Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

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

net/publication/228920696

Experiments of joint acquisition of seismic refraction and surface wave data

Article  in  Near Surface Geophysics · August 2003


DOI: 10.3997/1873-0604.2003002

CITATIONS READS

67 866

4 authors:

Sebastiano Foti Luigi Sambuelli


Politecnico di Torino Politecnico di Torino
180 PUBLICATIONS   4,675 CITATIONS    161 PUBLICATIONS   1,309 CITATIONS   

SEE PROFILE SEE PROFILE

Laura Valentina Socco Claudio Luciano Strobbia


Politecnico di Torino realtimeseismic
188 PUBLICATIONS   3,265 CITATIONS    113 PUBLICATIONS   2,053 CITATIONS   

SEE PROFILE SEE PROFILE

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

Geotechnical post-seismic reconnaissance View project

Understanding wave propagation in nonlinear media View project

All content following this page was uploaded by Luigi Sambuelli on 19 September 2020.

The user has requested enhancement of the downloaded file.


Near Surface Geophysics, 2003, 119-129

Experiments of joint acquisition of seismic refraction


and surface wave data

Sebastiano Foti1, Luigi Sambuelli2*, Valentina L. Socco2 and


Claudio Strobbia2
1
Dipartimento di Ingegneria Strutturale e Geotecnica, Politecnico di Torino, C.so Duca degli Abruzzi, 24 10129 Torino, Italy
2
Dipartimento Georisorse e Territorio, Politecnico di Torino, C.so Duca degli Abruzzi, 24 10129 Torino, Italy
Received February 2001, revision accepted April 2003

ABSTRACT

Dispersive Rayleigh waves, contained in conventional P-wave refraction records, can be used
to determine the shear-wave velocity profile. In the present paper, the synergies of a cross
interpretation of refraction and surface wave data are exploited using data collected at sites
where other geotechnical and geophysical information was available.
Selected examples are presented to emphasize the relative advantages and limitations of
the two techniques in the cases of hidden layer, velocity inversions and shallow water table.
Surface wave analysis was performed to estimate the experimental dispersion curves in
the f–k domain and the results of the dispersion curves inversion were compared with refrac-
tion results in terms of velocity profiles.
The experimental results prove that many advantages, in terms of resolution and reliabil-
ity, can be obtained with joint acquisition and cross interpretation of P-wave refraction and
surface wave data, without a significant increase in testing time.

INTRODUCTION Vs by means of surface wave analysis, utilizing the dispersive


nature of surface waves in heterogeneous media (Glangeaud
Seismic surveying techniques provide relevant data regard- et al. 1998). A variety of Surface Wave Methods (SWM) for
ing soil behaviour at a very low strain level for large portions shallow subsurface characterization has been proposed
of soil, tested in an undisturbed state, by using procedures using different testing configurations, processing techniques
that are time and cost effective. Since this is very useful for and inversion algorithms (Jones 1958; McMechan and Yedlin
geotechnical characterization of the shallow subsurface and 1981; Nazarian and Stokoe 1984; Gabriels et al. 1987; Al-
for assessing the dynamic site response, new techniques are Hunaidi 1994; Park et al. 1999; Rix et al. 2001).
being developed from traditional seismic techniques in Data obtained with other geophysical tests can be useful
order to provide reliable results in cases of complex distri- during the interpretation of surface wave data to improve
bution of physical parameters at an engineering scale the reliability of the inversion process. Joint inversion and
(Massarsch 1999). In particular, the estimate of shear-wave cross interpretation of one-dimensional methods have been
velocity (Vs) represents a very important information, since successfully implemented. In particular, the joint inversion
this parameter is directly related to soil stiffness. Vs profiles of Rayleigh wave dispersion curves and VES apparent resis-
can be estimated using shear-wave seismic methods, such as tivity curves has been shown to be very effective, solving
shear-wave refraction, but very often these techniques are some interpretation ambiguities inherent in the two tech-
affected by experimental difficulties related to the need for niques (Hering et al. 1995; Misiek et al. 1997; Comina et al.
specific equipment (polarized sources and receivers) and to 2002). VSP and downhole test data were used to provide
the inadequacy of the shear-wave data quality. geometrical constraints for surface wave interpretation,
Recently, much interest has arisen in the determination of increasing the reliability of the results (Godio et al. 2001).
Two-dimensional results from GPR and ERT indicate the
*luigi.sambuelli@polito.it presence of lateral variations that prevent the application of

© 2003 European Association of Geoscientists & Engineers 119


120 S. Foti et al.

surface wave methods due to the one-dimensional model be assumed for the interpretation of the refraction data.
adopted for the interpretation of the latter (Socco et al. Under this assumption, the acquisition of refraction data
2001). The positions of the bedrock and the watertable are can be performed with only two end-off shots: this allows
also extremely useful as constraints for the inversion the identification of plane, possibly slightly dipping, layers.
process (Foti and Strobbia 2002). In this framework, the presence of lateral variations can be
Cross interpretation of surface wave and body wave seis- recognized from traveltime curves, but cannot be interpret-
mic data is of particular interest. SH reflection and refrac- ed. On the other hand, any dip of refractors can be detect-
tion results can be quantitatively compared with surface ed, giving warnings about the inversion of SWM, which
waves so that an accurate subsurface model can be obtained. assume a model without dip. It is therefore implicit that the
Since surface wave acquisition and processing are prefer- joint acquisition of surface wave and refraction does not
ably performed with multistation approaches (Foti et al. exceed the SWM limitations due to the assumption of a one-
2001; Foti 2002), P-wave refraction data and surface wave dimensional model.
data can be acquired simultaneously (Shtivelman 1999). It is With the above assumption, the interpretation of refrac-
therefore interesting to explore the advantages of joint tion data leads to a vertical velocity profile that it is compa-
acquisition and cross interpretation of the two data sets. rable with the result of SWM.
Indeed, surface wave methods overcome some inherent lim- The survey design for these two seismic techniques is
itations of the refraction technique caused by refraction mainly based on the required depth of investigation and on
equivalence such as a hidden layer, gradual velocity varia- spatial resolution. For seismic refraction, the modelling of
tions, velocity inversions and the blinding effect of a shallow the P-wave traveltimes and the estimation of the expected
water table. On the other hand, refraction results provide crossover distances of the refractors allow the correct defi-
very useful information for surface wave interpretation: nition of the profile length and of the geophone spacing. The
constraints for surface wave inversion (e.g. the bedrock source is then chosen according to the overall distance.
position), indication of the water table depth (which allows Acquisition parameters are consequently chosen to ensure
a good estimation of Poisson’s ratio in surface wave model- adequate time resolution and recording windows, so that a
ling) and a warning in the case of lateral variation of veloc- reliable first-break picking is obtained. For the surface
ities (e.g. topography of refractors). When the two tech- wave, preliminary modelling allows the estimation of mini-
niques give independently reliable results that are mutually mum and maximum wavelengths, and the layout geometry
confirmed, the knowledge of the values of both Vp and Vs is therefore set according to considerations related to spatial
allows the estimation of mechanical properties and porosity aliasing, resolution and Rayleigh mode separation (Foti et
of the soil deposit (Foti et al. 2002a). al. 2002b). In some cases, non-equispaced arrays are used in
In this paper a series of joint acquisitions of seismic order to extend the frequency range in which the dispersion
refraction and surface wave data are presented in order to curve can be interpreted. As far as the sampling parameters
demonstrate the relative advantages and limitations of the are concerned, the first requirement is that the time window
two techniques in different situations and to emphasize must be long enough to sample the whole wave train at the
their possible synergies. furthest receiver. Furthermore, a wide time window pro-
duces a good resolution in frequency, while the sampling
rate can be quite low because high-frequency components
THE METHODS are not of interest.
Survey design for refraction and SWM often leads to
Acquisition very similar layout geometries and therefore the same array
Since the model used for SWM interpretation is a stack can be used to acquire data for the two tests. Moreover, ver-
of linear elastic homogeneous layers, the investigated site tical geophones with rather low frequency (usually around 5
should fit this assumption in order to obtain reliable results. Hz) can be adequate for both the techniques. On the other
The acquisition layout for SWM is usually an array of equi- hand, contrasting requirements have to be fulfilled for the
spaced vertical geophones with two end-off shots. The two acquisition parameters and therefore separate acquisitions
shots are usually performed in order to verify that the with different sampling rates and recording windows are
respective data are similar. If significant differences typically performed.
between the two data sets are evident, the site cannot be
considered as one-dimensional and therefore SWM are not Processing and interpretation
suitable for its characterization. As mentioned above, if the investigated site fits a one-
For a joint interpretation of surface wave and refraction dimensional model (or at most a very slight dip is present),
data, the interpretative models for the two techniques must very simple processing and interpretation procedures can be
be comparable, hence a one-dimensional model must also adopted for refraction data. After first-break picking, the

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
Experiments of joint acquisition of seismic refraction and surface wave data 121

TABLE 1 Acquisition parameters

Site Method Receiver spacing Profile length Source Sampling rate Recording window
1 Refraction seismic 2m 50 m Hammer 0.025 ms 51.2 ms
Surface waves 2m 50 m Weight-drop 2 ms 4096 ms
2 Refraction seismic - - - - -
Surface waves 0.5 m 12 m Hammer 1 ms 2048 ms
3 Refraction seismic 3m 75 m Hammer 0.050 ms 102.4 ms
Surface waves 3m 75 m Weight-drop 2 ms 4096 ms
4 Refraction seismic 2m 50 m Hammer 0.025 ms 51.2 ms
Surface waves 2m 50 m Weight-drop 2 ms 4096 ms

traveltime curves of the two end-off shots can be interpret- HIDDEN LAYER
ed with the intercept time method, thus obtaining a vertical
profile of the P-wave velocity. In the case of strong velocity contrasts (for example, a very
Surface wave processing is based on the estimate of the high velocity bedrock with a low velocity multilayer over-
dispersion curve of Rayleigh wave phase velocity, which cor- burden), waves refracted by a deeper layer may reach the
responds to the maxima of the f–k spectrum. Usually only surface before waves refracted by an intermediate layer. As
absolute maxima are considered but in cases where higher a consequence, the intermediate layer is not identified and
Rayleigh modes could play an important role, relative max- the bedrock depth is not properly estimated by seismic
ima are also sought in order to increase the information refraction techniques. SWM surveying can solve this so-
content of the dispersion curves. Zero padding in the space called hidden layer problem, which could severely affect
domain is usually applied to improve wavenumber resolu- refraction results.
tion (Foti et al. 2002b).
The interpretation is carried out on a selected branch of Case 1: A hidden layer
the dispersion curve. If the fundamental mode can be isolat- The site studied is located in Pontremoli (Tuscany, Italy),
ed, a weighted, damped least-squares inversion can be where the local geology suggests the presence of a fairly
adopted but, in the case of multimodal interpretation, a shallow bedrock. The seismic survey was carried out to
trial-and-error inversion is preferred. Indeed spatial sam- determine the dynamic site properties to be used for seismic
pling issues in a shallow subsurface characterization do not hazard evaluation and, therefore, an estimate of the bedrock
allow the experimental separation of Rayleigh modes in a depth and of the overburden characteristics were required.
complex layered situation, and mode superposition effects A geotechnical log was available about 200 m away from the
lead to the identification of an apparent dispersion curve geophysical survey location. It indicates the presence of a
(Foti et al. 2000), whose partial derivatives are not easily
computed. The density and Poisson’s ratio of the initial
model are fixed with particular attention to the water table
position (Foti and Strobbia 2002).
The software used for the interpretation is POLISURF
(Strobbia 2003), implemented in Matlab and based on the
algorithm for the resolution of the forward problem pro-
posed by Thomson (1950) and Haskell (1953) and modified
by Dunkin (1965). The software supplies both modal curves
and the apparent dispersion curve.

Experimental equipment and set-up


The acquisition was made using a 24-channel Mark VI
seismograph (ABEM) and 24 vertical 4.5 Hz geophones
(SENSOR). Two different impulsive sources were used: a
6 kg sledge-hammer and a 130 kg weight-drop system. The
former was used to generate high-frequency signals, while
the latter was necessary to guarantee a high S/N ratio at low FIGURE 1
frequency for SWM. Source type, array geometry and Case 1: experimental f–k spectrum showing the positions of the maxima associat-
acquisition parameters are reported in Table 1. ed with the Rayleigh experimental dispersion curve.

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
122 S. Foti et al.

FIGURE 2
Case 1: comparison between experimental and numerical dispersion curves.

FIGURE 3
Case 1: experimental velocity profiles.

FIGURE 4
Case 1: P-wave refraction seismo-
grams: direct and reverse shots.

FIGURE 5
Case 1: P-wave refraction travel-
times and best-fitting two-layer
model.

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
Experiments of joint acquisition of seismic refraction and surface wave data 123

FIGURE 6
Case 1: bedrock depth vs. velocity of the hidden layer under the assumption
of a single hidden layer with constant velocity.

very shallow zone of soft soils (about 2 m thick), underlain


by dense gravels. A sandstone bedrock, weathered in the
upper part, was found at a depth of about 8 m.
The SWM results showed good agreement with the geol-
ogy and with the information from the borehole. The seis-
mograms of the different shots were acquired separately,
allowing statistical evaluation of the coherence function for
pairs of receivers, which is a good indicator of the S/N ratio.
The absolute maxima of the f–k spectrum were used to esti-
mate the experimental dispersion curve for Rayleigh waves
(Fig. 1). In Fig. 2, the experimental dispersion is compared to
the numerical simulation for the last iteration of the inver-
sion process. The final result is the 5-layers-over-a-halfspace
model shown in Fig. 3.
The raw seismograms, for the two end-off shots of refrac-
tion, are shown in Fig. 4. The interpretation of the traveltime
curves (Fig. 5) suggested a two-layer model, which is shown
in Fig. 3. This result would locate a slightly dipping bedrock,
of fairly high velocity, at a depth varying between 3 and
3.7 m, but the high velocity contrast (330 m/s vs. 2620m/s)
between the two layers suggested the possibility of a hidden
layer with significant thickness.
No information about the existence and the properties of
this hidden layer can be obtained from the refraction survey,
but it is possible to calculate the thickness–velocity ratio of
such layer: this allows, at least, an estimate of the possible
error in the bedrock depth.
The maximum thickness of the possible hidden layer as a
function of its velocity is shown in Fig. 6: it refers to a single
layer whose refracted waves are, at any distance, preceded
by the arrival from the deep refractor or by the direct waves.
For each possible velocity (slope) of the intermediate layer,
the maximum thickness is computed such that the arrivals
from this layer are never first arrivals. Under this assump- FIGURE 7
tion of a single hidden layer, the error in the location of the Case 1: traveltimes for equivalent velocity profiles (multi-layer profile from
bedrock can be of the order of 50%. The situation is even SWM: two-layer profile from seismic refraction).
worse if the velocity of this blind zone is not constant but
increases with depth: the real depth of the bedrock can be determined by the SWM, can be compatible with the travel-
significantly greater. A simple modelling shows that a stack times observed in the field. The two traveltimes depicted in
of layers with velocities increasing with depth, like that Fig. 7 are equivalent, but correspond to very different strati-

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
124 S. Foti et al.

graphies: the two-layer model was obtained from the first In this situation the interpretation of the SWM tests is
interpretation of the refraction survey, while the multilayer somewhat more complex, indeed higher Rayleigh modes
model was obtained by scaling the Vs profile of the SWM play a very important role for surface wave propagation in
method assuming a Poisson’s ratio of 0.45 (Fig. 3). inversely dispersive media. Moreover the lack of spatial
The P-wave refraction survey proved to be inadequate at resolution in the usual applications makes it impossible to
the investigated site. Furthermore, the presence of the hid- separate the individual modes during the analysis and the
den layers, without other information, could only be sus- interpretation has to be done by considering the mode
pected. The results of the SWM, on the contrary, were in superposition effects evaluated using the apparent phase
good agreement with the available borehole data. velocity (Foti et al. 2000). The stiffness of the top layer
strongly affects the whole dispersion curve (Fig. 8), so even
if the characterization of the pavement system is not includ-
VELOCITY INVERSION ed in the scope of the testing program, information in the
high-frequency range is necessary for an adequate estimate
In general, if velocity inversions occur within the stratigra- of the shear-wave velocity profile.
phy, seismic refraction surveys are unable to detect low- The example reported below is part of a series of tests
velocity layers, while SWM are not affected by this problem. performed to evaluate the potential use of SWM for feasi-
In the following, two examples of velocity inversion are bility studies of directional drilling for pipeline positioning.
presented. At the first site, seismic tests were carried out in Trenchless technologies are becoming increasingly popular
an urban environment in the presence of a stiff top layer for underground service installation and they require infor-
(the road paving) and with a high level of background noise mation regarding the mechanical properties of the first 2 – 3
caused by human activity. The second site is a fluvial deposit m of soil. The site is located in the city of Biella (Italy). No
with a shallow water table where the velocity of shear waves detailed information is available for the stratigraphy at this
does not always increase with depth. site, but the drilling operation revealed the presence of a
fairly hard material at shallow depth (about 3 m below
Case 2: Velocity inversion in an urban area ground level).
Construction activity in an urban environment requires The design of the array geometry (see Table 1) was based
site characterization under very difficult conditions. Indeed, on the required depth of investigation (very shallow in this
very often space for the execution of tests is very limited case) and on the need for high-frequency information to
and noise levels are quite high, also because the traffic can characterize the stiff, thin top layer. To ensure an adequate
seldom be stopped. Moreover, tests very often have to be spatial sampling, a short array with a small geophone spac-
done in paved areas and seismic refraction cannot supply ing was chosen for the surface wave data acquisition. The
information at shallow depths below the very stiff top layer. data quality was very good, even though the acquisition was
SWM can provide a shear-wave velocity profile in this situ- made in an extremely noisy environment, due to intense
ation because measurements are not very sensitive to back- traffic nearby. The very fast direct wave travelling in the top
ground noise and the pavement system can be adequately layer made the seismic refraction survey totally blind below
modelled. the pavement system.

FIGURE 8
Effect of different thin, stiff top lay-
ers on the Rayleigh-wave apparent
dispersion curve.

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
Experiments of joint acquisition of seismic refraction and surface wave data 125

FIGURE 9
Case 2: comparison between experimental and numerical dispersion curves.

As mentioned above, a multimodal approach is needed


for the interpretation of the SWM in this case. Due to the FIGURE 10
presence of the stiff top layer, higher Rayleigh modes are Case 2: shear-wave velocity profile.
dominant at high frequencies. Energy partition between dif-
ferent modes of propagation is not easily evaluated and the Case 3: Velocity inversion with a shallow water table
modes cannot be isolated during the analysis. The approach The site is situated along the banks of the River Po
used for the interpretation is based on mode superposition (Italy), and is characterized by recent sandy deposits on a
and it computes the apparent phase velocity by considering sequence of fluvial sediments (clay layers over dense sands
the same receiver geometry as used for field testing. and gravels). The uppermost materials are characterized by
The experimental and numerical dispersion curves are very low seismic velocities (P-wave velocity of 200 m/s and
shown in Fig. 9. Both the Rayleigh modes and the apparent S-wave velocity of less than 100 m/s). The aim of the testing
phase velocity are shown for the numerical simulation, was the identification of clayey layers, which prevent seep-
showing that the fundamental Rayleigh mode is dominant age during floods.
only for frequencies less than 120 Hz. In the high-frequency Due to the vicinity of the river, the water table is quite
range the propagation is dominated by higher modes, with a shallow. In saturated porous media, the P-wave velocity is
continuous transition characterized by an asymptotic value strongly dependent on the fluid mechanical properties, and
associated with the stiff top layer. The inversion process was it is therefore less sensitive to the solid skeleton properties.
conducted using a trial-and-error approach; indeed, the For this reason, P-wave refraction does not provide useful
dependence of the apparent dispersion curve on higher geometrical and mechanical information for saturated
modes and on the testing geometry prevents the use of local materials. The only information that can be inferred from P-
search algorithms because the partial derivatives are not wave refraction in this situation is the water table depth. On
sufficiently stable. The estimated shear-wave velocity profile the other hand, surface wave data, which are less influenced
is shown in Fig. 10. The presence of a stiffer material starting by the degree of saturation, allow soil characterization, also
2.5 m below the ground surface is compatible with the below the water table. The information about water-table
observation made during the drilling mentioned above. depth provided by the P-wave refraction leads to a more

FIGURE 11
Case 3: P-wave refraction seismo-
grams: direct and reverse shots.

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
126 S. Foti et al.

FIGURE 12
Case 3: P-wave refraction traveltimes and best-fitting two-layer model.
FIGURE 15
Case 3: comparison between experimental (dots) and numerical (solid)
dispersion curves.

FIGURE 13
Case 3: time-domain data for SWM (weight-drop source).

FIGURE 16
Case 3: shear-wave velocity pro-
file from SWM.

FIGURE 14
Case 3: experimental f–k spectrum showing the positions of the maxima asso-
ciated with the Rayleigh experimental dispersion curves.

FIGURE 17
Case 3: example of SH-wave refraction seismograms.

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
Experiments of joint acquisition of seismic refraction and surface wave data 127

velocities. A borehole log at about 1 km from the investi-


gated site shows a similar profile. This interpretation is in
good agreement with the results of several other tests per-
formed along the same profile. A vertical electric sounding
showed the presence of a conductive layer from 9 to 16 m
depth (Comina et al. 2002). A Wenner resistivity tomogra-
phy confirmed the interpretation of SWM and VES,
although the ERT did not show an adequate depth of
investigation and so did not provide any information about
the bottom end of the clayey layer. It is also interesting to
compare the results of SWM with an SH refraction tomog-
raphy. The VSH data were acquired by using a polarized
source and performing 15 shots along the profile. 21 hori-
FIGURE 18 zontal receivers (Swyphone™ prototypes, Sambuelli et al.
Case 3: 2D map of shear-wave velocity from tomographic inversion of SH 2001) were used (for acquisition parameters, see Table 1). A
refraction data compared with the shear-wave velocity profile from SWM. sample seismogram shows evidence of the refraction events
(Fig. 17).
reliable surface wave inversion, supplying an estimate of The tomographic interpretation was carried out using
Poisson’s ratio, typically assumed a priori in the inversion Rayfract® (Intelligence Resources) software and the results
(Foti and Strobbia 2002). are shown in Fig. 18 together with an SWM profile. Obvious-
Refraction seismograms are shown in Fig. 11 and travel- ly the velocity inversion is not detected by seismic refraction
time graphs with apparent velocities are shown in Fig. 12. and this also affects the reliability of the depth values
The only information that can be obtained from the P-wave obtained by tomography.
refraction is the P-wave velocity of the unsaturated upper-
most material (about 200 m/s), the water-table depth (about
3 m), and the P-wave velocity of the saturated media (about CONSTRAINED INVERSION
1625 m/s).
SWM supplied more interesting information regarding In the case of a plane layered lithology, with significant
the investigation target. A sample raw seismogram is shown velocity contrasts between different layers, the seismic
in Fig. 13. Mode separation is evident in the time–offset refraction results give a vertical velocity profile that could
domain where the time separation of the two main events be used as a starting model or as a constraint for surface
increases with offset. The different modes of propagation wave inversion. This could help in obtaining more consistent
are easily located in the f–k spectrum (Fig. 14). results from the inversion process, limiting the ambiguity
The fundamental mode (associated with the higher related to the non-uniqueness of the solution. On the other
wavenumber) is dominant at low frequencies, below 27 Hz. hand, SWM could supply information about small changes
At higher frequencies a higher mode (probably the third) in the mechanical properties inside the layers, which would
becomes dominant, but by seeking relative maxima the fun- not be detected by refraction surveys. Moreover SWM sup-
damental mode can be identified up to more than 35 Hz. ply direct information concerning the shear modulus at very
Because of the uncertainty related to the exact definition low strains, which is very important for geotechnical appli-
of the higher mode detected, the inversion is performed by cations such as site amplification studies.
considering the fundamental mode only. The shape of the
fundamental mode (Fig. 15) suggests the presence of a Case 4: Constrained SWM inversion
velocity inversion that was confirmed by the interpretation. This example deals with a case of constrained inversion
The values of Poisson’s ratio used for the inversion were of surface waves using refraction results. The site is located
chosen according to the results of P-wave refraction; values in the Tuscany region, close to the site of Case 1. A very stiff
were assigned to the starting SWM model, taking into limestone bedrock starting at a depth of around 9 m was
account the effect of saturation (Foti and Strobbia 2002). found by a borehole log. The upper layers consist of a geo-
Figure 15 compares the experimental dispersion curve with logical formation of gravels, very similar to that described in
the numerical one, obtained for the final profile selected in Case 1. In this case, the shallow cover also consists of soft
the iterative inversion process. soils.
The velocity profile (Fig. 16) can be interpreted by con- P-wave refraction traveltime interpretation led to a
sidering the local geology as a sequence of clayey and grav- three-layer stratigraphy with very small dip (Fig. 19a). The
elly layers, respectively characterized by lower and higher velocity values and the depths of the bedrock and of the dif-

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
128 S. Foti et al.

FIGURE 19
Case 4: (a) P-wave refraction traveltimes; (b) comparison between experi-
mental and numerical dispersion curves; (c) experimental velocity profiles.

the case of refraction and acquiring data over an adequate


frequency range for SWM). The experiments described
show how a joint acquisition and a cross interpretation of
the results could, in some cases, minimize the consequences
of the following limitations:
• Equivalence problems such as velocity inversions and
hidden layers, which are responsible for significant errors
in layer depth estimation in seismic refraction, are over-
come by using SWM, which is not affected by such limi-
tations and which could therefore provide consistent
results and ‘warnings’ for refraction interpretation (see
Cases 1 and 3). Similarly P-wave refraction cannot be
used under certain layer conditions, e.g. where all the first
arrivals are associated with a very stiff top layer, while
SWM allows the characterization of such sites (see Case
2).
• Seismic refraction assumes sharp velocity contrasts
between layers and is not adequate in the case of gradual
changes in velocity with depth, while SWM can solve any
vertical velocity distribution with good resolution, espe-
cially at shallow depths (see Cases 1 and 4).
• SWM is scarcely influenced by the presence of the water
table and gives information about small velocity varia-
tions in the saturated zone. P-wave refraction gives infor-
mation about the water-table level but is less sensitive to
small velocity variations in saturated media. On the other
hand, information about the water-table level can be
ferent portions of the overburden were in good agreement used to determine the correct value of Poisson’s ratio to
with data from boreholes in the vicinity. SWM processing be used for SWM inversion (see Case 3).
was then performed using the velocity profile from the • Non-uniqueness of the SWM inversion solution can be
refraction results as the starting model for the inversion pro- greatly mitigated by the introduction of a priori informa-
cedure. The good fit between experimental and calculated tion, which could be inferred by refraction data (see Case
dispersion curves (Fig. 19b) and the small changes intro- 4).
duced in the starting model by the inversion procedure (Fig.
19c) were confirmation of the reliability of the information The above considerations focus on many situations in which
obtained from the two seismic techniques. The results of a the cross interpretation of P-wave refraction and SWM
seismic downhole test performed in the above-mentioned could be synergic, overcoming many limitations of the two
borehole are in good agreement with the shear-wave veloc- techniques. It is suggested that SWM should be associated
ity profile obtained with SWM. with P-wave refraction as a routine technique for geotechni-
cal site characterization.

CONCLUSION AND COMMENTS


ACKNOWLEDGEMENTS
Seismic refraction and surface wave methods suffer from
some intrinsic limitations, which are inherent in their The authors thank the Tuscany Regional Government and
respective basic assumptions and interpretation procedures. the Agency for the River Po for permission to publish the
They are also affected by experimental difficulties, which data. They are grateful to F. Bosco, C. Comina and R.
are different for the two techniques (mainly: acquiring data Maniscalco for their valuable help during data acquisition
with an adequate S/N ratio for reliable first-break picking in and processing.

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129
Experiments of joint acquisition of seismic refraction and surface wave data 129

REFERENCES

Al-Hunaidi M.O. 1994. Analysis of dispersed multi-mode signals of Hering A., Misiek R., Gyulai A., Ormos T., Dobroka M. and Dresen
the SASW method using the multiple filter/crosscorrelation tech- L. 1995. A joint inversion algorithm to process geoelectric and
nique. Soil Dynamics and Earthquake Engineering 13, 13–24. surface wave seismic data. Part I: basic ideas. Geophysical
Comina C., Foti S., Sambuelli L., Socco L.V. and Strobbia C. 2002. Prospecting 43, 135–156.
Joint inversion of VES and surface wave data. Proceedings of Jones R.B. 1958. In-situ measurement of the dynamic properties of
SAGEEP 2002, Las Vegas, USA, February 10–14, CD-Rom. soil by vibration methods. Geotechnique 8, 1–21.
Dunkin J.W. 1965. Computation of modal solutions in layered, elas- Massarsch K.R. 1999. Seismic field measurements applied to static
tic media at high frequencies. Bulletin of the Seismological geotechnical problems. 5th EEGS – ES meeting, Budapest,
Society of America 55, 335–358. Expanded Abstracts, SE1.
Foti S. 2002. Numerical and experimental comparison between 2- McMechan G.A. and Yedlin M.J. 1981. Analysis of dispersive waves
station and multistation methods for spectral analysis of surface by wave field transformation. Geophysics 46, 869–874.
waves. RIG (Italian Geotechnical Journal 1, 11–22. Misiek R., Liebig A., Gyulai A., Ormos T., Dobroka M. and Dresen
Foti S., Lai C.G. and Lancellotta R. 2002a. Porosity of fluid-saturat- L. 1997. A joint inversion algorithm to process geoelectric and
ed porous media from measured seismic wave velocities. surface wave seismic data. Part II: applications. Geophysical
Geotechnique 52, 359–373. Prospecting 45, 65–85.
Foti S., Lancellotta R., Sambuelli L. and Socco L.V. 2000. Notes on Nazarian S. and Stokoe II K.H. 1984. In situ shear wave velocities
fk analysis of surface waves. Annali di Geofisica 43, 1199–1210. from spectral analysis of surface waves. Proceedings of 8th
Foti S., Lancellotta R., Sambuelli L. and Socco L.V. 2001. Conference on Earthquake Engineering, San Francisco, Vol. 3,
Application of fk analysis of surface waves for geotechnical char- pp. 31–38, Prentice–Hall.
acterization. Proceedings of the Fourth International Conference Park C.B., Miller R.D. and Xia J. 1999. Multichannel analysis of sur-
on Recent Advances in Geotechnical Earthquake Engineering face waves. Geophysics 64, 800–808.
and Soil Dynamics, San Diego, CD-rom Rix G.J., Lai C.G. and Foti S. 2001. Simultaneous measurement of
Foti S., Sambuelli L., Socco L.V. and Strobbia C. 2002b. Spatial sam- surface wave dispersion and attenuation curves. Geotechnical
pling issues in fk analysis of surface waves. Proceedings of Testing Journal, ASTM, 350–358.
SAGEEP 2002, Las Vegas, USA, February 10–14, CD-Rom. Sambuelli L., Deidda, G.P., Albis G., Giorcelli E. and Tristano G.
Foti S. and Strobbia C. 2002. Some notes on model parameters for 2001. Comparison of standard horizontal geophones and newly
surface wave data inversion. Proceedings of SAGEEP 2002, Las designed horizontal detectors. Geophysics 66, 1827–1837.
Vegas, USA, February 10–14, CD-Rom. Shtivelman V. 1999. Using surface waves for estimating shear-waves
Gabriels P., Snieder R. and Nolet G. 1987. In situ measurements of velocities in the shallow subsurface onshore and offshore Israel.
shear-wave velocity in sediments with higher-mode Rayleigh European Journal of EEGS 4, 15–35.
waves. Geophysical Prospecting 35, 187–196. Socco L.V., Strobbia C., Sambuelli L. and Foti S. 2001. Integrazione
Glangeaud F., Mari J.L., Lacoume J.L., Mars J. and Nardin M. 1998. di dati sismici ed elettrici per la caratterizzazione di un sito cam-
Dispersive seismic waves in geophysics. European Journal of pione presso gli argini del fiume Po. Proceedings of GNGTS,
EEGS 3, 265–306. Roma, November 6–8.
Godio A., Sambuelli L., Socco L.V. and Strobbia C. 2001. Rilievo Strobbia C. 2003. Surface wave methods: acquisition, processing and
Geofisico multidisciplinare per la caratterizzazione di una discar- inversion. PhD dissertation Politecnico di Torino.
ica mineraria. Proceedings of GNGTS, Roma, November 6–8. Thomson W.T. 1950. Transmission of elastic waves through a strati-
Haskell N.A. 1953. The dispersion of surface waves on multilayered fied solid medium. Journal of Applied Physics 21, 89–93.
media. Bulletin of the Seismological Society of America 43, 17–34.

© 2003 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2003, 1, 119-129

View publication stats

You might also like