J. Phys.

Earth, 40, 285-295, 1992

Earthquake Response Analysis of Ground Surface

by Wave Front Tracing Method

Jingzhe Zheng,* Choshiro Tamura, and Kazuo Konagai

Instituteof IndustrialScience,The Universityof Tokyo,
Minato-ku,Tokyo 106,Japan

Earthquakeresponseof soft surfacelayer overlayingthe diluvialbasementwas

studiedby usingthe WaveFront Tracingmethod.We consideredtwo typesof surface
grounds:a thin layereddepositwith horizontal basement,and the surfacedeposit
with irregularshapeof basement.Mainattentionswere givento the followingaspects;
(I) the investigationsof the amplificationpropertiesof thin soft deposit; (2) the
comparisonbetweenthe responseof groundwith irregularshapeof basementand that
with horizontal basement.Towards this end, both theoreticalanalysis and the
simulationof observedearthquakemotionswerecarriedout.

I. Introduction
Investigations of the earthquake damage (e.g., JSCE, 1986) have revealed that
local ground structures largely affect the earthquake response characteristics at the
surface, and the destruction concentrates in the region that has irregular geological
structure and thick soft deposit. During these years, in order to apprehend the effects
of local site conditions on the seismicmotion at the surface, a number of experimental
studies (e.g., Shinozaki and Kobori, 1988; Tamura and Suzuki, 1988), theoretical
investigations (e.g., Aki and Lamer, 1970; Bard and Bouchon, 1985; Inoue and
Hakuno, 1981; Trifunac, 1971, 1990) and observed earthquake response analysis
(e.g., Kinoshita et al., 1982; Okamoto, 1973) have been carried out. These previous
studies have shown the importance of geological irregularities and local soil condi-
tions. In these studies, however, few efforts have been made to compare directly the
dynamic behaviors between surface grounds with and without geological irregularity,
and there have been few observations and analytic results for the thin soft deposit with
thickness less than 10m.
This paper attempts to investigate the following two subjects: (1) the earthquake
response of a thin soft layer with thickness of 9 m, which is modeled from the Subway
Hiroo observation site; (2) the comparison between the responses of soft surface
deposits with and without geological irregularity. Towards this end, the Wave Front

ReceivedNovember20, 1990;AcceptedSeptember,30, 1991

* Towhomcorrespondence shouldbeaddressed.Presentaddress:KajimaTechnicalResearch
Institute,KajimaCorporation,Chofu,Tokyo182,Japan.

285
286 J. Zheng et al.

Tracing method is used to simulate and analyze ground responses under plane SH
wave incidence.

2. Description of the Wave Front Tracing Method

We have proposed the Wave Front Tracing method (Zheng et al., 1989) for
the earthquake response analysis of one-dimensional and two-dimensional surface
grounds. Our method proceeds in two steps. First, we input a unit impulse from
the basement. Because of reflections and refractions at boundaries between different
media, the number of impulses increases, and the amplitude of impulses changes.
Impulse amplitude variations are calculated by using reflective and refractive
coefficients, which we can obtain from the impedance ratio and incident angle* (e.g.,
Saw, 1978). The travel times of impulses are acquired by tracing rays. Through the
above calculations, we can easily obtain impulse time series arriving at observation
point (Fig. 1). Second, based on the impulse time series, we compute the response at
the observation point for any kinds of input.
input: a(t)

response: (1)

And through the Fourier transform of the impulse time series, we derive the transfer
function as

(2)

The intrinsic attenuation is considered by using exp(-r), where =2h/(1-h2)L

and r is the propagation distance, L the wavelength and h the damping ratio. When
considering the damping effect, we use Ak exp(-rk) instead of Ak in Eqs. (1) and

(2), where rk is the propagation distance of impulse. As for diffractive waves, we

approximate wave amplitude by using the Fresnel integral.

Fig. 1. Calculation of impulse time series. Impulse amplitude is computed by

using reflective and refractive coefficients, and the arriving time of impulse
is acquired by tracing rays.

* In case of over critical incidence occurrence

, we assume the impulse shape will not be
changed. The propriety of this approximation has been testified by comparing with the results
from finite difference method_

J. Phys. Earth
 Earthquake Response Analysis of Ground Surface 287

(3)

where S is the propagation distance, L the half wavelength, and D the distance along
the wave front from the discontinuity point.

3. Analysis of Earthquake Response of Soft Surface Ground

In this section, first, we will analyze the earthquake response of the Subway Hiroo
observation site, where soil profile can be modeled as horizontally layered structure,
and then investigate the response property of soft deposit with a simple dipping
basement and that with concave basin-shaped basement by comparing them with
grounds overlaying a horizontal basement.

3.1 Response analysis of the Subway Hiroo observation site

We have conducted the earthquake observation at the Subway Hiroo observation
site since 1970. Figure 2(a) shows the soil profile of the site, which is obtained from
boring tests. Through the Fourier spectrum analysis of the earthquake records
obtained at the surface and inside the mudstone, we inferred the ground structure as
shown in Fig. 2(b), taking the results of PS logging for the similar ground condition
into account.
In order to understand the behavior of the thin soft surface during earthquakes,
we simulated and analyzed several earthquake records by using our method. We
assumed the wave is vertically incident from the basement, and used the records in the
deep site as input wave. Figure 3 is the calculated transfer function of the ground; the
fundamental frequency is 4.44 Hz. Figure 4 represents results in time domain. The
wave recorded in the deep site, which we used as input for this example, is shown in
Fig. 5. It can be seen that the calculated result is in quite good agreement with the
observed one both in the waveforms and amplitudes (Good agreement was obtained
for all other simulations). Table I shows the computed and observed maximum
acceleration at the free surface and observed maximum acceleration at mudstone for
four earthquakes. The predominant frequencies of observed ground motions at
mudstone are also shown in the same table. In the case of earthquake 1, the maximum
acceleration ratio between free surface and mudstone is as large as 5.87 because
the predominant frequency (4.31 Hz) of ground motion at mudstone is near the
fundamental frequency (4.44 Hz) of the ground. On the other hand, in the case of
earthquake 2, the maximum acceleration ratio becomes 1.49, since the predominant
frequency (1.71 Hz) lies in the frequency range where ground amplification ratio is
small. Through the investigations of a lot of observed earthquake motions, it was
found that the acceleration amplification ratio has the value from 1.02 to 5.78, which
depends on the predominant frequency range of earthquake motion at mudstone. The

Vol. 40, No. 1, 1992

288 J. Zheng et al.

(a) (b)

Fig. 2. (a) Soil profile from the boring tests. (b) Inferred ground structure.

Fig. 3. Computed transfer function of the ground at SubwayHiroo observation

site. Fundamental frequencyof the ground is 4.44Hz.

important fact is that, if the predominant frequency of the input is close to the
fundamental frequency of the ground, even such a thin soft layer (thickness H=9 m)
can largely amplify earthquake motions. Thus, in earthquake resistance design, we
should also consider the amplification effects of thin soft layer.

J. Phys. Earth
 Earthquake Response Analysis of Ground Surface 289

Fig. 4. Observed and calculated accelerations at the surface.

Fig. 5. Acceleration wave recorded at mudstone during the earthquake that

occurred on June 24, 1986. The wave is used as input for the simulation
shown in Fig. 4.

Table 1. Observed and calculated maximum acceleration and the predominant frequency of
acceleration recorded at mudstone.

3.2 Response analysis of ground with irregular shape ofbasement

3.2.1 Surface ground with a dipping basement
The next model is with a dipping basement as shown in Fig. 6. The impedance
ratio between soft deposit and hard basement is 1/6, and plane SH wave is vertically
incident from the basement. A is an arbitrary observation point at the free surface with

Vol. 40, No. 1, 1992

290 J. Zheng el al.

Fig. 6. Analytical model of a soft deposit with a dipping basement. The

impedance ratio between soft deposit and basement is 1/6. A is an arbitrary

observation point. The broken line represents the base for horizontal

basement case.

Table 2. Dipping angle and the number of impulses. Incident angle is 90.

thickness of H. Now, we investigate the relationship between dipping angle and the
pattern of impulse time series arriving at the A point. After a unit impulse is incident
from the basement, it repeats reflection and refraction inside the soft deposit, and at
last, propagates in a direction of the dip. The number of impulses arriving at the
surface decreases as the dip angle of the base increases. The relationship between the
number of the impulse and the dip angle can be expressed by

where is the dipping angle, and N the number of impulses. Table 2 shows the number

of impulses for the case that dipping angle is from 5 to 45.

The number, amplitudes, and time interval of impulses vary with the dipping

angle. Figure 7 shows the impulse time series when dipping angles are 10, 18, and
30% respectively. The results for a horizontal base case with the surface layer thickness

of H are compared by broken line in the same figure. The horizontal axis indicates the

dimensionless time t/T, where T is the time interval of impulses for horizontal base

case. The time durations between adjacent impulses are shorter than the case with

horizontal basement, and the time durations become smaller and smaller along with

time axis. These indicate that the ground with a dipping base largely amplifies the

higher frequency components of input waves. We can also see that impulse time series

pattern depends on the dipping angle.

Knowing the impulse time series, we can obtain the transfer function of the

surface ground and the response in time domain for arbitrary input. The transfer
function for the case with the dipping angle of 30 (damping coefficient 5 and 20%) is

shown in Fig. 8. Horizontal axis is the dimensionless frequency x/V, where x is the

distance from the merged point and V the shear wave velocity; = 27f, where f

represents wave frequency. Vertical axis indicates the amplification ratio. When

damping coefficient is 5%, the amplification ratio of the case with dipping base

J. Phys. Earth
 Earthquake Response Analysis of Ground Surface 291

Fig. 7. Comparison of impulse time series patterns. Impedance ratio is 1/6.

Fig. 8. Comparison between the transfer function of ground with a dipping

base (dipping angle 30) and that of the ground with horizontal basement.

obviously differs from that with a horizontal base. In the case that the damping ratio

is 20%, however, the distinction between the two cases becomes much smaller. For the
convenience of the following discussions, we define as

amplification ratio of the case with ay= horizontal base /

amplification ratio of the case with a dipping base

Vol. 40, No. 1, 1992

 292 J. Zheng et al.

Fig. 9. Analytical model of a soft deposit with a concave basin-shaped rigid

basement. S is the observation station above the intersecting point of dipping
basement and horizontal basement. A, B, C are three observation points.

With h =5%, reaches about 6 when x/V is around 1.0, and about 0.5 when x/V

is in the range of 1.5-2.1. If the predominant x/V of input wave is within 1.5-2.1,

the ground with a dipping base will show larger response than that with a horizontal

base. With h=20%, is almost 1.0 when x/V is larger than 2.6. If the predominant

x/V of input wave is larger than 2.6, the response of the cases with and without a

dipping base will perform in a similar manner. Therefore, the degree of difference in
the response of ground between with horizontal and dipping bases depends on the

dimensionless predominant frequency x/V range of input and the damping ratio.

3.2.2 Surface ground with concave basin-shaped basement

Figure 9 shows the last ground model with concave basin-shaped basement. The

plane SH wave is vertically incident from the basement, and the basement is rigid.
In order to evaluate the extent of the area affected by the dipping base, we have

used the transfer function difference index (Zheng et al., 1990), which was defined as

where fI represents the transfer function of ground with irregular shape of basement,

indicates that with horizontal basement. To make the index have more fHpractical

meanings, the transfer functions used are filtered first as shown in Fig. 10, taking the

acceleration response spectrum for the earthquake resistance design into account.
Here, we use the above index to evaluate the effects of the dipping angle on the

response at the free surface.

We first discuss the analyzed results for the case when dipping angle is 30 and

thickness is 30 m. In Fig. 11, the thick line denotes the transfer function of the point

S, which is the observation station above the intersecting point of dipping basement

and horizontal basement, and the thin line represents the result of horizontal

basement case with surface layer thickness of H. If x (the distance from S point)

increases, the transfer function of ground with concave basin-shaped basement will
change. The distribution of the transfer function difference index along with the x axis

is shown by solid line in Fig. 12. Figure 13 shows the responses at three observation

J. Phys. Earth
 Earthquake Response Analysis of Ground Surface 293

Fig. 10. Pass filter used for the calculation of the transfer function difference
index.

Fig. 11. Comparison between transfer function of S point for ground model
with concave basin-shaped basement and that for ground with horizontal
basement.

Fig. 12. Distribution of the transfer function differenceindex along with x axis.
Deposit thicknessis 30m.

points, A (x/H=1.0), B (2.0), and C (3.0) when a real earthquake wave is incident
from the basement. Transfer function index decreases as the distance x increases, and
when the transfer function index decreases, the difference between the responses of
these two grounds become smaller. It can also be seen that transfer function index
distribution is not smooth in the region where x/H is less than 1.0, which is
considered as one of the characteristics of the effects resulting from the dipping base.

Vol. 40, No. 1, 1992

294 J. Zheng et a!.

Fig. 13. Comparison between acceleration responses of ground model with

concave basin-shaped basement and that with horizontal basement. Accelera-

tion wave, which was recorded at Hiroo during the earthquake that occurred
on Jan. 7, 1987, is used as input.

To comprehend the effects of the dipping angle, the index distribution for the

case with dipping angle of 10 (case 1) is also shown in Fig. 12. The case with dipping

angle of 30 is defined as case 2. From the figure, it can be seen that index value for
the case 2 is larger than that of case 1 when x/H is smaller than 2.2, and the index for

the two cases tends to be the same value if x/H is larger than 2.2. Similar results were

obtained from the comparisons of the index distributions for other dipping angles.

Thus, we can infer that the large dipping angle has large effects when x is small, and

the effects of different dipping angles are similar in the region where x/H>2.2.

4. Conclusion

By using the Wave Front Tracing method, we investigated the earthquake

response of a thin soft deposit, and that of surface ground with basement of irregular

shape. The following conclusions are drawn:

1) When the predominant frequency of earthquake motion is near the funda-

mental frequency of the deposit, even a very thin soft deposit can largely amplify the

earthquake motion. At the Hiroo site (thickness is 9 m), the maximum acceleration

ratio between free surface and mudstone reached 5.87.

J. Phys. Earth
 Earthquake Response Analysis of Ground Surface 295

2) The degree of difference between the response of ground with a dipping base
and that with horizontal base depends on the dimensionless predominant frequency

(x/V) range of the input wave, damping ratio, and dipping angle. If the predominant
x/V is in the range where the difference between the transfer functions
of these two

grounds is large, they will perform very differently.

3) In the case of ground with concave basin-shaped basement, the effects of

dipping basement become smaller as the horizontal distance from the intersecting

point of dipping base and horizontal base increases. In the area where x/H is less than
2.2, larger dipping angle is more effective. In the region x/H> 2.2, response does not

depend on the dipping angle.

REFERENCES

Aki, K. and K. L. Larner, Surface motion of a layered medium having an irregular interface
due to incident plane SH waves, J. Geophys. Res., 75, 933-954, 1970.

Bard, P. Y. and M. Bouchon, The two-dimensional resonance of sediment-filled valleys, Bull.

Seismal. Soc. Am., 75, 519-541, 1985.

Inoue, R. and M. Hakuno, The influence of underground rigidity distribution to ground

motion, Proceedings of JSCE, No. 310, 1-10, 1981.
JSCE, Damage report of the Nihon-kai Central Part Earthquake in 1983, 1986.

Kinoshita, S., K. Itoh, T. Mikoshiba, and H. Suzuki, Observation of earthquake response of

thick sedimentary layers, Proceedings of the sixth Japan Earthquake Engineering Symposium,
169-176, 1982.

Okamoto, S., Introduction to Earthquake Engineering, the University of Tokyo Press, Tokyo, pp.
97-112, 1973.

Sato, Y., Elastic Wave Theory, Iwanami Press, Tokyo, pp. 75-77, 1978 (in Japanese).
Shinozaki, Y. and T. Kobori, Ground motion of subsurface irregular soil excited by vibrator,
Proceedings of ninth world conference on earthquake engineering, 671-676, 1988.
Tamura, C. and T. Suzuki, Proposal of a mathematical model for earthquake response analysis

of irregularly bounded surface layer, Proceedings of ninth World Conference on Earthquake

Engineering, 665-670, 1988.
Trifunac, M. D., Surface motion of a semi-cylindrical alluvial valley for incident-plane SH

waves, Bull. Seismol. Soc. Am., 61, 1755-1770, 1971.

Trifunac, M. D., How to model amplification of strong earthquake motions by local soil and

geologic site conditions, Earthq. Eng. Struc. Dyn., 19, 833-846, 1990.
Zheng, J., C. Tamura, and K. Konagai, A method for earthquake response analysis of surface

ground by tracing wave front, Seisan-Kenkyu, 41, 922-925, 1989.

Zheng, J., C. Tamura, and K. Konagai, Extent of area affected by a dipping basement, Proceed-
ings of annual convention of JSCE, 1128-1129, 1990.

Vol. 40, No. 1, 1992