are complicated
by the pore aspect ratio spectrum, fracture alignment, pressure
effects, etc.
Shales are reservoir cap rocks that have a wide range of ve-
locities and impedances. However, shales always have higher
V p /Vs ratios than reservoir sands. As a result, when the cap
shale and reservoir sands have similar P-wave impedances,
their S-impedances are different. An example was shown by
MacLeod et al. (1999) in Alba field, North Sea. Because of
similar P-wave impedances between the cap shale and reser-
voir sand, streamer P-wave seismic data was inadequate for
defining reservoir boundaries. But the S-wave impedance con-
trast at top of the reservoir is very large, resulting in successful
reservoir imaging by use of P to S converted waves.

DENSITY

In theory, seismic velocities do not have to increase with bulk

density. For example, anhydrite has a higher bulk density but a
lower velocity compared to dolomite. Furthermore, increasing
the bulk density by adding more water in a partially gas/water
saturated rock will also decrease the velocities because the
FIG. 5. V p (a) and Vs (b) versus temperature in a heavy oil added water increases the bulk density but not the bulk modu-
sandy. Both Vp and Vs decrease by about 15% as temperature lus. Empirical relationships do exist that relate seismic velocity
increase from 22◦ to 177◦ C. V p drops further between 120◦ and increases with increasing bulk density. One of these is Gardner
177◦ C at low pore pressures (50 and 100 psi) as water inside
the rock transforms to steam [the top two curves in (a)], adding et al.’s (1974) classic relationship:
another 10% or so V p decrease. In contrast, Vs increases by
about 5% as results of steam. ρ = 0.23V p0.25 , (26)
406 Wang

where V p is the compressional velocity in ft/s and ρ is the bulk TEXTURE

density in g/cm3 . Gardner et al.’s relationship, however, can
Seismic properties are also controlled by texture such as
only estimate compressional velocities from bulk densities in
grain-to-grain contacts, roundness, sorting, cementation, etc.
water-saturated sedimentary rocks. Although Gardner et al.
Poor grain-to-grain contacts normally result in lower seismic
did give separate curves for individual lithologies, equation (26)
velocities and impedances, whereas cementation increases seis-
treats all sedimentary rocks as a single group, so that only a
mic properties sharply. Large grain-size sands show higher
single V p − ρ relationship exists for all the sedimentary rocks.
seismic velocities compared to fine sands due to larger con-
Wang (2000b) also divided sedimentary rocks into several
tact areas among grains. Poorly sorted sands show higher seis-
subgroups based on lithology and developed a set of relation-
mic velocities because poor sorting reduces porosity. Spencer
ships between bulk density and both V p and Vs . These rela-
et al. (1994) suggest that sorting and grain sizes of unconsol-
tionships are based on a large laboratory data set (over 500
idated sands may affect both the V p /Vs and Poisson’s ratio
data points) and hold for a variety of lithologies and satura-
in unconsolidated sands. The roundness or angularity of the
tions. Figure 6a shows that Gardner et al.’s (1974) relationship
sand grains may also affect seismic velocities and V p /Vs : well-
(dotted line) in general underestimates V p in water-saturated
rounded grains lead to better grain contacts and hence higher
reservoir rocks but overestimates V p in water-saturated shales.
velocities. Because textures of sedimentary rocks are difficult
Wang’s (2000b) refined relationships (solid lines) work better if
to quantify and cannot be described without cores, the effect
lithology is known. They also work for gas-saturated rocks. Fig-
of texture on seismic properties is in turn poorly quantified.
ure 6b shows that systematic relationships also exist between
Further investigation on this subject is needed.
bulk density and Vs . These results are similar to the findings of
Castagna et al. (1993).
COMPRESSIONAL VERSUS SHEAR VELOCITY

Although dipole sonic logs are nowadays commonly ac-

quired in wells, the need to derive Vs from the available V p
is still routine due to the amount of existing legacy data in
many developed fields where only monopole logs are available.
Castagna et al. (1985) published an empirical relation that re-
lates V p to Vs velocities for water-saturated clastic silicate rocks.
The correlation is known as the mudrock line:

V p = 1.36 + 1.16Vs (km/s). (27)

Although the mudrock line is useful in deriving shear ve-

locities when other alternative correlations are unavailable, it
suffers some obvious drawbacks: (1) it underestimates Vs with
known V p in soft unconsolidated sands and some clean lithified
sands, and (2) it is valid only for water-saturated clastic rocks.
Krief et al. (1990) presented a model relating bulk and shear
moduli in dry rocks to the Biot coefficient (Biot, 1941). For
fluid-saturated rocks, the authors use the Gassmann equation
(Gassmann, 1951) and give a straight-line relationship involv-
ing the squares of compressional and shear velocities in fluid
saturated rocks:
2
V p,sat − V f2 2
V p,m − V f2
2
= 2
, (28)
Vs,sat Vs,m

where V p,sat and Vs,sat are compressional and shear velocities,

respectively, in a fluid-saturated rock; V p,m and Vs,m are com-
pressional and shear velocities, respectively, in the rock’s solid
matrix (mineral); and V f is the compressional velocity in the
pore fluid.
Krief et al.’s (1990) model assumes that the frame V p /Vs
equals to the solid matrix’s V p /Vs . This is only approximately
true for quartz sands and sandstones. As a result, equation (27)
cannot be used for rocks other than sands and sandstones.
Greenberg and Castagna (1992) presented an iterative
model for estimating shear velocity in porous rocks. The model
combines the Gassmann equation and the Voigt-Reuss-Hill ve-
FIG. 6. V p (a) and Vs (b) versus bulk density in sedimentary
rocks. Gardner et al.’s (1974) relationship [dotted line in (a)] in locity average to yield shear velocity. The required input pa-
general underestimates V p in water-saturated reservoir rocks rameters to this model are lithology, saturation, porosity, and
but overestimates V p in water-saturated shales. compressional velocity.
 Fundamentals of Rock Physics 407

Murphy et al. (1993) observe that the frame shear to bulk assumed isotropic, always have higher V p /Vs ratios than reser-
modulus ratio is close to a constant value of 0.9 for quartz sands, voir sands. In carbonates, Rafavich et al. (1984) showed that
which yields a constant V p /Vs ratio of 1.55. Murphy et al. also V p /Vs could be used to discriminate limestones from dolomites.
give a set of relationships between frame bulk and shear moduli Wilkens et al. (1984) showed that V p /Vs could be a good
to porosity. For porosity less or equal to 0.35, the frame bulk discriminant of mineral composition for siliceous limestones.
and shear moduli are second-order polynomial functions of Wang and Szata (1999) presented data showing that nonporous
porosity: limestones and porous dolomites in an off-reef setting could
have similar impedances but distinct V p /Vs or impedance ra-
K d = 38.18(1 − 3.39φ + 1.95φ 2 ) (29) tios. Figures 8a and 8b show their results, namely the impedance
G d = 42.65(1 − 3.48φ + 2.19φ ). 2
(30) ratio versus P- and S-wave impedance, respectively. A clear
separation exists between the reservoir (porous dolomite) and
For porosity greater than 0.35, the relationships become expo- nonreservoir (nonporous limestone) rocks. Wang and Szata’s
nential: rock physics results led to a successful field trial with multicom-
ponent seismology, whose objective was to discriminate porous
K d = exp(−62.60φ + 22.58) (31) reservoir dolomites from nonporous limestones.
G d = exp(−62.69φ + 22.73), (32) V p /Vs or impedance ratios have also been used successfully
for direct hydrocarbon detection, especially with AVO tech-
where K d and G d are the bulk and shear moduli, respectively, niques. Because shear waves are insensitive to fluid changes,
of the frame rock (rock with empty pores)in GPa; φ is porosity while compressional waves are, changes in fluid type and satu-
in volume fraction. ration will result in V p /Vs changes. For a comprehensive review
When porosity, density, and shear velocity data are available, of this subject, see Castagna and Backus (1993).
the shear modulus can be calculated with both G d = ρVs2 and
equation (30) or (32). The two G d values are then compared.
If the two are close, the rock is a sand. If the calculated frame
shear modulus with G d = ρVs2 is significantly lower than the G d
from equation (30) or (32), the rock is a shale.
Murphy et al.’s (1993) findings are comparable to Wang’s
(2000c) results, who used a larger data set: the frame shear
to bulk modulus ratio averages 0.9639 for granular materials,
sands, and sandstones. This corresponds to a frame V p /Vs ratio
of about 1.54 (Poisson’s ratio is 0.135). Figure 7 shows that,
in general, a linear relationship exists between the frame bulk
(K d ) and shear (G d ) moduli in clastic granular rocks. This also
agrees with Castagna et al. (1985), who found that the frame
bulk and shear moduli are approximately equal to each other
in clastic rocks.

V p /V s RATIO

It has been long recognized that V p /Vs can be used as a lithol-

ogy indicator (e.g., Tatham, 1982; Domenico, 1984). Shales, if

FIG. 8. Impedance ratio versus P-wave (a) and S-wave (b)

impedance in carbonate rocks. A clear separation (solid line)
FIG. 7. A linear relationship exists between the frame bulk exists between the reservoir (porous dolomite) and nonreser-
(K d ) and shear (G d ) moduli in clastic granular rocks. R is the voir (nonporous limestone) rocks. Pd is different pressure (net
correlation coefficient. overburden pressure).
408 Wang

ANISOTROPY AND SHALE PROPERTIES and dolomite have a similar shear modulus, but dolomite is
denser, resulting in lower Vs in pure dolomite compared to solid
Many hydrocarbon reservoirs are relatively deep, ranging
quartz. However, because of the poor connectivity of quartz
from a few hundred to a few thousand meters in depth. This
grains, sandstones normally have lower Vs than dolomite.
means that thick nonreservoir formations can overlie reser-
In seismic applications of rock physics, I have been fre-
voir rocks. Furthermore, many hydrocarbon reservoirs are
quently asked for rules of thumb such as “Which way does
sealed by massive, impermeable shales that are intrinsically
V p go, and by how much, if reservoir pressure increases?”, etc.
anisotropic. Seismic waves travel through these overburden
Because rocks are so complicated and heterogeneous micro-
rocks before reaching the reservoir. An understanding of the
scopically, these rules of thumb only work in a general sense.
overlying shale properties and anisotropy is therefore impor-
The magnitude of change in seismic properties in response
tant for seismic imaging. Migration result of a Gulf Coast seis-
to changes of a petrophysical parameter varies from rock to
mic section that accounts for anisotropy shows much sharper
rock. As a result, the rules of thumb presented here are only
images of the reservoir and faults, as compared to migra-
qualitative.
tion without taking anisotropy into account (Meadows and
Table 1 shows some of the factors influencing seismic prop-
Abriel, 1994). Until recently, however, rock physics data on
erties of sedimentary rocks. These factors are grouped in three
seismic velocities and anisotropy in shales have been scarce,
categories: rock properties, fluid properties, and environment.
largely due to the time-consuming nature of laboratory mea-
Elaborations can be found in Tables 3, 4, and 5. The list is com-
surements of shales. Because at least five independent mea-
prehensive, but by no means complete. Most of these factors
surements are needed to describe a transversely isotropic (TI)
also interact with each other; it is likely that changing one fac-
medium, three separate, adjacent core plugs are traditionally
tor would change several others. For purposes of clarity, each
required.
factor is discussed separately in Tables 3, 4, and 5, so that other
Two types of anisotropy exist in rocks: intrinsic anisotropy
factors are assumed to remain constant. In practice, one should
and induced anisotropy. Intrinsic anisotropy is caused by the
be aware of the interactions among these factors.
preferential alignment of elongate grains or pores and fine lam-
ination. Intrinsic anisotropy in sedimentary rocks is normally in DISCUSSION
the form of transverse isotropy, so that five independent elas-
tic constants are needed to describe the elastic properties of Although impressive headway has been made in recent years
the rocks. Most shales are intrinsically transversely isotropic. in the understanding of the physical properties of rocks and
Kaarsberg (1959) was probably the first to study the sonic prop- in the application of this knowledge to seismic monitoring, to
erties of shales. Later on, many authors presented results on direct hydrocarbon detection, to lithology, fluid, and porosity
shale properties and anisotropy based on limited amounts of inversion, and to seismic interpretation, there are still many
data (e.g., Jones and Wang, 1981; Tosaya, 1982; Lo et al., 1986; unsolved problems.
Vernik and Nur, 1992; Hornby, 1994; Johnston and Christensen,
1995; Vernik and Liu, 1997). All the measured shales show a Know the range of applicability before using a theory or em-
substantial degree of seismic anisotropy, ranging from a few pirical equation
percent to as high as 50%.
Induced anisotropy is caused by stress anisotropy and frac- Because rocks are so complicated and variable from field to
turing. Nur and Simmons (1969) were the first to study stress- field, a universal rock physics theory or empirical relationship
induced anisotropy. Stress anisotropy preferentially aligns is impossible to obtain. When theories and models are used,
pores, grains, cracks, and fractures so that an otherwise one has to be aware of their range of applicability and the
isotropic rock becomes seismically anisotropic.
Figure 9 shows the measured compressional and shear ve-
locities in a brine-saturated soft shale. The compressional and
shear velocities show about 20% and about 15% anisotropy,
respectively. From many laboratory measurements of shale
anisotropy, I have seen as much as 50% velocity anisotropy
in both P- and S-waves. Obviously, such high magnitude
of anisotropy cannot be neglected in seismic processing and
interpretation.

SUMMARY: ROCK PHYSICS RULES OF THUMB

Although rocks are extremely complicated, their seismic

properties (V p , Vs , and impedance) depend largely on a com-
bination of connectivity and elasticity and density of what is
connected. As a result, many rules of thumb can be explained
by such combinations. For example, dolomite has a high bulk
modulus and the individual “grains” are well connected in a FIG. 9. Measured V p and Vs in a brine-saturated soft shale.
dolomite rock, so that compressional velocity and impedance Three V p s were measured at propagation directions of 81.2◦ ,
45.7◦ , and 8.8◦ to the symmetry axis. Two shear velocities were
are high. Compared to dolomite, quartz has a lower bulk mod- measured at a propagation direction of 81.2◦ to the symmetry
ulus and the grains are not as well connected, resulting in lower axis (the S H - and SV -waves are polarized at 81.2◦ and 8.8◦ to
compressional velocities in sandstones. For shear waves, quartz the symmetry axis, respectively).
 Fundamentals of Rock Physics 409

Table 3. Rock physics rules of thumb, I: Relations between Table 4. Rock physics rules of thumb, II: Relations between
seismic properties and rock properties.∗ seismic properties and fluid properties.∗

Compaction Better compacted rocks have higher seismic Viscosity Rocks with more viscous oil tend to have
properties (compressional and shear higher seismic properties. When the oil’s
velocities and impedances) due to better viscosity is extremely high, the oil may act
connectivity/contact. as a semisolid in the rock, resulting in
Consolidation Better consolidated rocks have higher higher seismic properties.
seismic properties due to better Density Higher density oils have higher bulk moduli
connectivity/contact. (lower compressibilities). As a result,
Age Older rocks have higher seismic properties rocks with higher density oils have higher
due to better compaction. compressional properties and shear
Cementation Cemented rocks have higher seismic impedance, but shear velocity may be lower.
properties due to better connectivity Wettability Wetting changes the interfacial energy
and contacts among grains. between the rock and the saturating fluid
Texture Texture includes grain size, sorting, (the interfacial forces are attractive between
roundness, etc. In general, sands with the wetting fluid and the solid surface and
larger grain size and sands with poorly repulsive between a nonwetting fluid and
sorted grains have higher seismic the solid surface). A rock fully saturated
properties due to better contacts and lower with a nonwetting fluid has slightly higher
porosity. Sands with angular grains have seismic properties compared with the same
lower seismic properties but higher V p /Vs rock saturated with the same but wetting
compared with sands composed of fluid. Most sands are water wet, so seismic
spherical grains. properties decrease when water is added
Bulk density Statistically, rocks with higher bulk density to a dry rock (softening effect).
have higher seismic properties. Fluid Heavier oils contain higher carbon-number
Clay content The effect of clay on seismic properties composition hydrocarbons so that their bulk moduli are
depends on the position of the clay and type higher. As a result, rocks saturated with
particles inside the rock. heavier oils show higher compressional
Statistically, rocks with higher clay properties. Shear properties are much less
content have lower seismic properties affected with shear impedance being
and higher V p /Vs . slightly higher and shear velocity being
Anisotropy Two types of anisotropy exist: intrinsic and slightly lower.
induced. Almost all shales are In general, high salinity brine has the
intrinsically anisotropic. A good highest bulk modulus, followed by low
approximation of shale anisotropy for flat salinity brine, fresh water, oil, and gas. But
layers is vertical transverse isotropy. Older, some heavy oils may have a bulk modulus
consolidated shales usually as high as, or even higher than, water’s.
have higher degree of seismic anisotropy. Fluid phase Rocks saturated with fluids in their gaseous
Stress anisotropy is the major cause of phase (hydrocarbon gas, steam, CO2 gas)
induced anisotropy. Most reservoir rocks have lower compressional seismic properties
can be approximated being intrinsically and shear impedances but slightly higher
isotropic unless they are finely layered. shear velocities due to the fluids’ higher
Fractures Fractures cause seismic properties to compressibility and lower bulk density,
decrease. Compressional properties are resulting in lower V p /Vs in the rocks.
affected much more by fractures compared Seismic properties hardly depend on the
to shear properties when the waves are amount of gas when gas saturation is
polarized across the fractures. Waves between 5 and 100%.
polarized perpendicular to a fracture have Gas-oil and The larger the amount of gas dissolved in an
low V p /Vs . Aligned fractures also cause Gas-water oil, the lower bulk modulus of the oil has.
seismic anisotropy. ratios Some high gas-oil-ratio (GOR) oils act like
Porosity Statistically, seismic properties decrease gas seismically, resulting in “bright spots” at
as porosity increases. V p /Vs is oil-water contacts. Rocks saturated with high
normally not a strong function of porosity. GOR oils have lower compressional seismic
Lithology Dolomite has high seismic properties, properties compared to the same rocks
followed by limestone and sand. Shales saturated with gas-free oils. Gas dissolved in
may have higher or lower seismic water has little effect on the bulk modulus
properties than sands. Limestone has high because of the unique molecular structure
V p /Vs , followed by dolomite and sands. of water. Little gas can be dissolved in water
Shales always have higher V p /Vs than anyway.
sands. Uncompacted sands may also have Saturation Full saturation of a liquid in a rock increases
high V p /Vs . the compressional seismic properties and
Pore shape Pore shape is the most important factor shear impedance but decreases Vs , resulting
influencing seismic properties yet is the in increases in V p /Vs . When gas is introduced
hardest to quantify. In general, rocks with to a fully liquid-saturated rock, the
flat pores have lower seismic properties. compressional seismic properties and shear
Pore shape difference is the major cause impedance all decrease but Vs increases,
of scatter in velocity-porosity plots. resulting in lower V p /Vs .
The magnitude of saturation effect is higher
∗ in rocks with weak frames and/or flat pores
Seismic properties increase with the elasticity and the micro-
connections in the rock. Note that each factor is assessed inde- (cracks, fractures).
pendently here assuming others are constant. In reality, many ∗
Rocks with more incompressible or viscous fluids have higher
of these factors interact. compressional properties. Shear properties in rocks are much
less sensitive to fluid saturation. Note that each factor is as-
sessed independently here assuming others are constant. In
reality, many of these factors interact.
410 Wang

Table 5. Rock physics rules of thumb, III: Relations between assumptions. Whenever a theory is stretched outside of its
seismic properties and environment.∗ range of applicability and its assumptions are violated, erro-
neous results are generated and in turn seismic data are misin-
Frequency Normally, seismic properties are higher at terpreted. One typical example is the misuse of Wyllie et al.’s
high frequencies (dispersion). However, the
magnitude of dispersion is hard to assess (1956) time-average equation to calculate the effect of gas sat-
due to difficulties in measuring seismic uration on seismic velocities (the time-average equation was
properties as functions of continuous derived for liquid-saturated rocks). On the other hand, empiri-
frequency across seismic (10–200 Hz), log cal equations are often generated from a specific data set within
(around 10 kHz), and lab frequency certain pressure, temperature, saturation, etc., ranges Beyond
(100 kHz–2 MHz) bands.
Stress history Knowing stress history of a rock under the data set and its physical range, empirical equations should
measurements can help plan the be used with great caution.
measurements and interpret the data. Hard
rocks are usually brittle. If a rock was under
great stress, the relief of stress generates Use rock physics data and knowledge whenever possible
microcrocks in the rock, which reduces
seismic velocities. Simply restoring the rock Although rock physics has played fundamental roles in de-
to the original stress may or may not veloping seismic technologies such as 4-D seismology, AVO,
eliminate the induced microcracks. rock property inversions, etc., it has been underused in inter-
The stress relief-induced microcracks may preting seismic data. With ever-increasing computing power
help identify the maximum/minimum
horizontal stress directions in situ. and constantly improving seismic processing technologies, I be-
Depositional Depositional environment variously affects lieve that seismic data are going to be interpreted in more and
environment seismic properties. For example, authigenic more detail for lithology and pore fluid discrimination, porosity
and detrital clays affect seismic properties inversion, and extraction of reservoir parameters such as pres-
differently. The rate and source of sure and temperature. Without rock physics, these tasks would
deposition and depositional energy also
affect seismic properties (e.g., sand/shale be hard to achieve. However, the challenge is that quantita-
sequences are results of alternating tive interpretation needs large amounts of rock physics data to
depositional energy, rate, and source). yield statistical relationships and to decouple the combined ef-
Temperature Seismic properties decrease as temperature fects of reservoir properties and parameters. Although skilled
increases. Dry (empty pores) and water- laboratory measurements seem to be relatively expensive for
saturated rocks normally show small large amounts of data, the potential cost savings and economic
seismic property decreases as temperature
increases. However, rocks, especially soft benefits of a more quantitative seismic interpretation could be
sands, saturated with heavy oils show large enormous.
seismic property decreases as temperature
increases.
Reservoir In seismic reservoir characterization True magnitude of dispersion is still unknown
process and 4-D seismic reservoir monitoring, it is
paramount to understand the reservoir One of the daunting issues in rock physics is frequency and
process. For example, waterflood increases dispersion. The Gassmann (1951) equation is an inherent zero-
water saturation and pressure in a reservoir. frequency theory. Although Biot’s (1956a, b, 1962) theory in-
It may also decrease the temperature corporates frequency effects, it is inadequate to account for
slightly. In general, the effect of reservoir
process on seismic properties is a the overall dispersion in rocks (Winkler, 1985; Wang and Nur,
combination of several factors. These 1990a). Depending on the pore structure and fluid viscosity,
factors may cancel or enhance each other’s seismic frequencies may well be considered in the high fre-
effect on seismic properties. quency range relative to the Gassmann equation. Due to ex-
Production The effect of production history is similar perimental difficulties, the Gassmann equation has not been
history to that of reservoir process on seismic
properties. In seismic reservoir character- rigorously verified in the laboratory or in the field. Further-
ization, production data and reservoir more, there exist many so-called high-frequency theories in the
parameters (fluids, saturation, pressure, literature that should be used with great caution. In essence,
temperature) need to be carefully gathered knowing the assumptions and limitations of the theories, mod-
at the time of the seismic survey in order els, and equations is paramount in order to produce meaningful
to better interpret the seismic data.
Net reservoir Net reservoir pressure is the difference results.
pressure between overburden and reservoir pressures. While Biot’s theory considers only the effect of large-scale
Because overburden pressure does not fluid flow (regional flow) on seismic wave dispersion, the local
change (or changes very little) in a flow model (Mavko and Nur, 1975; O’Connell and Budian-
reservoir, the effect of net reservoir sky, 1977) considers the small, pore-scale squirt flow. The two
pressure is just the opposite of the effect
of reservoir pressure on seismic properties. mechanisms in combination seem to explain well the fluid flow
Seismic properties in all rocks increase as effect on seismic wave dispersion. In practice, seismic data gath-
net reservoir pressure increases. The ered at 10–200 Hz are close to the Gassmann results for high-
magnitude of such an increase depends on porosity, high-permeability rocks saturated with low-viscosity
several other factors (pore shape, porosity, fluids. For low-porosity, low-permeability rocks saturated with
pore fluids, lithology, etc.). It can only high-viscosity fluids, data gathered at seismic frequencies are
be quantified through measurements.
∗ close to log- or lab-measured data (Wang, 2000a). However,
Pressure, temperature, and history (both depositional and
production) all affect the current state of the rock. Note that the true magnitude of wave dispersion in fluid-saturated rocks
each factor is assessed independently here assuming others are will remain unknown until several technical problems can be
constant. In reality, many of these factors interact. solved. For details on velocity dispersion and frequency effects,
 Fundamentals of Rock Physics 411

see the collection of papers in Nur and Wang (1989) and Wang Bourbie, T., Coussy, O., and Zinszner, B., 1987, Acoustics of porous
media: Editions Technip.
and Nur (1992, 2000). Brown, R. J. S., and Korringa, J., 1975, On the dependence of the elastic
properties of porous rock on the compressibility of the pore fluid:
Geophysics, 40, 608–616.
Castagna, J. P., and Backus, M. M., 1993, Offset-dependent
Scale up by adequate sampling reflectivity—Theory and practice of AVO analysis, Soc. Expl.
Geophys.
Another issue in rock physics is the scaling effect: laboratory- Castagna, J. P., Batzle, M. L., and Eastwood, R. L., 1985, Relationships
measured rock physics data only represent properties of a tiny between compressional wave and shear wave velocities in clastic
portion of the hydrocarbon reservoir. Rocks are known to be silicate rocks: Geophysics, 50, 571–581.
Castagna, J. P., Batzle, M. L., and Kan, T. K., 1993, Rock physics—The
heterogeneous, especially at core scales. One solution to this link between rock properties and AVO response, in Castagna, J. P.,
scaling problem is that laboratory measurements should not and Backus, M. M., Eds., Offset-dependent reflectivity—Theory and
practice of AVO analysis: Soc. Expl. Geophys., 135–171.
focus on a few samples. Instead, laboratory samples should be Dodson, C. R., and Standing, M. B., 1945, Pressure-volume-
carefully selected to represent facies, depositional sequences, temperature and solubility relations for natural-gas-water mixtures,
lithology types, etc. A sufficient number of samples should be in Drilling and production practices: Am. Petr. Inst, 173–179.
Domenico, S. N., 1984, Rock lithology and porosity determination from
measured to derive statistical relationships between seismic shear and compressional-wave velocity: Geophysics, 49, 1188–1195.
and reservoir properties. Such statistical relationships effec- Eberhart-Phillips, D., Han, D.-H., and Zoback, M. D., 1989, Empiri-
tively scale up core data to the seismic scale. It is these statis- cal relationships among seismic velocity, effective pressure, porosity,
and clay content in sandstone: Geophysics, 54, 82–89.
tical relationships, not the individual values or numbers, that Gardner, G. H. F., Gardner, L. W., and Gregory, A. R., 1974, Formation
should be used in seismic modeling and interpretation. velocity and density—The diagnostic basic for stratigraphic traps:
Geophysics, 39, 770–780.
In thin-bed or sand-shale sequence reservoirs, thin beds may Gassmann, F., 1951, Uber die elastizitat poroser medien: Verteljahrss-
remain below seismic resolution. In this case, cores should be chrift der Naturforschenden Gesellschaft in Zurich, 96, 1–23.
again adequately sampled. Once the representative properties Geertsma, J., and Smit, D. C., 1961, Some aspects of elastic wave prop-
agation in fluid saturated porous solids: Geophysics, 26, 169–181.
of each thin bed are obtained, Backus (1962) averaging can be Greenberg, M. L., and Castagna, J. P., 1992, Shear-wave velocity esti-
used to scaleup the lab or log data. mation in porous rocks: Theoretical formulation, preliminary verifi-
Because rock physics connects reservoir properties and pa- cation and applications: Geophys. Prosp., 40, 195–209.
Gueguen, Y., and Palciauskas, V., 1994, Introduction to the physics of
rameters to seismic data, it plays a fundamental role in many rocks: Princeton Univ. Press.
technologies attempting to derive reservoir rock and fluid Han, D. H., Nur, A., and Morgan, D., 1986, Effects of porosity and clay
content on wave velocities in sandstones: Geophysics, 51, 2093–2107.
properties and reservoir parameters from seismic data. With Hashin, Z., and Shrikman, S., 1963, A variational approach to the the-
more and more fields maturing and technologies further im- ory of elastic behavior of multiphase materials: J. Mech. Phys. Solids,
proving and progressing, it is conceivable that one day seismic 11, 127–140.
Hill, R., 1952, The elastic behavior of a crystalline aggregate: Proc.
surveys will be routinely used in taking three-dimensional snap Phys. Soc. London Ser. A, 65, 349–354.
shots of reservoirs for mapping porosity and fluid flow, diagnos- Hornby, B. E., 1994, The elastic properties of shales: Ph.D. thesis, Univ.
ing reservoirs for pressure, temperature, saturation changes, of Cambridge.
Johnston, J. E., and Christensen, N. I., 1995, Seismic anisotropy of
and even inferring hydrocarbon types. During such a process, shales: J. Geophys. Res., 100, 5991–6003.
rock physics will be an essential component for quantitatively Jones, L. E. A., and Wang, H. F., 1981, Ultrasonic velocities in Creta-
interpreting seismic data and for unraveling the superposed ceous shales from the Williston Basin: Geophysics, 46, 288–297.
Kaarsberg, E. A., 1959, Introductory studies of natural and artificial
effects of reservoir changes on seismic data. argillaceous aggregates by sound propagation and X-ray diffraction
methods: J. Geol., 67, 447–472.
Krief, M., Garat, J., Stellingwerff, J., and Ventre, J., 1990, A petro-
ACKNOWLEDGMENTS physical interpretation using the velocities of P and S waves (full-
waveform sonic): The Log Analyst, Nov.-Dec., 355–369.
I thank Chevron Petroleum Technology Company for allow- Kuster, G. T., and Toksoz, M. N., 1974, Velocity and attenuation of
ing me to publish this paper and Sven Treitel and Ali Tura for seismic waves in two-phase media—Part I: Theoretical formulations:
inviting me to contribute to the Y2K tutorial papers. Internal Geophysics, 39, 587–606.
Langan, R. T., Lazaratos, S. K., Harris, J. M., Vassiliou, A. A., Jensen,
reviews by Joseph Stefani and Ronald Behrens and external T. L., and Fairborn, J. W., 1997, Imaging of a stratigraphically complex
reviews by Sven Treitel, John Castagna, Ali Tura, Chris Liner, carbonate reservoir with crosswell seismic data, in Marfurt, K. J., Ed.,
Carbonate seismology: Soc. Expl. Geophys., 417–423.
and an anonymous reviewer greatly improved the readability Liu, Y., 1998, Acoustic properties of reservoir fluids: Ph.D. thesis, Stan-
and technical content. Sven Treitel meticulously edited the re- ford Univ.
vised manuscript and made it much more readable. Their help Lo, T. W., Coyner, K. B., and Toksoz, M. N., 1986, Experimental deter-
mination of elastic anisotropy of Berea Sandstone Chicopee Shale
is greatly appreciated. and Chelmsford Granite: Geophysics, 51, 164–171.
MacLeod, M. K., Hanson, R. A., Bell, C. R., and McHugo, S., 1999,
REFERENCES The Alba field ocean bottom cable seismic survey: Impact on field
development: The Leading Edge, 18, 1306–1312.
Backus, G., 1962, Long-wave elastic anisotropy produced by horizontal Mavko, G., and Nur, A., 1975, Melt squirt in the asthenosphere: J.
layering: J. Geophys. Res., 67, 4427–4440. Geophys. Res., 80, 1444–1448.
Batzle, M. L., and Wang, Z., 1992, Seismic properties of pore fluids: Mavko, G., Mukerji, T., and Dvorkin, J., 1998, The rock physics hand-
Geophysics, 57, 1396–1408. book: Tools for seismic analysis in porous media: Cambridge Univ.
Biot, M. A., 1941, General theory of three dimensional consolidation: Press.
J. Appl. Phys., 12, 155–164. Meadows, M. A., and Abriel, W. L., 1994, 3-D poststack phase-shift
——— 1956a, Theory of propagation of elastic waves in a fluid satu- migration in transversely isotropic media: 64th Ann. Internat. Mtg.,
rated porous solid. I. Low frequency range: J. Acoust. Soc. Am., 28, Soc. Expl. Geophys., Expanded Abstracts, 1205–1208.
168–178. Meyer, C. A., McClintock, R. B., Silvestri, G. J., and Spencer, Jr., R. C.,
——— 1956b, Theory of propagation of elastic waves in a fluid satu- 1979, ASME Steam Tables: Am. Soc. Mech. Eng.
rated porous solid. II. Higher frequency range: J. Acoust. Soc. Am., Murphy, W., Reischer, A., and Hsu, K., 1993, Modulus decomposition
28, 179–191. of compressional and shear velocities in sand bodies: Geophysics,
——— 1962, Generalized theory of acoustic propagation in porous 58, 227–239.
dissipative media: J. Acoust. Soc. Am., 34, 1254–1264. Nur, A., and Simmons, G., 1969, Stress-induced velocity anisotropy in
412 Wang

rock: An experimental study: J. Geophys. Res., 74, 6667–6674. Vernik, L., and Nur, A., 1992, Ultrasonic velocity and anisotropy of
Nur, A., and Wang, Z., Eds., 1989, Seismic and acoustic velocities hydrocarbon source rocks: Geophysics, 57, 727–735.
in reservoir rocks: Volume I: Experimental studies: Soc. Expl. Voigt, W., 1928, Lehrbuch der Kristallphysik: Teubner.
Geophys. Wang, Z., 2000a, The Gassmann equation revisited: Comparing labo-
O’Connell, R. J., and Budiansky, B., 1977, Viscoelastic properties of ratory data with Gassmann’s predictions, in Wang, Z., and Nur, A.,
fluid-saturated cracked solids: J. Geophys. Res., 82, 5719–5736. Eds., Seismic and acoustic velocities in reservoir rocks, 3: Recent
Osif, T. L., 1988, The effects of salt, gas, temperature, and pres- developments: Soc. Expl. Geophys., 8–23.
sure on the compressibility of water: SPE Res. Eng., 3, 175– ——— 2000b, Velocity-density relationships in sedimentary rocks, in
181. Wang, Z., and Nur, A., Eds., Seismic and acoustic velocities in reser-
Rafavich, F., Kendall, C. H. St. C., and Todd, T. P., 1984, The relation- voir rocks, 3: Recent development: Soc: Expl. Geophys., 258–268.
ship between acoustic properties and the petrographic character of ——— 2000c, Velocity relationships in granular rocks, in Wang, Z., and
carbonate rocks: Geophysics, 49, 1622–1636. Nur, A., Eds., Seismic and acoustic velocities in reservoir rocks, 3:
Reuss, A., 1929, Berechnung der fliessgrense von mischkristallen auf Recent development: Soc. Expl. Geophys., 377–383.
grund der plastizitatbedingung fur einkristalle: Zeitschrift fur Ange- Wang, Z., and Nur, A., 1990a, Dispersion analysis of acoustic velocities
wandte Mathematik aus Mechnik, 9, 49–58. in rocks: J. Acoust. Soc. Am., 87, 2384–2395.
Schön, J. H., 1995, Physical properties of rocks: Fundamentals and ——— 1990b, Wave velocities in hydrocarbon saturated rocks: Exper-
principles of petrophysics: Pergamon Press. imental results: Geophysics, 55, 723–733.
Sergeev, L. A., 1948, Ultrasonic velocities in methane saturated oils and Wang, Z., and Nur, A., Eds., 1992, Seismic and acoustic velocities
water for estimating sound reflectivity of an oil layer: 4th All-Union in reservoir rocks, 2: Theoretical and model studies: Soc. Expl.
Acoust. Conf. Izd. Nauk USSR (English translation). Geophys.
Spencer, J. W., Jr., Cates, M. E., and Thompson, D. D., 1994, Frame ——— 2000, Seismic and acoustic velocities in reservoir rocks, 3: Re-
moduli of unconsolidated sands and sandstones: Geophysics, 59, cent development: Soc. Expl. Geophys.
1352–1361. Wang, Z., and Szata, K., 1999, Rock physics lithology and porosity pre-
Tatham, R. H., 1982, V p /Vs and lithology: Geophysics, 47, 336–344. diction in carbonate reservoirs: Presented at the 69th Ann. Internat.
Timur, A., 1977, Temperature dependence of compressional and shear Mtg., Soc. Explr. Geophys.
wave velocities in rocks: Geophysics, 42, 950–956. Wilkens, R., Simmons, G., and Caruso, L., 1984, The ratio V p /Vs as
Tosaya, C. A., 1982, Acoustic properties of clay-bearing rocks: Ph.D. a discriminant of composition for siliceous limestones: Geophysics,
thesis, Stanford Univ. 49, 1850–1860.
Tosaya, C., and Nur, A., 1982, Effects of diagenesis and clays on com- Winkler, K. W., 1985, Dispersion analysis of velocity and attenuation
pressional velocities in rocks: Geophys. Res. Lett., 9, 5–8. in Berea sandstone: J. Geophys. Res., 90, 6793–6800.
Tosaya, C., Nur, A., Vo-Thanh, D., and Da Prat, G., 1987, Labora- Wood, A. B., 1941, A textbook of sound: G. Bell and Sons.
tory seismic method for remote monitoring of thermal EOR: SPE Wyllie, M. R. J., Gregory, A. R., and Gardner, L. W., 1956, Elastic wave
Reservoir Engineering, 2, 238–242. velocities in heterogeneous and porous media: Geophysics, 21, 41–
Vargaftik, N. B., 1975, Tables on the thermodynamical properties of 70.
liquids and gases: Hemisphere Pub. Co. Younglove, B. A., and Ely, J. F., 1987, Thermophysical properties of
Vernik, L., and Liu, X., 1997, Velocity anisotropy in shales: A petro- fluids. II. Methane, ethane, propane, isobutane, and normal butane:
physical study: Geophysics, 62, 521–532. J. Phys. Chem. Ref. Data., 16, 577–798.