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

net/publication/238361190

Applications of engineering seismology for site characterization

Article  in  Journal of Earth Science · June 2009

DOI: 10.1007/s12583-009-0045-9

CITATIONS READS

13 683

3 authors, including:

Murat Eser Mehmet M. Berilgen

Eagean Geophysical Yildiz Technical University
7 PUBLICATIONS   69 CITATIONS    31 PUBLICATIONS   285 CITATIONS   

SEE PROFILE SEE PROFILE

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

Performans of Deep Excavation Projects View project

All content following this page was uploaded by Mehmet M. Berilgen on 24 June 2015.

The user has requested enhancement of the downloaded file.

Journal of Earth Science, Vol. 20, No. 3, p. 546–554, June 2009 ISSN 1674-487X
Printed in China
DOI: 10.1007/s12583-009-0045-9

Applications of Engineering Seismology for

Site Characterization

Oz Yilmaz*, Murat Eser

Anatolian Geophysical, 28 Lalebayiri, Kemer Country, Kemerburgaz, Istanbul 34077, Turkey
Mehmet Berilgen
Yildiz Technical University, Istanbul, Turkey

ABSTRACT: We determined the seismic model of the soil column within a residential project site in
Istanbul, Turkey. Specifically, we conducted a refraction seismic survey at 20 locations using a receiver
spread with 484.5-Hz vertical geophones at 2-m intervals. We applied nonlinear tomography to
first-arrival times to estimate the P-wave velocity-depth profiles and performed Rayleigh-wave inver-
sion to estimate the S-wave velocity-depth profiles down to a depth of 30 m at each of the locations. We
then combined the seismic velocities with the geotechnical borehole information regarding the lithology
of the soil column and determined the site-specific geotechnical earthquake engineering parameters for
the site. Specifically, we computed the maximum soil amplification ratio, maximum surface-bedrock
acceleration ratio, depth interval of significant acceleration, maximum soil-rock response ratio, and de-
sign spectrum periods TA–TB. We conducted reflection seismic surveys along five line traverses with
lengths between 150 and 300 m and delineated landslide failure surfaces within the site. We recorded
shot gathers at 2-m intervals along each of the seismic line traverses using a receiver spread with
4 840-Hz vertical geophones at 2-m intervals. We applied nonlinear tomography to first-arrival times to
estimate a P-wave velocity-depth model and analyzed the reflected waves to obtain a seismic image of
the deep near-surface along each of the line traverses.
KEY WORDS: engineering seismology, geotechnical engineering, earthquake engineering, shear-wave
velocity.

INTRODUCTION we estimate an S-wave velocity-depth profile for the

A geotechnical project involves interdisciplinary soil column from the analysis of surface waves. The
coordination among the seismologist, the geomor- parameters associated with the soil geometry, soil pe-
phologist, and the geotechnical engineer (Fig. 1). We dology, and the soil dynamics constitute the geotech-
obtain a seismic image of the near-surface from the nical model of the soil column, which is then used for
analysis of reflected waves to delineate fault geometry the geotechnical design for soil remediation.
and geometry of layers within the soil column. We Figure 2 shows the location of the site. The size
then obtain a P-wave velocity-depth model of the soil of the area is approximately 40 acres. Elevations vary
column from the analysis of refracted waves. Finally, between 125 and 180 m. The topography is fairly flat
at the western half of the site while there is a downhill
*Corresponding author: oz@anatoliangeo.com slope in the northerly direction at the eastern half. The
top soil is entirely clay. The site has been designated
Manuscript received October 17, 2008. for a residential project that involves the construction
Manuscript accepted February 11, 2009. of multistorey apartment blocks.
Applications of Engineering Seismology for Site Characterization 547

Figure 1. Workflow for a geotechnical project in- Figure 3. Example common-shot gathers acquired
cludes seismic surveys to delineate the soil geome- by a 94-m receiver spread (the green bar on top) at
try and estimate the P- and S-wave velocities of the location KS08. The red asterisks denote the shot
soil column. locations—two shots with 2-m offset from the ends
of the receiver spread and a shot at the center of
the spread.

was chosen such that the elevation change along the

receiver spread was minimal.
Along each of the five reflection-survey lines
within the project site (Fig. 2), we deployed a receiver
spread with 4 840-Hz vertical geophones at 2-m in-
tervals. We used a hand hammer and a steel plate, and
acquired hundreds of shot records at 2-m intervals
along each of the seismic lines with lengths between
Figure 2. Location map for the seismic survey 150 and 300 m. The sampling rate for recording was
conducted within the residential development pro- 0.5 ms and the record length was 1 s.
ject site. The survey consisted of refraction profil-
ing at 20 locations (KS01–20) and reflection profil- ANALYSIS OF REFRACTED AND SURFACE
ing along five line traverses (YS05–09). The white WAVES
lines denote the orientation of the receiver spreads By applying a nonlinear traveltime tomography
for the refraction survey. The green curves denote (Zhang and Toksoz, 1997) to the first-arrival times
the extremities of the landslide failure zone. picked from the three shot records (Fig. 4), we esti-
mated a near-surface P-wave velocity-depth model
DATA ACQUISITION along the receiver spread at each of the 20 locations.
At each of the 20 refraction-survey locations By applying smoothing during the inversion and lat-
(KS01–KS20) within the project site (Fig. 2), we de- eral averaging after the inversion, we then obtained a
ployed a receiver spread with 484.5-Hz vertical geo- P-wave velocity-depth profile down to a depth of 30
phones at 2-m intervals. We used a buffalo gun at the m representative of each location.
bottom of a charge hole with a diameter of 10 cm and We applied a workflow (Yilmaz and Eser, 2002)
a depth of 30 cm, and acquired three shot records with for the analysis of the first-arrival times that are pri-
source locations at each end of the spread and at the marily associated with refracted waves. First, an initial
center of the spread (Fig. 3). The sampling rate for P-wave velocity-depth model was derived from the
recording was 0.5 ms and the record length was 2 s. traveltimes picked from the field records. Then, this
The orientation of the receiver spread at each location
548 Oz Yilmaz, Murat Eser and Mehmet Berilgen

Figure 4. First-arrival times (denoted by the red asterisks) picked from three shot records at location KS02
(top), and the P-wave velocity-depth model estimated by nonlinear inversion of the traveltimes (bottom).

‘initial’ model was perturbed iteratively by nonlinear

traveltime tomography to estimate a ‘final’ P-wave
velocity-depth model (Fig. 4). At each iteration,
first-arrival times were modeled and compared with
the actual (picked) traveltimes. Iteration was stopped
when the discrepancy between the modeled and actual
traveltimes was reduced to an acceptable minimum.
Next, we identified at each KS location the off-
end shot record with the most pronounced dispersive
Rayleigh-type surface-wave pattern (Fig. 5), which
was first isolated from the refracted and reflected
waves by inside and outside mutes, then filtered using Figure 5. Dispersive surface waves isolated from
a 2, 4–36, 48-Hz passband to remove low- and high- the reflected and refracted waves on a shot record
frequency noise. We then performed plane-wave de- at one end of the receiver spread at location KS13
composition to transform the data from offset-time to (left), and the S-wave velocity-depth profile (the
phase-velocity versus frequency domain (Fig. 6). A blue curve) estimated by inversion of the surface
dispersion curve associated with the fundamental waves (right). The vertical axis in depth is associ-
mode of the surface waves was picked in the trans- ated with the S-wave velocity and the vertical axis
form domain based on the maximum-energy criterion in frequency is associated with the dispersion curve
(Fig. 6) and inverted to estimate the S-wave velocity as in Fig. 6. The S-wave velocity-depth profile was
as a function of depth (Fig. 5; Park et al., 1999; Xia et estimated by an iterative inversion of the disper-
al., 1999). In this procedure, initial depth-profiles for sion curve for the Rayleigh-wave fundamental
P- and S-wave velocities are iteratively perturbed until mode shown in Fig. 6. Iteration is stopped when the
a final S-wave velocity-depth profile is estimated. At discrepancy between the modeled (the black curve
each iteration, modeled dispersion values and the on the right) and actual dispersion values (the
picked (actual) dispersion values are compared. black dots on the right) is reduced to an acceptable
minimum.
Applications of Engineering Seismology for Site Characterization 549

Figure 8. The S-wave velocity map for a depth level

Figure 6. Dispersion spectrum of the surface waves of 12 m created by three-dimensional interpolation
in the shot record shown in Fig. 5, computed after of the S-wave velocity-depth functions as in Fig. 5
the plane-wave decomposition. In this figure, the associated with the 20 locations (indicated by the
vertical axis represents the phase velocity of Ray- red dots) where the refraction seismic survey was
leigh waves. Note that each frequency component conducted. The black polygon defines the bounda-
of the Rayleigh waves travels at a different speed, ries of the project site.
thus the dispersive character of these waves within
the soil column. The largest portion of the
Rayleigh-wave energy often, but not always, is as-
sociated with the fundamental mode. For this mode,
a dispersion curve that represents the change of
phase velocity with frequency is picked as shown.
Then, this dispersion curve is used in an inversion
algorithm to estimate the S-wave velocity-depth
profile for the soil column at the KS locations (Fig.
5).

Figure 9. The P- to S-wave velocity ratio map for a

depth level of 12 m created by three-dimensional
interpolation of the P- and S-wave velocity-depth
functions associated with the 20 locations (indi-
cated by the red dots) where the refraction seismic
survey was conducted. The P-wave velocity func-
tions were derived by lateral averaging of the
P-wave velocity-depth models as in Fig. 4 at each of
the KS locations. The S-wave velocity-depth func-
tions are as in Fig. 5. The black polygon defines the
Figure 7. The P-wave velocity map for a depth level
boundaries of the project site.
of 12 m created by three-dimensional interpolation
of the P-wave velocity-depth functions associated
Iteration is stopped when the discrepancy between the
with the 20 locations (indicated by the red dots)
modeled and actual dispersion values is reduced to an
where the refraction seismic survey was conducted.
acceptable minimum. The velocity estimation from
These velocity functions were derived by lateral
surface waves represents a lateral average over the re-
averaging of the P-wave velocity-depth models as
ceiver spread length in contrast with the velocity esti-
in Fig. 4 at each of the KS locations. The black po-
mation from borehole seismic measurements that are
lygon defines the boundaries of the project site.
influenced by localized lithologic anomalies and
550 Oz Yilmaz, Murat Eser and Mehmet Berilgen

borehole conditions. Given the S-wave velocity-depth profile, the geotech-

Following the analysis of refracted waves to es- nical borehole information, and the accelerogram that
timate the P-wave velocities and the analysis of sur- describes the ground motion at the rock site, we cal-
face waves to estimate the S-wave velocities at each culated the accelerogram that simulates the ground
of the 20 locations, we interpolated the velocities and motion at the soil site (Fig. 12) for each of the three
generated velocity contour maps for depth levels of 0 sections within the site. This one-dimensional site-
to 30 m at 4-m intervals (Figs. 7 and 8). Additionally, response analysis was performed using a frequency-
we computed the ratio of the P-wave velocities to the domain algorithm that models the nonlinear material
S-wave velocities as a function of depth and generated behavior of the soil column as an equivalent linear
the velocity-ratio contour maps (Fig. 9). Based on the system (Bardet et al., 2000; Schnabel et al., 1972).
velocity variations, the site was divided into three In earthquake engineering, the soil response to an
sections—the western section with relatively low earthquake motion is calculated based on the scenario
S-wave velocities, the eastern section with relatively that corresponds to a maximum possible peak ground
high S-wave velocities, and the transitional central acceleration that may occur at a location (Kramer,
section (Fig. 10). We then drilled geotechnical bore- 1996). Hence, the accelerogram at the rock site (equi-
holes within each section down to a depth of 50, 40, valently, at the soil-bedrock interface) was actually
and 30 m at locations coincident with the center of the upscaled to a maximum value of 0.3 g before the
spreads KS05, KS10, and KS14, respectively. modeling of the accelerogram at the soil sites within
each section (KS05, KS10, and KS14).
ESTIMATION OF GEOTECHNICAL From Fig. 12a, we determined for each section
EARTHQUAKE ENGINEERING PARAMETERS the ratio of the maximum ground acceleration at the
For each section within the project site (western, soil site to the maximum acceleration at the soil-
central, and eastern), the S-wave velocities within the bedrock interface—often referred to as the maximum
soil column combined with the borehole lithology (Fig. surface-bedrock acceleration ratio. For each section,
11) were used to determine the geotechnical earth- we also computed the maximum acceleration as a
quake engineering parameters. We began with a rock- function of depth, and determined the depth range for
site SH accelerogram that describes the time history of which the surface-bedrock acceleration ratio is sig-
the strong ground motion associated with the August nificant (Fig. 13a). Specifically, as the bedrock motion
1999 Izmit earthquake (Yilmaz et al., 2006). is upward propagated through the soil column, maxi-
mum acceleration values are extracted from the

Figure 10. The S-wave velocity map for a depth

level of 28 m. The red dots represent the spread
centers at the 20 locations where the seismic survey
was conducted. The blue circles denote the geo- Figure 11. Correlation of the S-wave velocities with
technical borehole locations for each of the three the lithology column from the geotechnical bore-
sections. The black polygon defines the boundaries hole at the center of the spread at location KS10
of the project site. (central section in Fig. 10).
Applications of Engineering Seismology for Site Characterization 551

accelerograms computed at each of the discrete depth amplitude frequency is called the natural period of the
levels (Bardet et al., 2000) to obtain the curve shown soil column (0.3 s). Also from Fig. 13b, we computed
in Fig. 13a. Next, we calculated the smoothed ratio of the ratio of the peak spectral amplitude at the soil site
the amplitude spectra of the measured and modeled to the spectral amplitude corresponding to the same
accelerograms (Fig. 12b) as shown in Fig. 13b, and frequency at the rock site (2.9). This is called the
determined the peak spectral amplitude at the soil site maximum soil amplification ratio, which is a measure
and the frequency at which this peak occurs (3.3 Hz). of how much the soil column amplifies the earthquake
The latter is called the natural frequency of the soil motion that occurs at the soil-bedrock interface.
column and the period that corresponds to the peak-

Figure 12. Analysis for geotechnical earthquake engineering parameters at site KS10 (central section in Fig.
10). (a) The accelerogram that describes the bedrock motion associated with the August 1999 Izmit earth-
quake measured at the rock site (green), which is assumed to be equivalent to the response at the soil-
bedrock interface at the soil site (upscaled to a maximum acceleration value of 0.3 g), and the accelerogram
that describes the ground motion at the soil site KS10 within the central section (red) modeled by using the
velocity and borehole information given by Fig. 11. The maximum surface-bedrock acceleration ratio is 1.5.
(b) The amplitude spectra of the measured and modeled accelerograms. Note that the significant amplifica-
tion within soil column occurs between 0 and 12 Hz bandwidth.

Next, for each section within the site, we com- sponse spectra at the ground level (8 g) and soil-
puted the response spectra (Fig. 13c) that describe the bedrock interface (3 g), and computed the maximum
response of structures (buildings) with a range of nat- soil-rock response as the spectral acceleration ratio
ural periods to the modeled ground motion at the soil (2.4). We also determined the design spectrum periods
site and the actual ground motion at the rock site. The TA–TB (0.05–0.55 s). TA and TB correspond to the
structure is defined as a spring system with a single minimum and maximum periods for which the spec-
degree of freedom, usually with a damping ratio of 5%. trum is nearly flat. Outside the TA–TB bandwidth, the
From Fig. 13c, we determined the maxima of the re- spectrum ramps down rapidly.
552 Oz Yilmaz, Murat Eser and Mehmet Berilgen

Figure 13. Analysis for geotechnical earthquake engineering parameters at site KS10 (central section in Fig.
10). (a) Maximum acceleration as a function of depth; (b) soil-bedrock amplification ratio; and (c)
site-specific design spectra for the rock site (green) and the soil site (red), and the code-specific design spec-
trum for the soil site (yellow). Note that the most significant acceleration values at the rock site occur in
TA–TB: 0.05–0.55 s.

Table 1 Geotechnical earthquake engineering parameters

Maximum soil Natural period of Maximum surface- Depth interval Maximum Design spectrum
Section amplification the soil column bedrock accelera- with significant soil-rock periods TA–TB
ratio (s) tion ratio acceleration (m) response (sec)
Western 2.2 0.4 1.3 0–10 1.3 0.05–0.60
Central 2.4 0.3 1.5 0–15 2.4 0.05–0.55
Eastern 2.8 0.15 1.7 0–10 2.1 0.05–0.65

Listed in Table 1 are the geotechnical earthquake includes application of deconvolution, time-variant
engineering parameters for each of the three sections spectral whitening, bandpass filtering, and automatic
within the site. Combined with the parameters for the gain control. Aside from deriving a seismic section
soil dynamics, such as the bearing capacity, these pa- that represents the subsurface image down a depth of
rameters are used by the geotechnical engineer for soil 150 m (Fig. 14), we also estimated the near-surface P-
classification and by the civil engineer for structural wave velocity-depth model, again using the nonlinear
design of buildings. traveltime tomography, for each of the five line trav-
erses (Fig. 15). The nonlinear tomography solution is
ANALYSIS OF REFLECTED WAVES based on not just the first-arrival times but also
In contrast to a comprehensive processing se- changes in traveltime gradient. As such, within the
quence applied to reflection seismic data used in ex- near-surface, we were able to resolve strong lateral
ploration for oil and gas fields (Yilmaz, 2001), shal- velocity variations associated with the landslide fail-
low reflection seismic data usually require a simple ure surfaces (Fig. 15).
processing sequence (Steeples and Miller, 1990) that
Applications of Engineering Seismology for Site Characterization 553

Figure 14. The seismic image of the deep near-surface along the YS09 line traverse (line location shown in
Fig. 2). This image was derived by analysis of reflected waves in the shot gathers recorded along the line
traverse at 2-m intervals using a receiver spread with 4 840-Hz vertical geophones. The black curve denotes
approximately the topography along the line traverse. The high-amplitude reflectors indicated by the red
arrows correspond to the sandstone interbeddings within the predominantly clayey soil column.

ing the subsequent geotechnical borehole investiga-

tions, have been confirmed as landslide masses that
are widespread within the project site (Fig. 2). Several
landslide failure surfaces have been delineated as
shown in Fig. 15.

CONCLUSIONS
Site investigations require multidisciplinary par-
ticipation by the geologist, the seismologist, and the
geotechnical and earthquake engineers. The contribu-
tion of the seismologist includes a seismic image of
the near-surface from the analysis of reflected waves
to delineate fault geometry and geometry of layers
within the soil column, a P-wave velocity-depth model
of the soil column from the analysis of refracted
Figure 15. The P-wave velocity-depth models along waves, and an S-wave velocity-depth profile for the
the YS line traverses (line locations shown in Fig. soil column from the analysis of surface waves. As
2). The pink zones are associated with the low- such, the role of the seismologist is to provide these
velocity landslide masses. The white curves denote intermediate products to the geotechnical engineer and
the potential landslide failure surfaces. earthquake engineer so as to define the parameters
associated with the soil geometry, soil pedology, and
The seismic images along the YS line traverses the soil dynamics which constitute the geotechnical
exhibit horizontal, piecewise continuous, but strong model of the soil column. In this case study, we dem-
reflectors that are associated with sandstone interbed- onstrated the role of the seismic method in defining
dings within the otherwise predominantly clayey soil the geotechnical model of the soil column, which can
column (Fig. 14), whereas the P-wave velocity-depth then be used for geotechnical design for soil remedia-
models along the YS line traverses exhibit low- tion.
velocity near-surface layers (Fig. 15), which, follow-
 554 Oz Yilmaz, Murat Eser and Mehmet Berilgen

ACKNOWLEDGMENTS Steeples, D. W., Miller, R. D., 1990. Seismic Reflection Meth-

We thank the Hanyapi Real Estate Development ods Applied to Engineering, Environmental, and Ground-
Corp., Istanbul, and Professor Ahmet M Işıkara for water Problems. In: Ward, S. H., ed., Geotechnical and
giving us the opportunity to conduct this integrated Environmental Geophysics. Soc. of Expl. Geophys., Tulsa,
seismic project and granting the permission to publish OK. 1–30
the results. Xia, J., Miller, R. D., Park, C. B., 1999. Estimation of
Near-Surface Shear-Wave Velocity by Inversion of Ray-
REFERENCES CITED leigh Waves. Geophysics, 64: 691–700
Bardet, J., Ichii, K., Lin, C., 2000. Manual of EERA: A Com- Yilmaz, O., 2001. Seismic Data Analysis—Processing, Inver-
puter Program for Equivalent-Linear Earthquake Site Re- sion, and Interpretation of Seismic Data. Soc. of Expl.
sponse Analysis of Layered Soil Deposits. University of Geophys., Tulsa, OK
Southern California, Los Angeles Yilmaz, O., Eser, M., 2002. A Unified Workflow for Engi-
Kramer, S. L., 1996. Geotechnical Earthquake Engineering. neering Seismology. Expanded Abstracts, 72nd Annual
Prentice-Hall, New Jersey. 273 International Meeting of the Society of Exploration Seis-
Park, C. B., Miller, R. D., Xia, J., 1999. Multichannel Analysis mologists, Houston, TX
of Surface Waves. Geophysics, 64: 800–808 Yilmaz, O., Eser, M., Berilgen, M. M., 2006. A Case Study for
Schnabel, P. B., Lysmer, P. B., Seed, H. B., 1972. SHAKE: A Seismic Zonation in Municipal Areas. The Leading Edge,
Computer Program for Earthquake Response Analysis of 25(3): 319–330
Horizontally Layered Sites. In: Report EERC 72-12, Zhang, J., Toksoz, M. N., 1997. Nonlinear Refraction Travel-
Earthquake Engineering Research Center. University of time Tomography. Geophysics, 63: 1726–1737
California, Berkeley

View publication stats