International Journal of IJGE 2015 1(1): 20-41

Geohazards and Environmen t

http://ijge.camdemia.ca, [email protected] Available at http://ojs.library.dal.ca/ijge

Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model
of Subsurface in Israel
Marina Gorstein and Michael Ezersky*
Geophysical Institute of Israel

Abstract: Estimating possible site effect is an integral part of evaluation of the seismic hazard and
reduction of earthquake damages. In regions with low or moderate seismicity as in Israel, the site
response should be determined by analytical tools. These computations require knowledge of the
subsurface geological structure in terms of shear-wave velocity (Vs) profile down to seismic bedrock.
Conventionally, this problem is resolved by joint implementation of Horizontal-to-Vertical Spectral
Ratios (HVSR or Nakamura’s) technique, which is based on ambient noise measurements and seismic
methods such as S-wave refraction or Multichannel Analysis of Surface Waves (MASW) method. The
first one does not allow deep penetration of seismic waves because of its weak source. The MASW
method using 4.5 Hz geophones is restricted in penetration depth of surface waves because of frequency
(wavelength) limitations. In this study, we have applied 2.5 Hz geophones and special data processing to
provide constructing Vs section to a depth of 100 m and deeper. In combination with HVSR
measurements, MASW enables constructing reliable subsurface model, which could be integrated into
the seismic hazard assessment. Testing of this combined methodology was carried out at a number of
sites with differing geological structures in Israel.

Keywords: ambient noise, HVSR method, shear wave velocity subsurface structure, surface waves

1 Introduction reliable estimates of the required input parameters

Subsurface ground conditions, which might cause including thickness, density, damping and S-wave
significant amplifications due to high impedance velocity of different soil and rock layers at a site.
contrast between soft soils and a firm basement, 2) Direct measuring of the site response on
requires the need to estimate the expected ground the basis of earthquake recordings on specific
motions and determine the main characteristics of stations located at carefully chosen sites.
the seismic response of the underground (i.e. 3) Methods based on ambient noise
resonance frequencies and amplifications) for the recordings.
seismic hazard assessment and risk mitigation. The first group of methods based on
Techniques developed to identify the main utilization of seismic exploration to determine
characteristics of site responses for soft deposits subsurface structural models may be very
(i.e. resonance frequencies and amplification expensive. Moreover, predicting site effect
factor) may be grouped into three main categories parameters based on models inferred from
(Bonnefoy-Claudet et al 2006): geological and geophysical information only,
1) A numerical simulation approach coupled may differ significantly from experimental
with classical geophysical and geotechnical tools estimation (Zaslavsky et al 2005, 2008, and 2009).
(such as seismic refraction, seismic reflection, The second technique (Jarpe et al 1988, Satoh
boreholes, penetrometers, etc.) in order to provide et al 1995) provides an unbiased experimental

* Corresponding author: Michael Ezersky, E-mail: [email protected], Tel: +972-8-978 5865

Copyright © 2015 Canamaple Academia Services, http://press.camdemia.ca DOI: 10.15273/ijge.2015.01.004
20
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

estimation of the site transfer amplification factor. Boor et al (1997) note that use of average
However its use in regions with relatively low shear-wave velocity to a depth of 30 m as a
seismic activity like in Israel is usually variable to characterize site conditions is a choice
impractical. dictated by the relative unavailability of velocity
Finally, the third group is a practical and low data for greater depths, and it is necessary to
cost tool and is becoming more and more popular develop deeper estimates of Vs. That is why one
over the last decades (Kagami et al 1982, of the objectives of this study is modifying the
Yamanaka et al 1994). It offers a convenient MASW method to increase penetration depth.
technique, especially through urbanized areas. In addition, Zaslavsky et al (2012), based on
In the last decade, the Geophysical Institute of synthetic acceleration response spectra of real
Israel (GII) used H/V spectral ratios from ambient sites, conclude that a single parameter Vs30
noise (HVSR) supplemented with on-site cannot be used for generalizing site classification
geophysical, borehole and geological information in the complex geological conditions of Israel.
to derive the required models of the subsurface. The main goal of this study is to combine H/V
Part of the information (Vs profile) is usually spectral ratio from ambient noise (HVSR method)
obtained from S-wave seismic refraction surveys with the modified MASW technique to obtain
(Palmer 1986). However, use of the latter is often reliable shear-wave velocity structure of the
hampered by problems in generating S-waves subsurface down to bedrock. Similar combination
because of weak source and the difficulties in of techniques is reported for the first time in
performing a geophysical survey in urban areas. seismological literature by Scherbaum et al
Another method allowing constructing the (2003). This combination allows deriving
subsurface Vs profile is the widely used quantitative information on S-wave velocity
Multichannel Analysis of Surface Waves sections for the study site and enables
(MASW). The MASW technique is based on the investigating deep properties of the medium. The
study of the dispersion of surface waves (Park et main objectives are as follows:
al 1999). Since a surface wave is frequency - Modifying the MASW method to increase
dependent, i.e. dispersive, the Vs structure can be the penetration depth and resolution,
obtained by inversion of surface-waves dispersion - Applying the improved MASW method in
curves (Xia et al 1999). Depending on how the its active (passive, combined) modes to construct
surface waves are generated, active and passive shear-wave velocity (Vs) profile,
MASW techniques can be known (Park et al - Evaluating 1-D subsurface models via
2007). The passive MASW is based on fitting an analytical transfer function, using
measurements of ambient noise (tidal motion, sea SHAKE code to an observed HVSR and taking
waves, wind, traffic, industry activities). In most into account Vs-depth profile from MASW
cases, passive MASW method is combined with measurements and available geotechnical and
its active modification when excitation of surface geological data as constraints,
waves is performed using a sledgehammer and - Testing combination of HVSR and
other active seismic sources (like dropping MASW methods in different geological
weight, for instance) (Park et al 2005). conditions.
The MASW method was largely developed
taking into account recommendations of the
2 P- and S-wave velocities in soils and rocks
National Earthquake Hazard Reduction Program
of Israel
(NEHRP) about site classification averaging
shear-wave velocity (Vs) over 30 m (BSSC 1994). Shear-wave velocity measurements are an
Therefore, the MASW conventionally uses 4.5 important tool in designing buildings for site
Hz geophones and the penetration depth of specific conditions such as soil liquefaction,
surface waves is restricted (usually, by uppermost ground-spectral earth quake response, etc. Being
30 m) because of frequency (wavelength) slightly dependent on soil saturation, shear-wave
limitations (Park 2006). velocities are much more of a diagnostic tool for

21
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

engineering properties than P-wave ones. and Surface Wave prospecting (SWP) methods
Seismically, shear-wave velocity (Vs) is the best (Park et al 1999, Socco and Strobia 2004, Stokoe
indicator of shear modulus that is directly linked II et al 2006). If refraction method is based on
to a material’s stiffness and is one of the most direct S-wave excitation and measurements, SWP
critical engineering parameters. There are several method is based on the nature of Rayleigh waves
methods of Vs measurements in laboratory and whose phase velocity depends on Vs and its
in-situ conditions. distribution with depth. Latter waves are of
In situ geophysical methods include borehole dispersive origin that testifies possibility to
and surface measurements. The most accurate penetrate to different depths. Although methods
method is the cross-hole one (ASTM like shear-wave refraction, downhole, and cross-
D4428/D4428M, 1991 (1995)). This method hole surveys can be used, they are generally less
requires at least 2 highly parallel boreholes and is economical than SWP in terms of field operation,
limited to a depth of 30 m. Another method is the data analysis, and overall cost. Multichannel
downhole one (ASTM D7400, 2007) allowing Vs Analysis of surface Waves (MASW) is one of
measurement of a single borehole. The borehole SWP techniques.
is encased by a PVC pipe or filled with bentonite Numerous surveys carried out in Israel allow
to stabilize the borehole walls. In the first case, estimating the range of seismic compression
measurements can be affected by the quality of wave (Vp) and shear-wave (Vs) velocities of the
the borehole walls and fill geometry. In addition sediments and rocks down to a depth of 100 m.
soil characteristics in borehole can differ from The schema characterizing ranges of in-situ
those in the site located 100 m away. seismic velocities in the soils and rocks of Israel
Surface methods of measurements include is presented in Figs. 1a and 1b, respectively.
seismic refraction measurements (Palmer 1986)

Fig. 1 Range of Vp and Vs in soils (a) and rocks (b) measured in Israel in-situ and in laboratory (in brackets) conditions

Analysis of Figs. 1a and 1b allows us to The higher Vp values could be explained by the
conclude that ranges of both Vp and Vs seismic high stress and better rock quality in deep
velocities in soils and rocks generally conform to measurement conditions. Shear-wave velocities
those measured by other researchers in shallow measured throughout Israel vary by a wide range:
subsurface (Jakosky 1957). However, Vp values from 100 m/s in sands and lime carbonates to
in rocks reported in literature are higher (for more than 600 m/s in gravels, and from 400 m/s
instance, Vp = 5800 m/s in granite, Vp = 6100 in conglomerates to more than 2400 m/s in
m/s in limestone) than those measured in Israel. basalts and dolomites. Variability of Vs and

22
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

thickness of soft sediments overlying hard rock for higher impedance contrasts (by a factor of 4).
on the one side, and very limited availability of It was demonstrated through many studies
densely distributed geotechnical information such (Zaslavsky et al 2005, 2008, 2009) that, when
as Vs at depth call for less expensive and less noise measurements are made near boreholes
time consuming methods to provide the required and/or near refraction surveys, the fundamental
parameters for site effect assessment. Data frequency and its corresponding H/V amplitude
presented in Figs. 1a and 1b allow consideration are practically the same as the fundamental
of approximate range of Israeli sediments and frequency and its corresponding amplification
rocks for preliminary evaluation of possible site level derived from the computed transfer function
response and modeling. Separately, we consider of SH-waves at low strains propagating through a
velocities in the salt constituting firm layers along relatively simple 1-D model of the site, as known
the Dead Sea coastal area. from geotechnical and geophysical surveying.
Computer code SHAKE (Schnabel et al 1972) is
3 Methods used to analytically evaluate the site response
function. The specific parameters required for this
3.1 HVSR method
analysis are:
3.1.1 General - S-wave velocity, thickness, density and
Nakamura (1989) hypothesized that site response damping of each layer in unconsolidated
could be estimated from the spectral ratio of sediments, and
horizontal versus vertical components of noise - S-wave and density of the hard rock
(reflector).
observed at the same site (site of interest).
The HVSR technique has become the primary 3.1.2 HVSR data acquisition
tool of choice in many of the ambient noise The methodology of HVSR data acquisition and
related studies and it has been successful in processing was presented in details in Zaslavsky,
seismology to estimate the local transfer function et al. (2009). Ambient noise measurements were
in the site response problem in Israel and made using portable instruments (Shapira and
worldwide (Lerma and Chávez-García 1994, Avirav 1995) consisting of a multi-channel
Malischewsky et al 2010, Mucciarelli and amplifier, a Global Positioning System (GPS) for
Gallipoli 2004, Seekins et al 1996, Zaslavsky et timing, and a laptop computer with a 16-bit
al 2005, 2008, 2009). Nakamura's method is analogue-to-digital conversion card to digitize
based on the assumption that micro-tremors and store the data. Each seismograph station
consist of body waves. Mucciarelli and Gallipoli consists of three (one vertical and two horizontal)
(2004) claimed that the H/V spectrum of ambient L4C velocity transducers (Mark Products) with a
noise is dominated by the upward propagation of natural frequency of 1.0 Hz and damping ratio
SH wave through the layered media. On the other 65-67% of critical. The sample rate was 100
hand, an explanation based on the opposite samples per second and filter band-pass was
assumption that micro-tremors mainly consist of between 0.2 Hz and 25 Hz. All the equipment:
surface (Rayleigh) waves is also successful (e.g. sensors, power supply, amplifiers, personal
Fäh et al 2001, Lachet and Bard 1994). Both computer, and connectors were portable allowing
models agree that the H/V spectra and the site performing of measurements in autonomous
response function for SH wave are the results of mode.
the velocity structure of the media, that both
exhibit the same fundamental resonance 3.1.3 HVSR data processing
frequencies with similar amplitudes at least when For each site, the average H/V spectral ratios and
considering small motions. Malischewsky and their corresponding standard deviations were
Scherbaum (2004) demonstrated analytically the determined by applying the following process:
coincidence of H/V from the ellipticity of time windows, each of 30-60 seconds long
Rayleigh waves and the fundamental frequency depending on fundamental frequency. The

23
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

selected time windows were Fourier transformed, curves if the propagation mode signatures are
using cosine-tapering (1 second at each end) well separated. The dispersion curves are then
before transformation and then smoothed with a inverted for a 1D Vs profile with depth. When
triangular moving Hanning window. The H/V data are collected in a roll-along mode, each 1D
spectral ratios were obtained by dividing the profile is represented at its corresponding
individual spectrum of each of the horizontal midpoint spread, allowing a pseudo-2D Vs
components by the spectrum of the vertical section to be drawn. The inverse problem
component. To obtain consistent results from the formulation imposes that the investigated medium
spectra of ambient noise, 60-70 time windows is assumed as one-dimensional under the spread.
were used and then averaged with the spectral Long spreads are required to record wavelengths
ratios. Data processing was carried out using large enough for increasing the investigation
"SEISPECT" software developed in the depth and mitigating near-field effects (Bodet et
Geophysical Institute of Israel (Perelman and al 2005, Socco et al 2009).
Zaslavsky 2001). Passive MASW. The passive surface waves
3.2 MASW method generated from natural (e.g. tidal motion) or
cultural (e.g. traffic) sources are usually of a low-
3.2.1 General
frequency nature with wavelengths ranging from
Surface-wave dispersion inversion (SWDI) is a a few kilometers (natural sources) to a few tens
standard approach for inferring a 1D Vs structure. (or hundreds) of meters (cultural) (Okada 2003),
Surface waves, commonly known as ground roll, providing a wide range of penetration depths and
are always generated in all seismic surveys, have therefore a strong motivation to utilize them. The
the strongest energy, and their propagation ambient noises are recorded using receiver arrays
velocities are mainly determined by the medium’s (antennae) arranged as different geometrical
shear-wave velocity. figures (linear, circular, cross layout, etc.). The
The development of multichannel equipment most accurate estimation is obtained through a
has led to exploiting the methodology known as survey using a true 2D receiver array (Park and
Multichannel Analysis of Surface Waves Miller 2006). However, because the true 2D
(MASW) (Park et al 1999, Xia et al 1999). The receiver array, such as a circular and cross-layout
MASW method is basically an engineering ones are not a practical or possible mode of
seismic method dealing with frequencies mainly survey in built-up urban areas, a method that can
3-30 Hz recorded by using a multichannel (24 or be implemented with the conventional 1D linear
more channels) recording system and a receiver receiver array can be effective in this case (Louie
array deployed over a 2-200 m distance. The 2001). The data processing scheme can be found
active MASW method generates surface waves in Park et al. (2007).
actively through an impact source like a sledge Combination of active and passive MASW
hammer, whereas the passive method utilizes measurements. Dispersion images processed from
surface waves generated passively by cultural (e.g. active and passive data sets should be combined
traffic) or natural (e.g. thunder and tidal motion) to obtain improved dispersion curves. The active
activities (Park et al 2007). MASW method generates signals in about 30 Hz,
Active MASW. Active-source surface-wave- whereas the passive one allows for widening that
dispersion measurements are made with typical to the low frequency range down to geophone
seismic shot gatherers that are a collection of frequency. Combining two signals, we widen the
seismic traces, which share some common range of frequencies from 3-30 Hz, and sometime
geometric features. The wave field is transformed even up to 50 Hz. Thus, the penetration depth can
into a frequency-wave number (or frequency- be increased to a deep range (low frequencies)
slowness) domain in which the maxima should and the uppermost depth can be decreased (high
correspond to surface-wave signatures. Several frequencies) (Park et al 2007).
modes can be picked out for such dispersion

24
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

3.2.2 Data acquisition refraction one is the quality of the field records.
We illustrate it comparing two above acquisition
3.2.2.1 Practical aspects of data acquisition
methods in Fig. 2. Surveys were carried out along
Although methods like shear-wave refraction, the same length line by different hammers. The
downhole, and cross-hole surveys can be used, data acquisition of S-waves with the refraction
they are generally less economical than any other method was carried out using the horizontal
seismic methods in terms of field operation, data stroke of a 30kg sledge hammer (Fig. 2a)
analysis, and overall cost. The great advantage of allowing a most clear excitation of SH waves at
the surface wave method in comparison with the the background of other waves.

Fig. 2 Comparison of seismic methods for S-wave acquisition. (a) Excitation of S-waves in the seismic refraction technique
using horizontal stroke on metallic beam by a 30 kg hammer; (b) raw data obtained at such stroke complicated by strong
noise; (c) excitation of surface wave of R (Rayleigh) type using vertical stroke by a 200 kg hammer forced by slingshot; and
(d) field records of surface waves

However, because of weak source, S-wave 3.2.2.2 Modified data acquisition

records are complicated by ambient noise (Fig.
Active MASW. Conventional seismic data (i.e. the
2b) allowing penetration of S-wave as a rule to vertical component of the wave field from
25-30 m. On the other hand, surface waves,
common shot records obtained in shallow
because of strong vertical stroke by the 200 kg- refraction surveys) were used. To increase
hammer (in our case, forced by slingshot) (Fig.
penetration depth of surface waves, we used
2c), are characterized by the strongest energy, and vertical low frequency 2.5 Hz geophones
their propagation velocities are mainly
implemented to seismic profile. Receiver spacing
determined by the medium’s shear-wave velocity.
was varied with respect to necessary penetration
The quality of the field records is significantly
depth from 2.5-10 m; shot location was 5-10 m
higher (Fig. 2d) and allows penetration down to
away from nearest trace (off-end shooting). Data
depths determined by wavelength (that is usually
excitation was carried out using power Digipulse
some tens to hundreds of meters).
hydraulic source mounted on a Chevrolet pickup

25
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

truck (Fig. 2). Both geophone frequency and applied to the solution in order to reduce the
power source facilitate recording raw data of high degree of non-uniqueness. Constraints are data
quality and more penetration depth. A Summit II available from independent geological and
plus seismic recorder was used with a 24-48 geophysical information, such as longitudinal
geophone spread (vertical 2.5 Hz geophones). (compression) wave velocity (Vp), depth to
The number of geophones as well as line length reflector layer or half space, firm layer thickness
selected in accordance with the depth of target. and composition, geological section of the site
At first, P-wave seismic refraction study was under investigation. These data are intended to
carried out along study line using 2.5 Hz vertical confine result of inversion by possible parameters.
geophones. Record length for P-wave refraction The dispersion equation depends mainly on Vs,
was 500 ms. Then record length was increased to and thickness value in the layers. An appropriate
2000 ms and surface waves were recorded. choice of these parameters (the initial model) is
Passive MASW. Passive MASW considered as a fundamental issue for the
measurements were carried out using linear successful application of inversion (Socco and
system located along the roads (roadside schema) Strobia 2004).
with 2.5 Hz geophone separation of 5 m. Other The most important part of the MASW data
array used in Israel was circular array with 5-10 processing is constructing an initial layered
m separation between geophones (Ezersky et al model for inversion. Generally, bounds of wave
2013b). velocities are presented in Fig. 1. Some
Then active and passive records were parameters can be selected using well-known
combined using SurfSeis v3 software. rules of thumb (Xia et al 2003): (a) Vp versus Vs
ratio can be considered bounded for near surface
3.2.3 Data processing
materials, assuming Poisson's ratio, with values
Data processing is applied to (a) P-wave ranging from 0.20 to 0.48, and (b) The Rayleigh-
refraction data, and (b) surface wave data. wave velocity is 0.92 of Vs in an uniform half-
a) P-wave refraction processing is intended space (for Poisson’s Ratio of 0.25). As a
for constructing Vp depth layered model and reference model for inversion, Vs can be
determination depth to firm layer (reflector). approximated by the phase velocity multiplied by
These data are used for generating of initial a correction factor (less than unity for
model for surface wave inversion. fundamental-mode data). The examples of field
b) Surface wave data are used for: (1) records and data processing will be considered in
generation of dispersion image, (2) extraction of continuation.
dispersion curve, and (3) its inversion. All these Inversion was carried out at constant Vp
were carried out using SurfSeis v3. Software of values, whereas Poisson’s Ratio varied (Ezersky
Kansas Geological survey (KGS). This software et al 2013a). Such inversion procedure allows
allows any combination of active and passive stabilizing the inversion results.
dispersion images with different parameters of
measurements (different frequency, separation of 3.3 Combination of HVSR and MASW
geophones, etc.). methods
An inversion of the dispersion data was Data collected from a few seismic profiles
carried out using linearized inversion with a provide information on the S-wave velocities and
gradient-based iterative method (Park et al 1999, thickness of shallow sediments (down to 50-100
Xia et al 1999) implemented to above mentioned m) within the accuracy and resolution of the
software. The Root Mean Square Error (R.M.S.E.) geophysical technique. Seismic MASW profiles
between the theoretical dispersion curve and the are normally designed to obtain maximum
measured one is usually used as an indicator of information on Vs of the lithological units
the closeness between measured and iteratively represented in the study area and in the vicinity of
calculated dispersion curves. Usually, in boreholes. Measurements of ambient vibrations
linearized inversion methods, constraints are were also carried out either very close to or

26
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

directly at drilling sites where detailed 0.5 λmax (Shtivelman 1999) to (0.5-1) λmax (Rix
information on the subsurface was available. The and Leipski 1991). The latter affirms that the best
logging data are incorporated to obtain more overall accuracy and resolution would be
detailed and reliable information about the obtained when the dispersion data is evenly
subsurface. Then, the borehole and geophysical distributed between the minimum and maximum
information were combined with the observed wavelengths and the maximum wavelength is one
spectral ratios to estimate the depth S-wave to two times the maximum desired depth of the
velocity profile. The iterative procedure based on shear-wave velocity profile.
the stochastic optimization algorithm (Storn and It is important to note that in principle,
Price 1995) was applied in order to fit an penetration depth can reach λmax (and even, more).
analytical transfer function, estimated using
SHAKE code (Schnabel et al 1972) to an It depends on some factors such as shear-wave
observed H/V spectral ratio, focusing mainly on velocity of overburden, frequency, and signal
the resonance frequencies and considering the quality. Our experience (Ezersky et al 2013a)
shape of H/V curve. Thus, combining the shows that in most cases (with 2.5 Hz geophones)
borehole and geophysical information with the criterion of Z max = 0.5 λmax is enough to estimate
observed spectral ratios 1-D the depth – Vs properties (Vs) of the foundation down to 40-60
velocity profile was derived. m depth. However, in rare cases deeper reflector
penetration depth should be checked as an
3.4 Resolution of MASW method
exception. Example of such a case is presented
The resolution of MASW method is considered here for Ramat Hakovesh site described in this
usually as a rule of thumb (Bodet et al 2005, article, where active and passive MASW
Richart et al 1970, Park et al 1999, O’Neill 2003, measurements were carried out:
Shtivelman 1999). Many conclusions are based
on experimental results. Some issues should be V ph max = 708m / s
considered here in this context: (1) maximum and
f min = 2.96Hz
minimum penetration depth, (2) their
relationships with seismic line length, and (3) λmax = V ph max / f min = 239m
reliability of measured velocities. The resume is
based on literature data and our experience Where, f min = minimum frequency measured
(Ezersky et al 2013a, 2013b). in combined image, V ph max = maximum phase
Maximum penetration depth. At the beginning
of our study, we checked the main concerns and velocity, measured at this frequency, and
possibilities for increasing penetration depth. The λmax = maximum wavelength. Thus, we can
normally accepted axiom (Richart et al 1970) is accept the criterion of Rix and Leipski (1991) that
that the penetration depth ( Z max ) of ground roll is maximum penetration depth can be evaluated as
approximately equal to its wavelength ( λ ) and between 120 and 239 m. However, it is
may be up to 2 λ . Then the question arises, what reasonably to accept maximum penetration depth
is the maximum penetration depth ( Z max ) for as 120 m at least.
Relationship between λmax and length of
which Vs can be reasonably (exactly enough)
calculated. Park et al. (1999) evaluate it as geophone line (array). This issue is debated by
0.5 λmax . Rix and Leipski (1991) evaluate it as different investigators. It is generally considered
that length of receiver spread (L) is directly
(0.5-1.0) λmax . The maximum depth depends on related to the longest wavelength ( λmax ) that can
the maximum reliably estimated wavelength. be analyzed, e.g. L ≈ λmax (Park et al 1999).
Different authors suggest Z max as a rule of thumb.
L = X1 ∗(N −1) , where X 1 is separation between
Some estimations have been suggested. As a rule, geophones and N is number of geophones in
it is estimated as 0.25 λmax (Bodet et al 2005),
spread. Parameter X 1 = H 1 is connected with a

27
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

minimum measured depth in limits of which Vs (both, 2.5 Hz geophones) and separation of 5 m
can be considered as a reliable value (Stokoe et al and 10 Hz geophones show:
1994), e.g. H 1 > 0.5λ min = 0.5V ph min / f max , where 2.5 Hz geophones, 10 m apart:
f max is the highest frequency at which minimum V phmin = 219m / s
phase velocity V ph min can be measured. Socco and
f max = 15Hz
Strobia (2004) concluded that really, wavelengths
longer than the array could be observed, and the H 1 = λmin / 2 = 7.3m
maximum wavelength depended mainly on the
quality of the dispersion image. Our experience 2.5 Hz geophones, 2.5 m apart:
shows the correctness of this conclusion.
V phmin = 219m / s
Our concept is to use frequencies as low as
possible to reach a maximum penetration depth. f max = 23Hz
To improve quality of signal (S/N ratio) we have
used in Israel: (1) low frequency geophones, (2) H 1 = λmin / 2 = 4.7m
power Digipulse source based on Chevrolet
pickup truck (Fig. 2b), (3) long seismic line, and 10 Hz geophones, 5 m apart:
(4) all available information (borehole data, V phmin = 256m / s
refraction seismic and TEM measurement).
The increase in penetration depth is reached f max = 59 Hz
also by a combination of active and passive
sources (Park et al 2007). That is why we also H 1 = λmin / 2 = 2.2m
consider (0.5-1.0) λmax criterion as determining It is clearly seen that a more effective way to
Z max suggested by Rix and Leipski (1991). This improve the resolution of MASW at small depth
criterion well corresponds to our conditions and is to increase geophone frequency instead of
data. We have checked it with passive MASW, decreasing separation.
which has a penetration depth Z max > 0.5 λmax .
4 Parametric study
Resolution in uppermost subsurface. The
thickness H 1 of the uppermost (irresolvable) Within the framework of this paper, we consider
layer limiting the resolution of sub-surfaces is two typical subsurface structures forming
determined by the wavelength and parameters of conditions for site effect (Fig. 3).
the data acquisition system according to the half- In the first model, one or two hard layers are
wavelength criterion (Rix and Leipski 1991). The over- and underlain and by soft soil (“Layer
normally accepted criterion is that the minimum model”). This case is presented by salt layers of
5-30 m thickness located at a depth of 20-50 m
penetration H 1 is determined by (Stokoe et al
along the Dead Sea coast that is a national and
1994): international resort area. The subsurface structure
H 1 ≥ 0.5λ min = 0.5V phmin / f max (1) in the second case is widespread in the Israel
plain, and is formed by soft sediments overlaying
where V phmin is the minimum phase velocity of rigid rock (“Half-space model”). With respect to
the fundamental mode and f max is corresponding the surface waves prospecting method, the
penetration depth depends on geophone
frequencies at the phase velocity measured.
frequency used. Forward modeling carried out for
The problem can be resolved in two ways.
the two above models of subsurface (Fig. 3b
The first way is to vary the geophone separation
dashed and solid lines) has shown that dispersion
at active MASW. The second way is to use high
curves are significantly differed (dashed and solid
frequency array. The real example of data
graphs in Fig. 3a, respectively). Fig. 4 shows
acquisition with separations of 10 m and 2.5 m

28
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

examples of two combined dispersion images depth of 70-100 m and even more (Ezersky et al
obtained in subsurface of above structure. 2013a, 2013b). Thus, the MASW method in our
Usually, geophones of 4.5 Hz are mainly used modification allows for resolving the objectives
for surface-wave prospecting. Fig. 3a shows that formulated in Section 1.
in the frequency range of more than 5.5-6 Hz
both curves are non-distinguishable. To 5 Testing combination of HVSR and MASW
distinguish between recorded curves frequency methods at two sites of Israel
should reach the frequency range between 4.5-6
Hz. As a rule, to obtain such frequency the source 5.1 North of the Dead Sea
should be powerful. 5.1.1 Ein Boqeq site
Otherwise, surface waves cannot penetrate
overburden. Geophones used in our study (2.5 Hz) The Ein Boqeq area, 1.5-2.0 km long and 250-
together with power Digipulse source allowed 300 m wide, extends along the southern Dead Sea
low frequencies down to 2.5 Hz (Fig. 4a). These (DS) basin (recently evaporation ponds were
(together with proper line long) allow penetration artificially filled by pumping DS water from the
of both overburden and solid layers down to a northern basin (site EB in Fig. 5a). Several hotels

Fig. 3 Forward modeling. (a) Two dispersion curves corresponding to subsurface models shown in (b). Dashed line
corresponds to firm salt layer of 20 m thickness located under alluvial sediments at a depth of 25 m. Salt layer is also
underlain by alluvial sediments; Salt velocity Vs is 1650 m/s. Solid graph corresponds to the rigid half space (rock) overlain
by alluvial sediments. Vs of half space is also 1650 m/s

Fig. 4 Examples of combined dispersion images with extracted dispersion curves for subsurface models discussed above. (a)
Ein Boqeq site - “Layer model” (dashed line in Fig. 3b); (b) Ramat Hakovesh site - the “Half space” (solid line in Fig. 3b) is
located at a depth of 74m deep under alluvial sediments. In both cases the dark zone near 2.5Hz is an anomaly associated
with maximum energy near resonant frequency

29
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

Fig. 5 Ein Boqeq north site. (a) Site location map, (b) EB-2 borehole section (after Yechieli et al 2004), and (c) Geophysical
measurements layout

are located in Ein Boqeq, which is also named the active MASW measurements along Line 7ss (see
Hotel area. The study area is situated in the Fig.5c for location) are shown in Fig. 7. Active
northern part of Ein Boqeq, about 2-3 km east of signal is of a good quality (Fig. 7a) that allows
the Judea Mountains (see Fig. 5a and 5c). extraction of dispersion image of a good quality
According to data of boreholes EB-1 and EB-2 (S/N Ratio is in limits of 1.0-0.9) in a wide
(Yechieli et al 2004), geological section of this frequency range between 5-30 Hz (Fig. 7b).
area is represented by pebbles, cobbles and silty- Combined MASW measurements. Combined
sandy sediments (Fig. 5b). Borehole EB-3E image shown in Fig. 4a integrates the dispersion
crossed a sandy-gravel alluvium layer overlying images of passive and active MASW surveys
the salt unit at 26 m deep. At a depth of 32 m, presented in Figs. 6c and 7b, respectively. The
borehole met crushed salt and at 35 m, deep refraction P-wave depth velocity section used as
drilling was stopped. The sediments revealed by constraint for inversion is shown in Fig. 8a.
EB-3E borehole conform to those from boreholes Combined inversion of the dispersion curve for 5-
EB-1 and EB-2 located some 200-400 m south of layered model is presented in Fig. 8b
the study site. The thickness of the salt unit is 28 (R.M.S.E. = 12%) and in Table 1. The depth to
m. the top of salt layer is accepted as 26 m.
5.1.2 MASW measurement results 5.1.3 HVSR measurement results
Passive MASW measurements. Passive MASW Ambient noise measurements were carried out
measurements were carried out using a roadside close to borehole EB-2 (see Fig. 5c for location).
technique along a 24-channel line located south Individual and average H/V spectral ratios
of the Hod Hotel, 35-45 m away from a local road obtained at station EB-2 are shown in Fig. 9a.
(Fig. 6a). Two prominent resonance peaks observed at a
An example of ambient noise record from frequency of 0.65 Hz and in the range 3.5-6 Hz
traffic is shown in Fig. 6b. Noise was recorded associated with amplitude 2.5 and 3.5,
during 1 minute with a sample rate of 4 ms. The respectively, are shown in Fig. 9a.
dispersion image with the extracted dispersion
curve is depicted in Fig. 6c. The frequency range 5.1.4 Construction of the subsurface model
is between 2.5 and 12 Hz. using HVSR and MASW
Active MASW measurements. Results of The geotechnical data, which could contribute to

30
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

constructing the subsurface model in the Ein (dashed line in Fig. 9b) and empirical (solid line)
Boqeq north site, are shown in Table 2. They are estimations is reached by assuming the thickness
collected from borehole EB-2 (Fig 5b), refraction of the gravel-clay-limestone layer equal to 135 m
survey and MASW measurement. Note that the and Vs of the deep reflector (the most likely
shallow salt layer at a depth of 27 m is clearly another salt layer) is 1500 m/sec. We note that
detected by both refraction survey and MASW thickness and Vs of layers in the upper part of the
methods (Figs. 8a and 8b). Additionally, the subsurface model are kept as they are in the
MASW method provides us Vs of this salt layer MASW section or with minor changes.
and the underlying layers composed of gravel and Thus, the deep reflector depth estimated using
lime carbonate. The optimal analytical function HVSR method combines with geophysical survey
superimposed on the average spectral ratio is is 190 m. The optimal subsurface model is given
shown in Fig. 9b. The best fit of analytical in Table 3.

Fig. 6 Ein Boqeq site results. (a) Passive MASW roadside array, 5 m separation; (b) field records, 30s length, sample rate,
4 ms; and (c) dispersion image with extracted dispersion curve (See Fig. 5c for location). Relative coordinates in meters;
Intensity of traffic: 1 car and 1 bus during 1 minute

Fig. 7. Active MASW data along line EB7SS in the Ein Boqeq north site. (a) Field records acquired using active source; (b)
Dispersion image (DT) generated from active record: squares are points of the extracted dispersion curve. The DT image
comprises mainly fundamental mode and allows extracting the fundamental mode of dispersion curve

31
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

Fig. 8 Ein Boqeq north site. (a) Seismic refraction Vp depth velocity section along line 7ss; (b) Velocity versus depth
function obtained by inversion of combined dispersion curve shown in Fig. 4a for 5 layered model. Denoted: Initial model
(thin blue dotted line), corresponding initial dispersion curve (thin green dotted line), experimental points (blue rhombs) and
fitted dispersion curve (blue solid line), inverse depth velocity model (blue layered graph). Lower horizontal axis is frequency
(Hz) and top axis is depth (m)

Table 1 Parameters of the inversion procedure for 6 layered model (Ein Boqeq North, Line EB7ss)

Layer Depth range Initial model Inverted model R.M.S.E of Vs Vp Density

No m m/s m/s % m/s kg/m 3
1 0–6 380 365 11.7 930 1550
2 6.0 – 16.0 300 330 6.02 706 1640
3 16.0 –26.0 500 480 8.7 1224 1730
4 26.0 – 46.0 1500 1150 12.0 3500 2100
5 Under 46.0 800 570 10.00 1220 1910

Fig. 9 (a) Individual and average H/V spectral ratios ± standard deviation obtained at point EB-2 (digits show frequency of
the ratio maximum); (b) Analytical transfer function (dashed line) in comparison with H/V spectral ratio of ambient noise
obtained at the Ein Boqeq north site (solid line)

5.2 Ramat Hakovesh, (the Town of Tira, Mountains slope represented by hard carbonates
Central Israel) of the Judea Gr. (Turonian-Cenomanian age). The
plain is filled by heterogeneous sediments of the
5.2.1 Geological outline
Pleistocene age, mainly represented by clayey
The investigated area is situated in the center of and sandy soils of 70-100 m thick as it is seen
Israel (Fig. 10a) close to the town of Tira (Fig. from section of the borehole 1A (Fig. 10c)
10b), a few kilometers west to the Shomron located at the investigation site (Fig. 10d).
32
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

Table 2 Geotechnical parameters used for analytical site response determination (Ein Boqeq north site)

P-wave Refraction Line EB-

Borehole EB-2 MASW EB7SS array
7ss
Lithology Depth range Depth range Vp Depth range Vs
m m m/sec m m/sec
Sand and gravel 0-5 0 to 5-8 690 1-5 LRZ*
5-12 300
Gravel and sand 5-27 5-8 to 22-25 2490
12-27 410
Salt with sand, clay
27-62 Below 22-25 2590-3930 27-47 1200
and gravel layers
Gravel, clay, and
Below 62 -- -- Below 47 550
limestone
*Low resolution zone

Table 3 Optimal 1-D subsurface model for determination of analytical site response function at the Ein Boqeq north site

Thickness Depth to layer bottom VS Density

Layer Damping %
m m m/sec kg/m3
1 5 5 200 1600 5
2 5 10 300 1600 5
3 17 27 410 1800 4
4 28 55 1200 2100 1
5 135 190 600 1800 2
Half-space ∞ -- 1500 2000 --

5.2.2 Schema of measurements (Fig. 11b), and passive image extends the low-
frequency range to 2.5 Hz, whereas high-
The measurement schema is shown in Fig. 10d.
Two seismic refraction lines (P-wave line of 320 frequency decreases to 9 Hz (Fig. 11c).
Combining the different images allows getting a
m long and S-wave line of 250 m long) were shot
in 2007. MASW lines SS-10, SS-5, and SS-2.5 of frequency range between 2.5 Hz and 23 Hz and a
250 m, 150 m and 72.5 m long, respectively, reach-resolution (penetration of waves) of 1
comprising of 2.5 Hz geophones (with separation wavelength in the deep part and the resolution in
shallow depth to 2.5 m from surface.
of 10 m, 5 m, and 2.5 m, respectively) were shot
in 2014. HVSR of ambient noise was obtained at Inversion was carried out with the following
constraints: top of reflector is located at a depth
point 11w situated close to borehole 1A.
of 73.5 m, Vp = 3230 m/s based on seismic
5.2.3 MASW results refraction section (Fig. 12a)
Dispersion images obtained for different data Inversion was carried out with Vp fixed and
(different separation between geophones, active Poisson’s Ratio varying. 5-layered model with
and passive sources, combined images) are half space located at 73.5 m gives best result with
presented in Fig. 11. Note that active dispersion R.M.S.E. = 9.9%. Vs of half space is evaluated as
curves for 10 m separation (Fig. 11a) are in the 1480 m/s. Inversion results for 5-layered model
range of 4-12 Hz, similar images for 2.5 m are shown in Fig. 12b and in Table 4.
separation wide, the frequency range to 3-23 Hz
33
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

Fig. 10 General map of Ramat Hakovesh site location. (a) Israel map and test site location; (b) Geological map showing the
study area relative to the Tira town; (c) General lithological section of the Ramat Hakovesh based on 1A test borehole (see
Fig.10d for location); (d) measurement layouts. RF (top of dolomite) is located at a depth of 73.5 m in borehole 1A (test well)

Fig. 11 Four dispersion curves extracted from surface wave records from different MASW measurements: (a) active source
along a 26-channel line with a geophone separation of 10 m (250 m length); (b) active source along 31 channel line with
separation 2.5 m between geophones (77.5 m length); (c) ambient noise (passive source) along 26 channel line 10 m
separation; (d) combined dispersion curve combined from (a) – (c) images; (e) and (f) are field records of active and passive
sources, respectively

34
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

Fig. 12 Ramat Hakovesh site. Velocity versus depth function obtained by inversion of combined dispersion curve shown in
Fig. 5b for 5 layered model with half space as foundation. Denoted: as Fig. 8b. Vertical axis is Vs, upper horizontal axis is
depth, and lower one is the frequency

Table 4 Depth Vs profile derived from MASW and optimal 1-D subsurface model for determination of analytical site
response function at the Ramat Hakovesh site

MASW MASW+HVSR optimal model

Lithology
Depth Vs Depth Vs
m m/sec m m/sec
Alluvium 0-5 205 0-5 170
5-30 300 5-30 235
Sand, loam, clay, calcareous
30-50 500 30-55 450
sandstone
50-73 490 55-70 520
Broken dolomite, limestone 73-240 1480 70-260 1430
Dolomite below 240m -- -- 2400

5.2.4 HVSR results section of the borehole 1A (Fig. 10c) layered Vs

Measurements of ambient vibrations were carried may be correlated with sandy loam (Vs = 205
m/sec); sand (Vs = 300 m/sec); clay and sandy
out close to the experimental borehole 1A where
detailed information on the subsurface down to a loam (Vs = 500 m/sec) that overlies the broken
dolomitic limestone (VS = 1440 m/sec). These
depth of 75 m is available. The prominent feature
of the individual and average spectral ratios data are presented in Table 4. Slightly different
section was extracted from refraction survey
shown in Fig. 13a is two inseparable peaks at
along P-S lines, shown in Fig. 12a. The response
frequencies of 1.4 and 1.9 Hz. Such distribution
function calculated using solely geophysical data
of peaks could be construed as follows: while the
first resonance peak is related to the hard rock at and the broken dolomitic limestone at a depth of
73 m as a fundamental reflector is shown in Fig.
depth, the second peak is caused by intermediate
hard layer directly overlying the deep reflector. 13b (blue solid line). It is clearly seen that there is
no satisfactory consent between the calculated
5.2.5 Combination of HVSR and MASW function and HVSR neither in amplitude nor in
methods shape. In particular, the calculated function
In the velocity-depth section obtained from exhibits a single resonance peak with amplitude
MASW measurements (Fig. 12b), four layers of 3.5 at resonance frequency 1.6 Hz, while HV
could be identified. As seen in the lithological spectral ratio yields two resonance peaks at 1.4

35
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

and 1.9 Hz with associated amplitudes of 7 and (dolomite) at a depth which should be estimated.
5.5 respectively. Therefore, in the second step, we The result of the optimization procedure is shown
supposed the subsurface model, in which the in Fig.13b. The optimal model providing the best
broken dolomite-limestone is an intermediate fit between HVSR and calculated response
hard layer and overlays the fundamental reflector function is given in Table 4 as well.

Fig. 13 (a) horizontal-to-vertical spectral ratios (individual, average and standard deviation) obtained in Ramat Hakovesh site.
Arrows show the fundamental f 0 and second resonance f1 frequencies; (b) HVSR curve (red solid line) superimposed on
analytical function calculated using MASW results only (blue line) and obtained by combination of HVSR and MASW
methods (dashed line)

6 Discussion and conclusions downhole methods were applied in the earlier

Any analytical procedure (making use of a investigations. Noting the high quality of the S-
computer code) needs a model of the site’s wave refraction method in constructing
subsurface to facilitate the computations. Based subsurface models (Palmer 1986) we have to
on numerous investigations it can be concluded resume very shallow, as a rule, Vs sections
that there is no one single technique that will derived from this survey (Fig 2). Deeper
penetration of surface waves enables constructing
provide a model good enough for site response
evaluations. The spectral ratio method reveals key deeper Vs sections and facilitates their
information about the dynamic characteristics of optimization in combination with the HVSR
the subsurface and should be used to help method. Moreover, the system of shooting used in
the MASW method enables us also
construct the subsurface model. In most of recent
researches carried out in Israel, the reliable simultaneously to carry out a seismic refraction
line at the same disposition, geophones and stroke
subsurface models could be reached by
to get the Vp section used as a constraint.
integrating information from empirical HVSR,
MASW is usually performed in 3
and available geological, borehole, and
modifications: (1) passive, (2) active, and (3)
geophysical data (Zaslavsky et al 2009). While
combined (passive + active). Our experience
combination of seismic methods provides the Vs
model for the uppermost tens of meters, HVSR testifies that generally, all three MASW
modifications would give similar Vs sections
method adds information and constraints for the
development of a subsurface model at when the signal is strong enough (Signal-to-Noise
Ratio = 0.8-1.0) and the same initial layered
significantly greater depths (hundreds of meters).
In this study, we explored the potential of the model is used for inversion procedure. In this
Multichannel Analysis of the Surface Wave case, frequency range can be very similar. Such a
(MASW) method to provide shallow Vs models case is the Dam 5 study in the south of the Dead
for two sites where seismic refraction or Sea (Ezersky et al 2013b). However, in most

36
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

cases, passive and active MASW modes differ in deep reflector are adjusted.
the frequency range. Passive MASW is shifted to This case is an example of successful use of
the low-frequency range, while active MASW to the combined MASW and HVSR methods for
the relatively high-frequency range. estimating subsurface model down to the depth of
In the Ein Boqeq site passive MASW signal about 200 m when the geological model
was measured in the range of 2.5 - 12.0 Hz (Fig. comprises two hard layers within the soft
6c), whereas active signal was measured in the sediments.
range of 5.9 - 30 Hz (Fig. 6b). Such a difference In the Ramat Hakovesh site the following
in the frequency range results in the different geophysical measurements were carried out close
wavelength and, respectively, in the resolution of to 1A borehole at different times: P and S-wave
inversion results at the maximum depth and refraction survey, downhole and MASW
minimum (shallow) depth. In the latter example, measurements (Fig. 10d). The MASW survey
passive MASW shows higher wavelength at low- was performed in passive, active, and combined
frequency λmax = 174 m than the active one modifications. All the Vs versus depth graphs,
namely S-wave refraction section (thin solid line),
( λmax = 117 m). At the same time, the active
downhole (thick dashed line) and MASW based
MASW shows lower wavelength at higher ones (thick dashed line) derived from the
frequencies ( λmin = 11.6 m) as against combined dispersion curve (Fig. 4b) are shown in
λmin = 27.3 m measured with the passive method. Fig. 14. The optimal model calculated from a
Combining the two methods enables widening the combination of MASW and HVSR methods is
frequency range to 2.5-30 Hz and estimating the also shown by a thick solid line. General Vs
maximum penetration depth Z max in the range depth trend is presented by a dashed line marked
by circles.
of 87-174 m (e.g. 90 m, at least) and the From the comparative graph presented, it is
minimum penetration depth H 1 (= λmin /2) as 5.8 clearly seen that the depth of Vs sections obtained
m with determined resolution in the upper part of from different methods vary significantly. The
the section in active MASW.
Table 2 shows that seismic refraction method
allows constructing the Vs section down to 25 m
deep, whereas MASW enables constructing Vs
section below 47 m deep (including salt layer and
underlain sediments) that considerably facilitate
construction of optimal model.
A common feature of H/V spectral ratios
obtained at Ein Boqeq north site is two resonance
peaks at 0.65 Hz and in the range of 3.5-6 Hz.
(Fig. 8a). While the fundamental frequency is
related to the hard rock in depth, the position of
the second resonance peak is determined by an
intermediate hard layer in the subsurface. The
shallow salt layer connected with the second
resonance frequency is identified by the number
of boreholes (Fig. 4b, for instance) as a salt layer
found at a depth of 20-30 m and it is confidently
detected by both refraction and MASW methods
(Fig. 7a and 7b). We suggest that the deep
reflector is most likely also salt. Such assumption
is supported by data from deep boreholes Fig. 14 Comparison between different Vs versus depth
(Gardosh et al 1997). The depth and Vs of the models derived by different methods

37
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

refraction Vs model based on 250 m line with As is seen from a comparison of penetration
separation of 10 m between geophones has depths, S-wave refraction graph characterizes
penetration depth of maximum 75 m only uppermost part of rock foundation (~73-80
approximately. The downhole characterizes Vs m), whereas the MASW method characterizes
structure down to a depth of 68-70 m, which is depths down to 120 m, at least. No variations of
the depth where drilling met a karstified Vs were revealed in this depth range. We also
carbonate and was stopped. The extraction of the note that quality of the MASW signal is
drilling instrument from the borehole caused a significantly higher than the refraction one (Figs.
collapse and the lowermost part of borehole was 2d and 2c, respectively). This provides a more
filled with sediment. MASW penetration depth is reliable velocity-depth section obtained from the
evaluated from wavelength; that is MASW method. A good match between the
λmax = V ph max / f min = 239m and penetration depth analytical function and experimental spectral ratio
was reached when we extended the intermediate
is evaluated as between 120 m and 240 m
hard layer down to a depth of 250 m while
(accepting Z max = (0.5 − 1.0) ⋅ λ max ), e.g. 120 m at
retaining Vs profile obtained from MASW (Fig.
least. 13b) and added fundamental reflector composed
Comparing the velocity structure of different of dense dolomite with Vs = 2400 m/s. This
graphs with the optimal one derived from velocity value agrees well with Vs measured in
combined MASW and HVSR methods one can samples in the Technion laboratory (Frydman
see that the Vs versus depth trend shows the 2007).
gradual increase of Vs with the depth from Conclusions. In this study, we replaced S-
200 m/s near surface to 550 m/s at a depth of wave seismic refraction survey by MASW
refractor (70-73 m). With the general similarity, method in its active and passive modifications
graphs slightly differ from the optimal graph (by using 2.5 Hz geophones, and optimized data
50-100 m/s) at small (up to 40 m) depths. The acquisition and processing that improved
largest deviation is observed between refraction resolution of Vs section down to the bedrock
and downhole graphs. At depths greater than 40 located at 100 m depth and more.
m, all graphs are converged. Finally, shear-wave The MASW method has the following main
velocity of half space measured by S-wave advantages: (1) It is not limited by reverse
refraction method as 1400 m/s has been evaluated velocity structure; (2) It is based on highly
by MASW method as 1480 m/s. After energetic surface waves that provides good
optimization of the Vs-depth graph we obtain quality records; and (3) It is a low-cost, efficient,
Vs = 1440 m/s and overlain soil velocities as and relatively fast method. At the same time,
shown in Fig. 14 (thick solid graph). constructing an initial Vs model and applying
The discrepancy between the Vs for shallow constraints on the results of inversion are
subsurface sediments can be caused by several necessary.
reasons. The downhole located at approximately The methodology of joint use of HVSR
70 m north of Refraction and MASW lines is (based on ambient noise) and Multichannel
affected, evidently, by features of shallow Analysis of Surface Waves (MASW) based on
subsurface properties. In addition, the downhole surface wave records excited by seismic source
is also affected by the construction of the (active MASW) or ambient noise (passive one)
borehole walls, quality of drilling, cementation has been improved and tested in Israel, where the
and properties of the cement, required by the salt layers on the Dead Sea coast and hard
ASTM D4428/D4428M, (1991(1995)). It requires carbonates in the plain areas of Israel serve strong
an experienced drilling crew. Deviation of seismic wave reflectors. The main issue discussed
refraction Vs section can be caused by a large is: whether resolution of the MASW method is
separation (10 m) between geophones that enough to adequately replace and improve the
provides with low depth resolution at shallow refraction techniques.
depths and more than that at greater depths. At the two characteristic test sites, a

38
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

combination of MASW and HVSR methods Equations for estimating horizontal response
made it possible: spectra and peak accelerations from western
1. on the Ein Boqeq site (south of the Dead North American earthquakes: a summary of
Sea), to construct the subsurface structure down recent work. Seismological Research Letters,
to a depth of about 200 m and to detect two firm 68(1): 128 - 153.
reflectors composed of salt rock separated by soft BSSC, 1994. NEHRP recommended provisions
sediments; for seismic regulations for new buildings, Part
2. on the Ramat Hakovesh site (plain area in 1– Provisions, FEMA 222A. Federal
the Centre of Israel), to reveal that unlike Emergency Management Agency, 290p.
borehole and refraction survey data, the Ezersky, M., L. Bodet, E. Akkawi, A. Al-Zoubi,
fundamental reflector is the compact high C. Camerlynck, A. Dhemaied and P.Y.
velocity dolomite located at a depth of 260 m, Galibert, 2013a. Seismic Surface-wave
while the fractured limestone located at a depth of prospecting methods for sinkhole hazard
73 m overlying the dolomite is an intermediate assessment along the Dead Sea shoreline.
layer. Journal of Environmental and Engineering
Geophysics, 18(4): 233 - 253. DOI:
Acknowledgement 10.2113/JEEG18.4.233.
This publication was made possible through Ezersky, M., M. Gorstein, M. Kalmanovich and
support provided by the Ministry of Energy and N. Perelman, 2013b. Combining H/V spectral
Water Resources of Israel (Contracts No. 211-17- ratio (HVSR) technique with method using
007 and 213-17-004). We thank the M. surface waves (MASW) to map the
Kalmanovich for providing geological materials. subsurface: implementation in the Dead Sea
We are grateful also to V. Giller and P. Portnov area. Unpublished GII Report 025/759/13.
who acquire the data in complicated climate Fäh, D., F. Kind and D. Giardini, 2001. A
conditions. We thank G. Rix for cooperation. We theoretical investigation of average H/V ratios.
are very grateful to anonymous reviewers whose Geophys. J. Int., 145(2): 535 - 549. DOI:
great efforts allowed improvement of the 10.1046/j.0956-540x.2001.01406.x.
manuscript. Frydman, S., 2007. Geotechnical study of site
response Ramat Hakovesh. National Building
research Institute. Unpublished Report No.
References
2006686, 166p.
ASTM, 1991(1995). Standard D4428/D4428M. Gardosh, M., E. Kashai, S. Salhov, H. Shulman
Standard Test Methods for Crosshole Seismic and E. Tannenbaum, 1997. Hydrocarbon
Testing. exploration in the southern Dead Sea area.
ASTM, 2007. Standard D7400. Standard Test The Dead Sea: the lake and its setting. Oxford
Methods for Downhole Seismic Testing. University Press, Oxford, pp: 57 - 71.
ASTM International, USA. Jakosky, J., 1957. Exploration geophysics: Trija
Bodet, L., K. Van Wijk, A. Bitri, O. Abraham, P. Publishing Company, Los Angeles, 1195p.
Côte, G. Grandjean and D. Leparoux, 2005. Jarpe, S. P., C. H. Cramer, B.E. Tucker and A.F.
Surface-wave inversion limitations from Shakal, 1988. A comparison of observations
laser-Doppler physical modeling. Journal of of ground response to weak and strong motion
Environmental and Engineering Geophysics, at Coalinga, California. Bull. Seism. Soc. Am.,
10: 151 - 162. 78: 421 - 435.
Bonnefoy-Claudet, S., F. Cotton and P.Y. Bard, Kagami, H., C.M. Duke, G.C. Liang and Y. Ohta,
2006. The nature of the seismic noise wave 1982. Observation of 1- to 5-second
field and its implication for site effects studies: microtremors and their application to
a literature review. Earth Sci. Rev. 79(3-4): earthquake engineering. Part II. Evaluation of
205 - 227. site effect upon seismic wave amplification
Boor, D.M., B.V. Joyner and T.E. Fumal, 1997. deep soil deposits. Bull. Seism. Soc. Am., 72:

39
IJGE 2015 1(1): 20-41 Gorstein and Ezersky

987 - 998. Heljeson (ed), Kansas Geological Survey.

Lachet, C. and P.Y. Bard, 1994. Numerical and Kansas, USA, 40p.
theoretical investigations on the possibilities www.kgs.ku.edu/software/surfseis/index.htm
and limitations of the Nakamura’s technique. l
J. Phys. Earth, 42: 377 - 397. Park, C.B., R.D. Miller and J. Xia, 1999.
Lermo, J. and F. J. Chávez-García, 1994. Are Multichannel analysis of surface waves.
microtremors useful in site response Geophysics, 64(3): 800 - 808.
evaluation? Bull. Seism. Soc. Am., 84: 1350 - Park, C.B., R.D. Miller, N. Ryden, J. Xia and J.
1364. Ivanov, 2005. Combined use of active and
Louie, J.N., 2001. Faster, better: shear-wave passive surface waves. JEEG, 10(3): 323 -
velocity to 100 meters depth from refraction 334.
microtremor arrays. Bull. Seism. Soc. Am., Park, C.B. and R.D. Miller, 2006. Roadside
91(2): 347 - 364. Passive MASW. SAGEEP 2006, 12p.
Malischewsky, P. G. and P. Scherbaum, 2004. Park, C.B., R.D. Miller, J. Xia and J. Ivanov,
Love’s formula and H/V-ratio ellipticity of 2007. Multichannel analysis of surface waves
Rayleigh waves. Wave Motion 40: 57–67. (MASW) – active and passive methods. The
Malischewsky, P.G., Y. Zaslavsky, M. Gorstein, Leading Edge, 26(1): 60 - 64.
V. Pinsky, T.T. Tran, F. Scherbaum and F. Perelman, N. and Y. Zaslavsky, 2001. Analysis of
Estrela., 2010. Some new theoretical Seismic Signals in Frequency Domain,
considerations about the ellipticity of (SEISPECT). Unpublished GII Report
Rayleygh waves in the light of site-effect 569/345/01.
studies in Israel and Mexico. Geofisica Richart, F. E., J. R. Hall and R. D. Woods, 1970.
Internacional, 49(3): 141 - 152. Vibrations of Soils and Foundations. Prentice
Mucciarelli, M. and M.R. Gallipoli, 2004. The Hall, Englewood Cliffs, 414p.
HVSR Technique from Microtremor to Rix, G.J. and A.E. Leipski, 1991. Accuracy and
Strong Motion: Empirical and Statistical resolution of surface wave inversion. In:
Considerations. 13th World Conference of Bhatia S.K. and Blaney G.W. (eds.), Recent
Earthquake Engineering, Vancouver, B.C., Advances in Instrumentation, Data
Canada, Paper 45. Acquisition and Testing in Soil Dynamics:
Nakamura, Y., 1989. A method for dynamic Am. Soc. Civil Eng. Geotechnical Special
characteristics estimation of subsurface using Publication No. 28, pp: 17 - 23.
microtremor on the ground surface. Quarterly Satoh, T., T. Sato and H. Kawase, 1995.
Report of RTRI, 30: 25 - 33. Nonlinear behavior of soil sediments
Okada, H., 2003, The microtremor survey method. identified by using borehole records observed
Geophysical Monograph Series, no. 12. at the Ashigara valley, Japan. Bull. Seism.
published by Society of Exploration Soc. Am., 85: 1821 - 1834.
Geophysicists (SEG), Tulsa, OK. Scherbaum, F., K.G. Hinzen, M., Ohrnberger,
O'Neill, A., 2003. Full-waveform Reflectivity for 2003. Determination of shallow shear wave
Modelling, Inversion and Appraisal of velocity profiles in the Cologne, Germany
Seismic Surface Wave Dispersion in Shallow area using ambient vibrations, Geophys. J. Int.
Site Investigations. Ph.D. thesis, The 152: 597 - 612.
University of Western Australia, School of Schnabel P.B., J. Lysmer and H. B. Seed, 1972.
Earth and Geographical Sciences. SHAKE: a Computer Program for Earthquake
Palmer, D., 1986. Refraction Seismics. The Response Analysis of Horizontally Layered
Lateral Resolution of Structure and Seismic Sites. Report No. EERC72-12, University of
Velocity. Geophysical Press, London- California, Berkeley.
Amsterdam. Seekins, L.C., L. Wennerberg, L. Margheriti and
Park, C.B., 2006. SurfSeis. Active and passive H.P. Liu, 1996. Site amplification at fife
MASW. User’s Manual. V.2.0. M. Adkins- locations in San Francisco, California: a

40
Combination of HVSR and MASW Methods to Obtain Shear Wave Velocity Model IJGE 2015 1(1): 20-41

comparison of S waves, codas and with fundamental and higher modes. Journal
microtremors, Bull. Seism. Soc. Am., 86: 627 of Applied Geophysics, 52: 45 - 57.
- 635. Yamanaka, H., M. Takemura, H. Ishida and M.
Shapira, A. and V. Avirav, 1995. PS-SDA Niva, 1994. Characteristics of long-period
Operation Manual. Unpublished IPRG Report microtremors and their applicability in
No. Z1/567/79. exploration of deep sedimentary layers. Bull.
Shtivelman, V., 1999. Using surface waves for Seism. Soc. Am., 84: 1831 - 1841.
estimating the shear-wave velocities in the Yechieli, Y., M. Abelson, A. Bain, V. Shtivelman,
shallow subsurface onshore and offshore C. Onn, D. Wacks, G. Bar, R. Kalbo and V.
Israel. Europ. Journ. Env. Eng. Geoph., 4: 17 Lyachovski, 2004. Sinkhole phenomena along
- 36. the Dead Sea shore line. Unpublished GSI
Socco, L.V. and C. Strobia, 2004. Surface-wave Report No. GSI/21/04 (in Hebrew).
method for near-surface characterization: a Zaslavsky, Y., A. Shapira, M. Gorstein, M.
tutorial. Near Surface Geophysics, 2: 165 - Kalmanovich, V. Giller, N. Perelman, I.
185. Livshits, D. Giller and I. Dan, 2005. Site
Socco, L.V., D. Boiero, S. Foti and R. Wisén, response from ambient vibrations in the towns
2009. Laterally constrained inversion of Lod and Ramle (Israel) and earthquake hazard
ground roll from seismic reflection records. assessment, Bulletin of Earthquake
Geophysics, 74(6): G35 - G45. Engineering, 3(3): 355-381. DOI:
Stokoe II, K.H., A. Kurtulus and K. Park, 2006. 10.1007/s10518-005-1243-1.
Development of field methods to evaluate the Zaslavsky, Y., G. Ataev, M. Gorstein, M.
nonlinear shear and compression moduli of Kalmanovich, A. Hofstetter, N. Perelman, T.
soil. Earthquake Geotechnical Engineering Aksinenko, Giller, V., H. Dan, D. Giller, I.
Workshop, Canterbury 2006 (EGE 2006), Livshits, A. Shvartsburg, and A. Shapira,
Christchurch, New Zealand, November 21 - 2008. Microzoning of site response
24, 2006. parameters in the towns of Dimona and Bet
Stokoe, K.H., S.G. Wright, J. Bay and J.M. Shean (Israel), Bollettino di Geofisica, teorica
Roesset, 1994. Characterization of and applicate, 49(1): 109 - 130.
geotechnical sites by SASW method. In : Zaslavsky Y., A. Shapira, G. Ataev, M. Gorstein,
Woods, R.D. (ed.), Geophysical T. Aksinenko, M. Kalmanovich, N. Perelman
Characterization of Sites (ISSMFE TC#10), and R. Hofstetter, 2009. Using Ambient Noise
15 - 25. Oxford and IBH Publishers. Measurements in the Process of Assessing
Storn, R. and K., Price, 1995. Differential Earthquake Hazards in Urban Areas:
Evolution: A Simple And Efficient Adaptive Examples from Israel. In: T. Miura and Y.
Scheme For Global Optimization Over Ikeda (eds), Earthquake Engineering: New
Continuous Spaces. Technical Report TR-95- research. Nova Publisher, 155 - 218.
012, International Computer Science Institute, Zaslavsky, Y., A. Shapira, M. Gorstein, N.
Berkeley. Perelman, G. Ataev, T. Aksinenko, 2012.
Xia, J, R.D. Miller and C.B. Park, 1999. Questioning the applicability of soil
Estimation of near-surface shear-wave amplification factors as defined by NEHRP
velocity by inversion of Rayleigh waves. (USA) in the Israel building standards.
Geophysics, 64(3): 691 - 700. Natural Science, 4: 631 - 639. DOI:
Xia, J., R.D. Miller, C.B. Park and G. Tian, 2003. 10.4236/ns.2012.428083.
Inversion of high frequency surface waves

41