You are on page 1of 19

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/320298106

Rock Quality, Seismic Velocity, Attenuation and Anisotropy

Book · January 2006

CITATIONS READS
2 1,068

1 author:

Nick Ryland Barton


Nick Barton & Associates
300 PUBLICATIONS   11,292 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Seismic measurements in the earth sciences View project

Statistical mechanics and Population Genetics View project

All content following this page was uploaded by Nick Ryland Barton on 10 October 2017.

The user has requested enhancement of the downloaded file.


Barton-01.qxd 8/17/06 5:19 PM Page 1

Part I
Barton-01.qxd 8/17/06 5:19 PM Page 2
Barton-01.qxd 8/17/06 5:19 PM Page 3

1 Shallow seismic refraction,


some basic theory, and the
importance of rock type

‘Nature has left us an incomplete and often well-concealed record of her activities, and no ‘as con-
structed’ drawings!’ (Stapledon and Rissler, 1983)

‘Tenders for the Tay pipeline crossing did not allow time for boreholes to locate bedrock. Seismic
refraction took one day to confirm that the trench would not encounter rock. The pipeline was laid
in sediments.’ (Gardener, 1992)

‘The time may come when the various relations between geophysical parameters and rock properties
can be usefully combined into a single classification system.’ (Darracott and Orr, 1976)

1.1 The challenge of the that are perpendicular to the major stress, and the more
near-surface in civil open state of those that are parallel will give the rock
engineering mass anisotropic stiffness. Consequently the rock mass
will frequently display anisotropic seismic velocities. By
Refraction seismics is by far the oldest method used in implications, hydraulic conductivities and deformation
exploration seismology, with its origin traced to R. Mallet moduli that show anisotropic distributions will be, at least
from 1848. Shallow refraction seismic measurements in part, detectable by seismic measurements. Anisotropy
using first arrival, compressional P-wave velocities close will also be caused by layered inter-beds, foliation and
to the surface often give a remarkable picture of near schistocity, and of course by a dominant joint set. Simple
surface conditions due to some fortuitous interactions of examples of (azimuthal) anisotropy, applicable in civil
physical phenomena. Firstly, weathering and the usual engineering, will be given in Chapter 3, while larger-
lack of significant stress near the surface has allowed joint scale examples of anisotropy detection will be described
systems, shear zones and faults to be exaggerated in both in much greater detail, and from various fields of the
their extent and severity. Secondly, stress levels are low earthsciences, in Chapters 13, 14 and 15 in Part II.
enough to allow joints and discontinuities to be seismic- Despite the obvious challenges of seismic interpret-
ally visible due to their measurable apertures. ation in fractured and faulted petroleum reservoirs at
So-called acoustic closure occurs at greater depths than many kilometers depth, or of mid-ocean ridge investi-
those usually penetrated by conventional hammer seis- gations beneath three kilometers of ocean, many geo-
mic, unless rock strengths are rather low (e.g., New and physicists insist that obtaining high resolution images
West, 1980; Hudson et al., 1980). (At this juncture, we from ground level to just 50 m depth, is still one of the
need to differentiate between two ‘J.A. Hudson’ authors, major challenges of modern geophysics. This happens to
one in geophysics, the other in rock engineering, and be the layer of the subsurface closest to most of our civil
both very prominent in their chosen fields. We will engineering endeavours, from tunnels, to dams, to the
occasionally refer to ‘rock’ Hudson in Part I, and later foundations for high buildings.
in Part II to ‘seismic’ Hudson). Undoubtedly, the ‘0 to 50 m’ challenge is mainly due
Since micro-fractures and rock joints are sensitive to to the extreme variability of the near-surface, resulting
stress levels, the more closed state of the discontinuities from the contrasting geological materials and weathering
Barton-01.qxd 8/17/06 5:19 PM Page 4

4 Rock quality, seismic velocity, attenuation and anisotropy

grades that are often present. There is also a velocity Sjøgren, 2000 suggested the following list of essen-
gradient that is extreme compared to anything found at tial information expected from near-surface seismic
greater crustal depths, where consolidation effects smooth surveys, performed for civil engineering geotechnical
out some of the differences. The first 5 m of unconsoli- investigations:
dated dry beach sand may see velocity increase from
150 m/s to 300 m/s, (Bachrach et al., 2000), giving a gra- ● The velocities of the overburden layers, including
dient of 30 s-1, which may be an order of magnitude the upper, less consolidated rock layers.
higher than the gradient over the next 50 to 100 m, where ● The thickness of the various overburden velocity
weathered and jointed rock may typically be found. layers, and the total depth to the main refractor.
There are an infinite number of challenges in the near- ● A detailed determination of the velocity distribu-
surface. Some of the worse may be karst phenomena in tion in the main refractor.
limestones, or the ‘inverse’ problems of core-stone anom- ● An estimate of the uncertainty of the velocity and
alies in the case of sparsely jointed but deeply weathered depth determinations.
granites and gneisses. These features have caused tun- ● An analysis of the (velocity-) depth structure.
nelling surprises in numerous countries, with nearly as ● An assessment of velocities in vertical and lateral
numerous arbitrations as a result. Although completely directions in relation to the geology.
weathered Grade V is an expected feature beneath the ● Seismic results in relation to results from other
Grade VI soil in tropical terrains, Grade V saprolite investigations, if available.
sometimes confusingly swaps places with the usually ● Conclusions and recommendations resulting from
deeper, and almost unjointed Grade I or II. (Saprolite the investigation that are of importance to the
is a weak, water sensitive, weathered in-place, some- project.
times beautifully structured and coloured relic of the
rock). Although reflection methods have eventually dom-
If this reversal of weathering grades appears in a tunnel inated the field of exploration seismics due to the various
arch beneath massive, high velocity core-stones, or if needs involved with deeper exploration, there is ‘univer-
there is a generally very undulating rock surface, with fre- sal’ use of shallow refraction seismic in sub-surface inves-
quent tunnel penetrations into weathered materials, tigations for civil engineering projects around the world,
there can be major delays. A tunnel collapse is difficult to due to its apparent simplicity and low cost. Further-
avoid when water is present, unless preparations have more, refraction seismics can be used to remove (from
been made, as a result of the more frequent exploratory the more deeply focussed reflection data), the ‘adverse’
drilling demanded when seismic anomalies such as these effect of the first meters or tens of meters of the hetero-
are suspected. geneous weathered layer, where differences in the ori-
Pre-injection ahead of the tunnel face, and heavier ginal rock quality may cause tens of meters of sub-surface
tunnel support, would be the very basic requirements ‘topography’ in the case of on-land exploration.
in a drill-and-blasted tunnel. (This is one of the purposes
of the ‘Q-system’ of rock mass characterization and
tunnel support selection). In the case of a TBM (tunnel 1.2 Some basic aspects concerning
boring machine) excavation, a change to a closed mode elastic body waves
in the case of a hybrid machine with earth-pressure-
balance (EPB) would be needed, especially if the wea- It is usually assumed that the strains associated with the
thered depressions in the bedrock contained water, as is passage of a seismic wave are of minute, sub-micron mag-
usually the case. nitude, and except in the neighbourhood of the source,
Best advice of all, as a direct result of a seismic refrac- the strains are generally assumed to be elastic. Based on
tion survey, would be to drive a deeper tunnel from the this assumption, the velocities of propagation of seismic
start. It is easy to imagine subway station construction waves are determined by the appropriate elastic moduli
under such heterogeneous conditions. It could be and densities of the materials passed through. The general
extremely time-consuming, and even dangerous. The form of the classic equations linking these three quan-
tities is V  (E/) ⁄ . Compressional bodywaves (primary
1
cost of deeper access to the stations, via longer escal- 2

ators, would be a small price to pay for much reduced or P-waves) propagate by alternating compression and
tunnelling and station costs. dilation (Figure 1.1 a) in the direction of the waves.
Barton-01.qxd 8/17/06 5:19 PM Page 5

Shallow seismic refraction, some basic theory, and the importance of rock type 5

The third important elastic modulus influencing the


conversion between dynamic properties is the bulk modu-
lus (K), defined as the ratio of the volumetric stress (P)
and the volumetric strain (v/v). Since the three mod-
uli are linked by the equation   K  4/3, it
follows that Vp can also be expressed as:

1
⎡ K  4 /3  ⎤ 2

Vp  ⎢ ⎥ (1.3)
⎢⎣  ⎥⎦

This equation therefore demonstrates the fundamen-


tally faster nature of Vp in relation to Vs. The ratio of these
two dynamic properties are also linked by the dynamic
Poisson’s ratio for the material, as will be shown in the
next section, which contains some standard equations.

1.2.1 Some sources of reduced


elastic moduli

In the case of micro-cracked, fractured, or jointed rock


masses, there is a correspondingly reduced set of moduli
in relation to the undisputed elastic nature of the intact
matrix, because of micro (and probably elastic) displace-
ments in normal and/or shear directions across and/
Figure 1.1 Elastic deformations and particle motions associated or along the micro-cracks, fractures or joints. These repre-
with the propagation of body waves: a) P-wave, sent an important part of the source of attenuation of
b) S-wave. Based on Bolt, 1982. the seismic waves in the dry state, due both to various
scales of wave scattering and due to the intrinsic micro-
deformations. Added losses are incurred if these micro-
The oscillating uniaxial strain involved in the case of a or macro-discontinuities are partly saturated, since there
confined body, means that the axial modulus () con- is communication with the pores and eventual pore
trols the velocity of propagation, thus: fluid, and minute flows may be initiated to equilibrate
pressures. These micro-imbalances will only be equili-
1
⎛⎞ 2 brated when the frequency is sufficiently low.
Vp  ⎜⎜ ⎟⎟⎟ (1.1) The above mechanisms mean that dynamic proper-
⎜⎝  ⎟⎠
ties, such as the velocities, Poisson’s ratio and attenuation
tend in practice to be dispersive, or frequency depend-
Shear bodywave waves, termed secondary, transverse ent. They are also of course rock quality and environment-
or S-waves propagate by a sinusoidal pure shear strain dependent, in the broadest possible meanings of these
(Figure 1.1 b) in a direction perpendicular to the direc- words. As rock quality declines, or the surface is
tion of the waves. The shear modulus (), which is approached, there develops a serious discrepancy between
given by the ratio of shear stress () divided by the shear the dynamic or elastic properties of the intact matrix
strain (tan ), will therefore control the (lower) velocity and the dynamic properties of the (partly discontinu-
of propagation, thus: ous) medium. The ratio between the dynamic proper-
ties of the (partly discontinuous) medium and the static
⎛ ⎞
1
2 deformation properties, such as the (rock mechanics)
Vs  ⎜⎜ ⎟⎟⎟ (1.2) deformation moduli and joint stiffnesses (the inverse
⎜⎝  ⎟⎠
of compliances), may rise into double figures in this
Barton-01.qxd 8/17/06 5:19 PM Page 6

6 Rock quality, seismic velocity, attenuation and anisotropy

complex region, where velocity-depth gradients are


often extreme.
At depth, under high confinement, and if rock quality
is high, it is assumed that there will be only small dis-
crepancies between the dynamic properties of the matrix
and the dynamic properties of the rock mass. The ‘static’
crack and joint stiffnesses, being so high, will be close to
the dynamic (inverted) crack and joint compliances.
There is controversy however, about the ratio of the
dynamic normal and shear compliances, and the
(inversed) ratio of the ‘static’ normal and shear stiffnesses.
There is even controversy over whether friction is a valid
attenuation mechanism, at the level of these micro-
displacements.
In the rock mechanics of ‘static’, ‘macro-deformations’,
we are familiar with a significant mismatch between the
high normal stiffness, and the much lower (and scale-
dependent) shear stiffness. Concerning the ratio of
dynamic compliances, geophysicists seem not to be so
sure, a dilemma that also probably affects whether fric-
tion is, after all, to be a valid attenuation mechanism, as
assumed in much of the geophysics literature, virtually
up to the present day.
Figure 1.2 Mean Vp and Vs statistics from 4 km of seismic profiles
in metamorphic and igneous rocks. Sjøgren, 1984.
1.3 Relationships between Vp and
Vs and their meaning in
field work
The rock mass quality Q-value, mentioned earlier in
the introduction, is composed of parameter pairs (RQD/
The advantages of using both P-wave and S-wave data to
Jn, Jr/Ja and Jw/SRF – see Appendix A for descriptions
interpret seismic results in hard rock was strongly empha-
and ratings). These effectively describe relative block size,
sised by Sjøgren, 1984. This has been reinforced by the
the inter-block friction coefficient, and an active stress. This
successively easier acquisition of multi-component, multi-
rock quality term will be utilised in various places in this
channel data, and rapidly developing PC analysis cap-
book, not least as a possible interpretation of seismic
abilities. In addition to many other sets of data, some of
quality Q (the inverse of attenuation). (Improvements
which will be referred to later, Sjøgren, 1984 presented
to equation 1.4 will be developed in Chapter 5, to allow
average Vp/Vs ratios from 93 rock sections from 5 differ-
for its application to weaker and more porous rock
ent sites in igneous and metamorphic rocks. These are
types, and to adjust it for depth or stress effects).
reproduced in Figure 1.2.
A rock quality Q-value of 1 tends to be heavily
P-wave velocities ranged from 3.3 to 5.7 km/s, and
jointed, containing some clay, while values
1, are tend-
S-wave velocities from 1.6 to 3.4 km/s. On average, Vp/Vs
ing towards better quality, with wider spacing of joints,
ratios were 1.89 in the rock mass with lower velocities
less joint sets, and no clay. (Q may reach values of about
(heavier jointing) and 1.80 in the rock mass with higher
1000 to 2000 in the case of massive, unjointed rock
velocities (sparser jointing).
masses, confined at depths of say 500 m or more).
These two ratios of Vp/Vs imply rock mass quality
The ratio Vp/Vs depends on dynamic Poisson’s ratio
Q-values of roughly 1 to 10, and 10 to 100 respectively,
( ) according to the following:
according to the following near-surface, hard rock Vp–Q
relationship (Barton,1991).
Vp 2 2
 (1.5)
Vp ≈ 3.5  log10 Q (1.4) Vs 1 2
Barton-01.qxd 8/17/06 5:19 PM Page 7

Shallow seismic refraction, some basic theory, and the importance of rock type 7

From equation 1.5 one can derive the value of dynamic In the case of massive rocks of low porosity, the static
Poisson’s ratio as follows: and dynamic values of the elastic constants (e.g., the
elastic moduli Estat. and Edyn.) are quite close, while for
( Vp / Vs )2 2 heavily fractured and clay bearing zones, large differ-
 (1.6) ences between Estat and Edyn are seen (e.g. Cosma, 1995).
2( Vp / Vs )2 1
Rock mass quality descriptions such as Q or RMR or
RQD, which are described in more detail later, correlate

The ratio Vp/Vs is about 3 for hard (zero-porosity) better with static moduli than with dynamic moduli.
rocks, for which  0.25. However, in the case of Numerous relations between these moduli will be given
unconsolidated sediments, the ratio Vp/Vs can even reach in Chapter 6.
values of 20 to 40 for near surface material, for which is
commonly greater than 0.45. Later in this chapter, high
values of (dynamic) Poisson’s ratio for a near-surface fault 1.4 Some advantages of shear
zone will also be seen, for similar reasons to the above. waves
A rock quality interpretation, linking these dynamic
parameters, can be added here, by taking Sjøgren’s (1984) In addressing the challenge of resolving the 0–50 m reso-
P- and S-wave results from 4.1 km of seismic profiles lution problem, Dasios et al., 1999 reported multi-
for hard but sometimes weathered metamorphic and component investigations at four shallow sites (thick
igneous rocks (Figure 1.2). The mean value of Vp/Vs  clays, clay/sand sequences over chalk, mudstone overly-
1.89 in the more heavily jointed rocks (perhaps a rock ing granodiorite bedrock, and landfill), using a combin-
quality Q  1–10), and the mean value of Vp/Vs 1.80 ation of both compressional and shear wave seismic.
in sparsely jointed rocks (perhaps a rock quality Q  The authors of course admit that there is a higher
10–100), can be used to calculate dynamic Poisson’s ratios level of effort required to conduct multi-component
of 0.30 and 0.28 respectively. seismic, requiring a three-component source configur-
As lower rock quality Q-values are approached in shear ation, and three-component geophones, but otherwise
zones and faulted zones (e.g. rock quality Q  0.1), the conventional multi-channel seismic recording systems,
ratio Vp/Vs increases to about 2.0, corresponding to a and PC-based processing software. Obviously the sur-
calculated value of dynamic Poisson’s ratio of about 0.33. veys are more difficult and more time consuming than
The corresponding Q-value (from equation 1.4, using compression-wave refraction or reflection, but the level
minimum Vp data from Figure 1.2) is indeed about 0.1. of geophysical information is that much more useful.
Extremely low Q-values, for example Q  0.01–0.001 They varied the acquisition geometry to optimize
(when Vp  1.5–2.5 km/s) will be needed before results. They found that under all the conditions, shear-
dynamic Poisson’s ratio values become as large as 0.45 waves penetrated with less attenuation than compres-
(as indicated for near-surface shear zones, in a later sional waves, also being unaffected by water saturation.
section of this chapter). Shear-wave reflections from shallow interfaces were in
Further basic equations linking Vp, Vs, dynamic some cases less affected by noise compared with the
Poisson’s ratio ( ), density ( ) and dynamic Young’s modu- equivalent compressional-wave reflections.
lus Edyn. are as follows (Darracott and Orr, 1976): They offered the following simple explanation of
why shear-waves offer better vertical resolution than
⎛ Edyn.(1 ) ⎞⎟
1
2 compressional-waves, particularly in shallow, unconsoli-
Vp  ⎜⎜⎜ ⎟⎟ (1.7) dated sediments. The dominant reason is that the
⎜⎝ (1  )(1 2 ) ⎟⎟⎠ shear-wave velocities in such cases, are only a fraction
(sometimes less than one fifth) of the compressional-wave
1 velocities. This results in very small wavelengths, despite
⎛ Edyn. ⎞⎟ 2

Vs  ⎜⎜⎜ ⎟⎟

(1.8) the fact that the dominant frequency of shear wave data
⎝⎜  2(1  ) ⎠⎟ is generally lower than is the case for compressional wave
data. In order to obtain the same level of resolution with
P-waves, energy of very high dominant frequency has to
3( Vp / Vs )2 4 (1.9) be generated, and this is correspondingly more attenuated
Edyn.  Vs2  
( Vp / Vs )2 1 in the low seismic Q sub-surface.
Barton-01.qxd 8/17/06 5:19 PM Page 8

8 Rock quality, seismic velocity, attenuation and anisotropy

(a)

(b)

Figure 1.3 a) Shear-wave velocities (km/s) and Vp/Vs ratios versus depth. b) Shear-wave velocities (km/s) and dynamic Poisson’s ratio versus
depth for a clay-over-mudstone-above-basement sequence, with an interpreted water table at 4.7 m depth. Results of multi-
component seismic at one of four shallow sites described by Dasios et al., 1999.

Since shear-waves are not attenuated at the water table, of Vp, Vs and the dynamic Poisson’s ratio, in the presence
and are little affected by changes in fluid saturation, they of a water table, and also show the effect of increasing
can more easily detect lithological changes with corres- depth in a uniform sedimentary rock. At their site, thin
pondingly less ambiguous velocity contrasts. The authors clays overlayed mudstone, with a basement of granodi-
found that under conditions of full water saturation, orite at 70 m depth.
P-wave velocity contrasts between lithologies were small, Figure 1.3 shows multi-component plots of shear-wave
whereas the shear-wave velocities reflected the true litho- velocity versus the ratio of Vp/Vs, and of shear-wave
logical changes. velocity versus (dynamic) Poisson’s ratio, each as a func-
In this brief summary, the results from a site they inves- tion of depth to the basement rock at 70 m depth.
tigated in Crewekerne, Dorset will be reproduced. Their Analysis of the P-wave first arrivals gave velocities of
results give a good illustration of the subtle interaction 496 m/s and 1,766 m/s for the unsaturated and saturated
Barton-01.qxd 8/17/06 5:19 PM Page 9

Shallow seismic refraction, some basic theory, and the importance of rock type 9

layers, and indicated a depth of 4.7 m for the top of the 1.5 Basic estimation of rock-type
saturated zone. The dynamic Poisson’s ratio showed a and rock mass condition, from
small decrease close to the surface as a result of the con- shallow seismic P-wave velocity
solidation, then a sharp increase as a result of the water
saturation, followed by a steady decrease with depth When first investigating the bedrock for suitability for
within the uniform water-saturated mudstones. near-surface tunnelling or other relatively shallow con-
As expected from the theoretical calculation of the struction in rock, the preliminary use of shallow refrac-
dynamic Poisson’s ratio, there is a certain accentuation tion seismic is very typical, where surface access (including
of the above trends for the case of the ratio Vp/Vs, except noise) do not present major problems. As a minimum,
that this ratio reduces faster at shallow depth (rather the information gives a Vp – depth profile of inestimable
than when deeper), due to the strong Vp gradient. The value for further planning of the sub-surface investiga-
authors point out that the P-wave velocity was not avail- tion, in particular the optimal siting of boreholes for
able at the greatest depths, due to attenuation, so the future core-logging and permeability testing.
data was extrapolated to 70 m. There were indications of Figure 1.4 reproduces four examples of shallow refrac-
shear-wave anisotropy in the uppermost meters of clay, tion results from Sjøgren 1984, demonstrating the help-
but whether due to desication fractures or some form of ful information about the location, width and depth of
layering is not certain. zones of lower velocity. Later in this chapter, and in sub-
Although outside the usual range of Vp/Vs (about 1.4 sequent chapters, we will be seeing the many ways of
to 2.0) for water saturated rocks, it is of interest to see interpreting such velocities in terms of rock quality and
details of the development of Vp/Vs ratios in unconsoli- degree of fracturing, each tempered by the effect of rock
dated (e.g., subsea) sediments, by noting the progres- type, density, porosity, depth (or stress level), and of
sion from soft soils through compacted soils, to rocks. course the possible anisotropy (or directional depend-
Hamilton, 1979 gave comprehensive Vp-depth, Vs-depth ence) of the result in relation to an anisotropic jointing
and Vp/Vs-depth data for silty clays, turbidites and mud- frequency, and horizontal stress anisotropy.
stones to 1 km depth. Vp values increased slowly from The seismic refraction survey provides numerous
about 1.5 to 2.3 km/s as porosity reduced with increased depth to bedrock and quality of bedrock assessments at
depth, while Vs increased rapidly from only .05 km/s a small fraction of the cost and time needed for drilling.
close to the surface, to 0.15 km/s at depth. Vp/Vs ratios Depths are given at the impact points (hammer or shal-
therefore reduced very rapidly from double figures down low explosive source) and at the detector points (geo-
to about 2.5 at 1000 metres depth. The dynamic values phones or 3D seismometers), so a close spacing of
of Poisson’s ratio decreased, as a result, from about 0.49 detectors gives the equivalent of a large number of sound-
near the surface to 0.41 at 1000 m depth. ings or borings.
In connection with these high values of Poisson’s Sjøgren, 1984, gives the example of 5 m detector and
ratio for sediments, it is significant to note the relatively 25 m source separations for a 10 m deep bedrock inves-
high values of (dynamic) Poisson’s ratio that tend to be tigation. A 100 m profile gives the equivalent of 250 m
recorded in shear zones and fault zones at much shal- of soundings, and a complete distribution of relative
lower rock engineering projects. Gardener, 1992, calcu- quality beneath the profile. With the 10 m source and
lated the values of Poisson’s ratio from Vp and Vs 50 m detector separations needed for a deeper survey to
measurements at the Transfynydd power station in Wales, 50 m depth, the equivalent of 650 m of soundings per
estimating 0.45 for the shear zones, where the Vp velocity 100 m profile is given.
range was 1.6–2.7 km/s. The knowledge and experience of the geophysical team
The higher values of Poisson’s ratio for shear zones is essential in setting out optimal profiles in relation to the
have psuedo-static parallels to the special feature of geology and structural geology, in particular in relation to
heavily jointed rock masses, which can show ‘expansion anisotropic, layered media, and in relation to fault and
ratios’ or pseudo-Poisson’s ratios far in excess of 0.5, and shear zones. ‘Correct’ interpretation of the calculated
even in excess of 1.0 as (shear) failure is approached information cannot be divorced from the geology, since a
(Barton and Bandis, 1982). Elastic continuum theory given velocity (Vp or Vs or dynamic Poisson’s ratio) is not
is of course ‘violated’ by the shear displacements tend- unique to any one material but part of a scale or gradation
ing to occur on the failing joint surfaces. in the specific geological profile at the site, and reflects
Barton-01.qxd 8/17/06 5:19 PM Page 10

10 Rock quality, seismic velocity, attenuation and anisotropy

Figure 1.4 Seismic refraction results illustrating the wealth of potential information obtained concerning near surface conditions. Sjøgren, 1984.
Barton-01.qxd 8/17/06 5:19 PM Page 11

Shallow seismic refraction, some basic theory, and the importance of rock type 11

Figure 1.5 Typical ranges of Vp for sediments and for little weathered, moderately fractured rocks. Sjøgren, 1984.

various ‘environmental’ factors acting on each rock A similar range of values from the SSDS Project
domain, as will be demonstrated in subsequent chapters. granites in Hong Kong (Gardener, 1992) gives a useful
The later geological and rock quality interpretation qualitative impression of variations caused by weather-
of core recovered from boreholes drilled close to the ing and jointing in the same rock type.
seismic profiles is the domain of engineering geologists,
who besides identifying rock type, will perform careful Table 1.2 Typical range of Vp (km/s) for granite (Gardener, 1992).
logging of RQD, joint or fracture spacing, joint rough- Decomposed granite (soil) 1.6–1.8
ness and discontinuity mineral filling identification (or Fracture zones 2.8–3.5
testing). The performance of rock quality characteriza- Jointed granite 3.5–4.5
tion of drillcore is also standard practice for civil engin- Intact granite 4.5–6.5
eering and many mining projects, using the Q-value
(Barton et al., 1974, Barton, 2002) and RMR
(Bieniawski, 1989) as a minimum. Although these two At the hazardous second Severn Estuary crossing
methods have similarities, and common goals, there are between England and Wales, tidal currents are so strong
differences, and care is needed in converting Q to RMR that 85% of the crossing had continuous rock outcrops
and visa versa, e.g. Barton, 1995. between low and high tide. Sonar buoys and bottom
Typical ranges of velocities for relatively competent (lit- drag cable gave the following relatively tight ranges of
tle weathered moderately fractured) rocks are given in velocities for five rock types that were confirmed with
Figure 1.5. Much lower velocities, covering most of the boreholes, enabling the rocks to be identified across
lower diagonal space between 1 km/s and 6 km/s are seen the site.
with extremes of weathering, jointing and fault related
fracturing. The following is an example of the effects of Table 1.3 P-wave velocities at the Second Severn Crossing
weathering for just one rock type, from Sjøgren, 1984: (Gardener, 1992).

Average velocity Velocity range


Rock description (km/s) (km/s)
Table 1.1 Typical range of Vp for gneiss (Sjøgren, 1984).
Triassic mudstone 2.1 1.7–2.3
500 m/s Soil (above water table) Triassic siltstone 2.4 2.2–2.6
1700 m/s Highly weathered biotite gneiss Triassic sandstone 2.6 1.8–3.1
2800 m/s Weathered biotite gneiss Carboniferous siltstone
3500 m/s Jointed biotite/granitic gneiss and sandstone 3.0 2.5–4.4
4900–5400 m/s Sound biotite gneiss Carboniferous sandstone 4.0 3.0–4.4
Barton-01.qxd 8/17/06 5:19 PM Page 12

12 Rock quality, seismic velocity, attenuation and anisotropy

Table 1.4 A list of typical Vp and Vs values from Press, 1966.

Material Vp (m/s) Vs (m/s)

Alluvium 500–2100
Clay 1100–2500
Sand 200–2000
Glacial Till 400–1700
Sandstone 1400–4500
Shale, Slate 2300–4700
Limestone
Soft 1700–4200
Hard 2800–6400
Crystalline 5700–6400
Dolomite 3500–6900
Granite, Granodiorite 4600–6000 2800–3200
Diabase 5800–6000
Gabbro 6400–6700 3400–3600
Basalt 5400–6400 2700–3200
Schist 4200–4900 2500–3200
Figure 1.6 Typical ranges of Vp for common rock types. Griffiths Gneiss 3500–7500 3300–3700
and King, 1987. Water 1450
Air 335

Griffiths and King, 1987, also give typical Vp ranges Table 1.5 P-wave velocities for different horizons in a river bank
for common rock types. These are reproduced in Figure terrace (Sen and Bandyoadhyay, 1990).
1.6, as a source of cross-referencing. Fractured, faulted 0.7–0.8 km/s Clayey and silty soil with pebbles
and heavily jointed zones extend the six major ranges 1.7–1.8 km/s Pebbles/cobbles in silty sand matrix
for these rocks far to the left on occasion. Note the 2.1–2.2 km/s Cobbles/boulders in silty sand matrix
extremely high velocities of the dense, ultramafic rocks,
which lie outside the common range of 1 to 6 km/s.
Table 1.6 P-wave velocities for phyllites at a Himalayan dam site
A comprehensive set of in situ seismic Vp values, and
(Dhawan et al., 1983).
some Vs values for common rock types, is also shown in
Table 1.4. The data are given by Press, 1966. Overburden/weathered phyllites 925–1200
The wide ranges of velocity for sandstone, shale, lime- Unweathered phyllites 2520–4500
stone and dolomite are mainly due to the wide ranges of
porosity (and density) for these materials. The surpris-
authors also give data for phyllites, which do not appear
ingly high range for gneiss is due to the wide range of
on the foregoing figures or tables of Vp data.
mineralogical composition (and density) for this rock.
The marked variation of velocities that are measured
in superficial deposits (0.5 to 2.0 km/s in Figure 1.5) 1.6 Some preliminary conversions
are partly caused by location either above or below the from velocity to rock quality
water table, as shown by Sjøgren’s 1984 data set. The
list given in Table 1.4 shows 0.2 to 2.0 km/s just for Due to the seismic ‘visibility’ of jointing in the upper 25
the case of sand, mostly for this reason. to 30 metres or so, Sjøgren et al., 1979 and Sjøgren, 1984
The following is perhaps a good example of the influ- and others, have been able to record significant correl-
ence of particle size in river born sediments. ations between Vp, RQD and joint frequency. These
The last line of Table 1.5 (for cobbles and boulders) authors compared the results from a total of 113 km of
differs significantly from the range 1.3 to 1.9 km/s for P-wave profiles from fifteen sites, with the results of
‘river boulders’ given by Dhawan et al., 1983, presum- 2.85 km of core-logging from seventy four drill holes at
ably due to differences in porosities. In the latter case, eight of the hard rock sites. (They also had 5 km of
the ‘silty sand matrix’ is presumably absent. These last S-wave surveys at five of the sites).
Barton-01.qxd 8/17/06 5:19 PM Page 13

Shallow seismic refraction, some basic theory, and the importance of rock type 13

The range of rocks occurring at the measured locations, Figure 1.7 reproduces Sjøgren et al., 1979 data in the
mostly in Norway, were: amphibolite, granite, gneiss, readily absorbed format used by the authors. Mean values
meta-anorthosite, pegmatite, porphyry, quartzite and for all the parameters apply. The Edyn modulus is the
mylonite. The authors were careful to emphasise that same as the symbol () used to denote the uniaxial or
the correlations they derived between P-wave velocity axial modulus in equation 1.1.
and jointing descriptions such as mean RQD and mean Since their measurements were shallow, the effect of
frequency F(m 1) were relevant only to unweathered stress-induced joint closure was minimised. They also
igneous and metamorphic rocks, and generally for the effectively removed other sets of variables by generally
upper 20 to 30 metres. recording correlations for hard and almost unweathered
igneous and metamorphic rocks. The usual variables of
depth, porosity, uniaxial compressive strength and dens-
ity were therefore largely removed.
A hard rock, near surface correlation of Vp and the
rock quality Q-value was derived by Barton, 1995, on
the basis of trial and error fitting to cases known to the
writer and also Q-logged. (These will be described in
Chapter 5). In Figure 1.8, the important effects of poros-
ity, uniaxial strength and depth are ignored, as for the
Sjøgren et al., 1979 data. Note that the RQD and F/m
mean data have been smoothly extrapolated beyond
both ends of the Sjøgren et al., ‘data-base’ represented
by Figure 1.7.

1.7 Some limitations of the


refraction seismic velocity
interpretations
Figure 1.7 Mean values of physical and dynamic properties for
hard, unweathered igneous and metamorphic rocks,
The seismic refraction method has some important limi-
based on shallow refraction seismics. Sjøgren et al., 1979.
Q-Scale added by Barton, 1995.
tations. One is that the horizontal or sub-horizontal ray

Figure 1.8 RQD and Fm 1 trends from Sjøgren et al., 1979. Q-scale, (as also in Figure 1.7) added by Barton, 1995. These results and
approximate correlations to seismic-frequency Vp are relevant for hard, low porosity, unweathered, near-surface rock masses.
Barton-01.qxd 8/17/06 5:19 PM Page 14

14 Rock quality, seismic velocity, attenuation and anisotropy

paths record only the upper part of each seismic layer. A


thin high-speed layer can mask underlying material, while
a low-velocity intermediate layer will not be recognised
for similar reasons. Depth calculations to underlying
refractors will be erroneous. The hidden low velocity
zones can of course be detected by up-hole shooting from
a borehole to the seismic spread (i.e. VSP), or by inspec-
tion and index testing of core, if available.
A useful review of refraction seismic methods and com-
monly used methods of interpretation is given in Whitely,
1990. Citing Sjøgren, 1984 that low velocity zones are
consistently interpreted as being shallower than the bore-
hole confirmation, Whitely, 1990 went on to compare
three interpretation methods of the simple yet frequently
occurring situation shown in Figure 1.9. The three solu-
tions were obtained by independent geophysics practi- Figure 1.9 Three independently derived depth estimates for a low
tioners. It is clear that control drilling of identified velocity zone detected by refraction seismics. Whitely,
features must be made before making important deci- 1990.
sions such as minimum depth of cover over a sub-sea
tunnel.
Stapledon and Rissler, 1983 emphasised the follow-
ing potential shortcomings of the seismic method, as engineering projects may be far different from these
related to near-surface investigations for civil engineer- apparently uniform conditions. Yet even the ‘simple’ site
ing projects. may contain velocity anomalies, which reduce the image
quality. The time horizons suffer ‘push-down’ beneath
1. Minor geological defects such as weathered seams or slow-velocity anomalies, and ‘pull-up’ beneath fast
minor faults may govern the engineering behaviour velocity anomalies.
of a site, especially if their orientation is unfavour- The geological ‘reality’ shown on the left in Figure 1.10
able. The seismic method ‘generally is unable to contrasts nicely with the equivalent post-stack seismic
detect such defects’. Is this pessimism justified or is time section, shown on the right. There is significant
it entirely correct? distortion of layer ‘horizons’, and perhaps surprisingly
2. P-waves are first to arrive at the geophones ‘and must persistent follow-through to depth. Armstrong et al.,
take the shortest path through the best rock’. Will 2001 describe a method for compensating for these
this mean that they do not represent the average discrete overburden velocity anomalies. (Their warning
local quality but the best local quality? about possible miss-calculation of reserves, or miss-
positioning of infill wells, or by-passing of incremental
Both shallow and deep refraction seismic are subject reserves in relation to petroleum engineering, can clearly
to velocity anomalies, causing time-distortion and there- apply in other contexts to shallow civil engineering
fore depth anomalies, as we have seen in the above shal- projects, where for example there are buried sediment-
low example. Although the following case of Armstrong filled channels that have been known to plague certain
et al., 2001 is strictly speaking a deeper reservoir case, it tunnel projects, due to high water storage, and
provides such a nice illustration of the hazards of depth ‘constant’ inflow pressure).
interpretation that it will be referred to in this first Also concerning the possibility of depth estimation
chapter. error, it was emphasised by Bradford and Sawyer, 2002,
As the authors point out, the overburden above a in the context of shallow seismic reflection measurements,
hydrocarbon prospect is often more or less horizontal or that larger depth and layer thickness estimates would
perhaps with gently dipping sedimentary layers. It is gen- occur when using conventional velocity analysis, if in the
erally paid little attention, in relation to the focussed presence of the extreme velocity gradient close to the dry-
investigation of the reservoir target at depth. Clearly, saturated transition, (usually fairly close to the surface in
the overburden at typical shallow refraction sites for civil temperate climates).
Barton-01.qxd 8/17/06 5:20 PM Page 15

Shallow seismic refraction, some basic theory, and the importance of rock type 15

(a) (b)

Figure 1.10 Schematic cross-section through an overburden containing velocity anomalies of different geological age. The idealized vertical
section is shown on the left, and the contrasting, time-horizon distortions, post stacking, are shown on the right. Armstrong,
et al., 2001.

The authors indicated (as also seen earlier), that the Sjøgren, 2000 expressed concern about the useful-
P-wave velocity could increase by a factor of four or ness of the GRM method for near-surface geotechnical
more at this transition, in the case of unconsolidated investigations, where details of the various overburden
sediments changing from dry, through partial satur- layers are required since they may have important con-
ation, to full saturation – which may occur just above sequences for the subsequent geotechnical design.
the water table, due to the action of capillary forces. A A relatively more recent technique for modelling of
large velocity gradient (e.g. 400 m/s to 1600 m/s from 1 travel times and travel time inversion in refraction seis-
to 10 m depth) apparently violates many of the assump- mics is the so-called Eikonal solver. In principle, this
tions made in conventional reflection data processing involves the calculation of travel times on a regular
schemes. velocity grid. Early versions, originating from the late
In a recent paper, Sjøgren, 2000, evaluated several 1980s were restricted to a plane topography for the
standard methods for interpreting travel time curves. recording surface.
He utilised the ABC method (originating from 1931), Lecomte et al., 2000, describe a first order Eikonal
the ABEM correction method (early 1950s, detailed in solver that can incorporate the exact topography of the
Sjøgren, 1984), the mean-minus-T method (mid-fifties, surface terrain, and any arbitrary lateral variation of
also adopted in the ABEM method), and the Hales velocity. There is no restriction on the velocity contrast.
method (1958), in order to critically evaluate a more In effect, the model is built up layer by layer, with the
recent (1980) generalized reciprocal method (GRM) of refractor imaging, and the velocity mapping being per-
Palmer, in particular Palmer, 1991. formed for each identified refractor at a time, as seen
Barton-01.qxd 8/17/06 5:20 PM Page 16

16 Rock quality, seismic velocity, attenuation and anisotropy

(a)

(b)

(c)

Figure 1.11 An arbitrarily chosen model for demonstrating a new method of refraction seismic inversion. Lecomte et al., 2000 used this
synthetic model for demonstrating some basic elements of the Eikonal solver, which calculates the travel time of the fastest
wave at any point of a regular velocity grid, using the head waves generated in refraction seismics. The three stages of model-
ling shown here are: a) Wave-fronts and raypaths at receivers along the surface, when considering the whole model (but
minus the acoustic wave velocity in the air, which was omitted for clarity. b) Wave-fronts and raypaths at receivers along the sur-
face, when masking the deepest layer. c) Wave-fronts and raypaths at receivers along the surface, when masking the two deepest
layers.

schematically in one of their illustrative figures, repro- Consider for example the jointed chalk marl at the
duced here in Figure 1.11. Chinnor Tunnel in the UK, where jointing in this weak
material were seismically closing at about 15 metres depth,
to give a stable 1.6 km/s field velocity, despite changes in
1.8 Assumed limitations may the degree of jointing (‘rock’ Hudson et al., 1980). This
hide the strengths of the can be contrasted with the jointed gneiss at the Gjøvik
method cavern in Norway, which gave a continuous rise in vel-
ocity from 3.5 to 5.5 km/s in the first 50 metres depth due
In later chapters the numerous factors influencing seismic to increased stress, yet had almost unchanged rock qual-
velocities such as joint frequency, porosity, rock (and joint ity. The joint frequency, RQD and rock quality Q-values
wall) strength, density, depth, stress, stress anisotropy, did not show an improvement with depth (Barton et al.,
degree of saturation, type of saturating fluid, will each be 1994). Both the above observations could be interpreted
reviewed. This will be done in order to emphasise that as ‘limitations’ of the method. In fact they are demonstrat-
the seismic method has numerous complications, but also ing specific and quite logical physical laws of behaviour.
inestimable advantages, and that some of the assumed The latter is an example of the need to interpret seismic
shortcomings can be due to misinterpretations (often an velocity with knowledge of depth and/or stress level, since
over-simplification, or perhaps even a pre-conceived a rock quality Q-value increase from perhaps 1 to 100
opinion about ‘seismic’ limitations). might otherwise be assumed in these first 50 metres,
Barton-01.qxd 8/17/06 5:20 PM Page 17

Shallow seismic refraction, some basic theory, and the importance of rock type 17

based on a ‘hard rock’, ‘stress-less’ interpretation (i.e. round of advance would also be demonstrating a
equation 1.4, and Figure 1.8). ‘reversed’ Vp-quality behaviour. However, the more dom-
The interpretation cannot be divorced from consider- inant effect of a reduced need for time consuming rock
ation of what is actually occurring in a rock mass as depth support would be consistent with the normal ‘high
increases, i.e. joint normal stiffness increases, joint aper- velocity  high quality’ concept. Since it is more logical
ture reduces, joint frequency reduces (usually), and the that Vp increases should accompany rock quality or
clay content in the joints reduces (usually). The list can Q-value increases, such cases of ‘reversed’ quality must
be lengthened considerably by adding that the stress be treated as separate, method-specific cases.
increases, the deformation modulus increases, the permea-
bility reduces, the pore pressure increases. Since there
may be changes of lithology with depth increase, the 1.9 Seismic quality Q and apparent
provision ‘usually’ should really be added to each of similarities to Q-rock
the above.
When considering the possibility of relating seismic A fundamental feature of the propagation of stress waves
velocity to a rock quality descriptor such as the rock in all materials is the absorption of energy and the result-
quality Q-value, another quite basic problem must ing change in shape of the transient waves. Non-linear
also be considered. A velocity of 2.5 km/s for massive friction has traditionally been assumed to be one of the
chalk marl of high porosity, as in the better parts of dominant attenuation mechanisms (Kjartansson, 1979),
the UK-France Channel Tunnel, will have entirely dif- but as will be discussed in Part II, there are a large num-
ferent engineering consequences to that of a regional ber of mechanisms that can explain the different degrees
fault-zone of the same 2.5 km/s velocity, crossing a of attenuation in different rock masses, including scat-
Japanese high-speed rail tunnel excavation, and delay- tering due to structure, and squirt flow.
ing progress by months. In this best quality chalk marl, Geophysicists commonly characterise seismic attenu-
near-world record speeds of (TBM) tunnel boring were ation by the seismic quality factor Q, which of necessity
achieved. has been termed Q seis throughout this book, to distin-
The natural velocity of the unjointed rock under in guish it from the rock mass quality Q. Intuitively Q seis
situ conditions (Sjøgren et al., 1979), and the contrast is related to, but no relation of, the rock mass quality Q
seen in low velocity zones is the main index of difficulty, of Barton et al., 1974, which is a ‘quality number’ also
since an order of magnitude reduction in the rock quality widely used in civil and mining engineering in the last
Q-value will generally accompany each 1.0 km/s reduc- several decades.
tion in seismic velocity, according to the simplified The seismic quality Q seis is often defined as the max-
model shown in Figure 1.8. imum energy (Emax) stored during a cycle, divided by
Low velocity and potentially high permeability zones the energy lost (E) during a cycle:
will be the natural focus of attention in most sub-surface
civil engineering projects. However, in a TBM (machine- 2
Q seis   E max (1.10)
bored) tunnelling project, there will be serious delays if E
there is too much high velocity rock, due to the slow
progress made in hard, sparsely jointed rock. This would For dry rocks, Q seis has been claimed to be independ-
give a ‘reversed’ Vp–rock quality indication, due specif- ent of frequency over a reasonable frequency range
ically to poor borability and the need for frequent (McKenzie et al., 1982). However, as will be seen in
cutter changes. Part II, Q seis is a remarkably sensitive indicator of
A Q-value based ‘Qtbm’ rock-machine quality factor anisotropy, and is frequency dependent in the case of
has been developed for this specific problem (Barton, saturated or partly saturated conditions. When attenu-
2000), which also allows for the fact that more jointing ation (Q 1seis) is high, Q seis obviously has a low value.
is good for progress – up to some limit, when other prob- Chapter 10 in Part II, will address Q seis in detail, and
lems may arise. (More tunnel support is needed, there also draw some tentative parallels between ‘Q and Q’,
could be gripper-setting problems, the cutter head could via a common link to deformation properties.
even be jammed). Intuitively, the rock mass quality Q (of Barton et al.,
In the case of drill-and-blasted tunnels, the drillability 1974) which has a high value in high-modulus, high-
and blastability components of the cycle time for one velocity rock masses (i.e., Q  100, Vp  5.5 km/s,
Barton-01.qxd 8/17/06 5:20 PM Page 18

18 Rock quality, seismic velocity, attenuation and anisotropy

Emass  45 GPa) would seem likely to have high energy Further discussion of Q seis is given in Chapter 4, based
storage and low energy loss under such conditions, and on recent seismic attenuation tomography (see alterna-
therefore qualify for a high value of Q seis and a low tive tomograms). However, a much fuller treatment of
value of attenuation (Q 1 seis). Conversely, in a rock mass Q seis is given of both lab-scale and in situ scale attenuation
characterized by many joint sets with clay coatings and phenomena, in Part II, chapter 10, and in relevant parts
fillings on many joints (i.e., Q  0.1, Vp  2.5 km/s, of Chapters 13 and 15.
Emass  5 GPa), both low energy storage and high
energy loss per cycle would be expected. (Q seis is low
and attenuation Q 1seis is correspondingly high.)

View publication stats

You might also like