Original Research
published: 27 February 2017
Edited by:
Katsuichiro Goda,
University of Bristol, UK
Reviewed by:
Hadi Ghofrani,
University of Western Ontario,
Canada
Naohiro Nakamura,
Hiroshima University, Japan
*Correspondence:
Izuru Takewaki
takewaki@archi.kyoto-u.ac.jp
Specialty section:
This article was submitted to
Earthquake Engineering,
a section of the journal
Frontiers in Built Environment
Received: 15December2016
Accepted: 08February2017
Published: 27February2017
Citation:
KamagataS and TakewakiI (2017)
Occurrence Mechanism of Large
Acceleration in KiK-net Seismic
Records during IwateMiyagi Nairiku
Earthquake in 2008.
Front. Built Environ. 3:13.
doi: 10.3389/fbuil.2017.00013
Frontiers in Built Environment | www.frontiersin.org
doi: 10.3389/fbuil.2017.00013
Occurrence Mechanism of Large
Acceleration in KiK-net Seismic
Records during IwateMiyagi Nairiku
Earthquake in 2008
Shuichi Kamagata1 and Izuru Takewaki2*
Nuclear Power Department, Kajima Corporation, Tokyo, Japan, 2Department of Architecture and Architectural Engineering,
1
Graduate School of Engineering, Kyoto University, Kyoto, Japan
The KiK-net is a vertical array system, and seismic motions are recorded at the ground
surface and in the borehole. With the peak acceleration of 24.5m/s2 at AKTH04 (EW)
and 38.7 m/s2 at IWTH25 (UD) during the IwateMiyagi Nairiku earthquake in 2008,
the occurrence mechanism of large accelerations is investigated by the non-stationary
Fourier spectra and the double integrated displacement profile. From the difference in the
occurrence time, the frequency, and the amplitude of dominant components between
the surface and the borehole, it is clear that the amplification in the surface soil caused
the large acceleration at the surface. In the deformation process of the surface soil,
the whipping mode is picked up at AKTH04. The relative displacement profile between
the surface and the borehole is analyzed by the phase curves. The orbit at AKTH04 is
identified by the sinusoidal waves of the sway, updown, and rocking modes. It can be
concluded that the large accelerations at AKTH04 may be due to an irregular local mode
caused by the collision between the base-mat and surrounding soil. From the orbit at
IWTH25, the vertical pulse can be regarded as an induced vertical motion due to the
collision between the base-mat of an observation house and the soil beneath it.
Keywords: KiK-net, pulse wave, non-stationary Fourier spectra, double integrated displacement profile, relative
displacement profile, amplification in surface soil, induced vertical motion, collision between base-mat and
surrounding soil
INTRODUCTION
After the Hyogoken-Nanbu earthquake in 1995 (moment magnitude Mw = 6.9), the importance
of seismic records was re-recognized for mitigation of seismic hazard, and the Strong-motion
Seismograph Networks (K-NET, KiK-net) were established by the National Research Institute for
Earth Science Disaster Prevention (NIED, 2016) in 1996. The number of seismic events with Mw
6.0 or larger measured by K-NET and KiK-net and the maximum magnitude from 1996 to 2017
(January) are illustrated in Figure1.
Forty-eight seismic records (as of January 2017) with the peak acceleration larger than the accel-
eration of gravity alerted us to the reevaluation of the seismic design force for nuclear facilities.
The Nuclear Safety Commission of Japan (NSC) started the reevaluation of the Nuclear Reactor
Regulations in June 2001. The specialist working group for the Nuclear Reactor Regulation reported
the revised version in May 2006, and the NSC revised the Regulation in September 2006 (The World
Nuclear Association, 2016). The electric companies were requested to certify the soundness of the
nuclear power plants for the revised regulation as the back-check. During the back-check process,
1 February 2017|Volume 3|Article 13
Kamagata and Takewaki
100
Mar. 11 2011 Mw9.0
80
Occurrence time
60
57
Maximum Magnitude
40 Occurrence times
more than Mw6.0
20
0 0
2000 2005 2010
Year
FIGURE 1 | Seismic events recorded by K-NET and KiK-net (as of
January, 2017).
three large seismic events attacked the nuclear power plants
every 2 years, i.e., the Niigata-ken Chuetsu-oki earthquake in
2007 (MW 6.6), the Suruga Bay earthquake in 2009 (MW 6.3), and
the 2011 off the Pacific coast of Tohoku earthquake (MW 9.0).
The 2011 earthquake forced the upheaval of the Japanese nuclear
policy. The Nuclear Regulation Authority was newly established
in place of the Nuclear and Industry Safety Agency in 2012. The
electric companies were forced to reevaluate the soundness of
the nuclear power plants as the back-fit, which is against the
non-retroactivity. The authors have investigated the occurrence
mechanism of large accelerations, in which the non-linear inter-
action between a building and the surrounding soil was detected
(Kamagata and Takewaki, 2013a,b; Kojima etal., 2014).
In the guideline of Japan Nuclear Energy Safety Organization
(Kobayashi et al., 2014), 16 seismic events are listed up as the
seismic events by hidden faults, 1 of which is the IwateMiyagi
Nairiku earthquake in 2008 (MW 6.9). The seismic records with
large accelerations, i.e., AKTH04 (EW) with 24.5 m/s2 and
IWTH25 (UD) with 38.7 m/s2, were evaluated as the basis of
seismic design motions with the peak value 20m/s2 or larger.
Regarding the seismic records at IWTH25, many researchers
investigated the occurrence mechanism of large accelerations
from various viewpoints. Mogi etal. (2013) focused on the non-
linearity of surface soil and evaluated the change of the shear
wave velocity by the Normalized Input-Output Minimization
analysis. Yamada etal. (2010) also focused on the non-linearity
of surface soil and evaluated the deterioration ratio of the surface
soil by the deconvolution method. Aoi etal. (2010) proposed the
trampoline effect of surface soil. Tobita etal. (2010) supported
the trampoline effect by the FEM analysis with the non-linear
multiple simple shear mechanism. Yamada et al. (2009) also
supported the trampoline effect by introducing the mechanism
of slapdown and evaluated the uplifted amplitude of the near
surface layer. Kinoshita (2011) also focused on the near surface
layer and explained it as the bouncing-ball effect. All the above
studies focus on the seismic behavior of surface soil from the
geo-mechanical research field.
On the other hand, Ohmachi etal. (2011) introduced a differ-
ent occurrence mechanism from the viewpoint of engineering.
They insisted that the collision between the base-mat of the
Frontiers in Built Environment | www.frontiersin.org
Occurrence Mechanism of Large Acceleration
observation house and the soil beneath it caused the pulse-like
10 wave with large acceleration. This may be an acceptable assump-
tion from the engineering viewpoint. The collision between the
8 base-mat and the below soil has been discussed as an induced
Maximum Mw
vertical motion by the uplift of the base-mat in the seismic
6
analysis of the nuclear power plant.
4 While Ohmachi et al. (2011) provided a hypothesis on the
occurrence mechanism of large amplitudes of ground motions
2 and investigated the validity of such hypothesis, the authors
investigate the occurrence mechanisms by introducing the fol-
2015 lowing unique approaches: (i) the non-stationary Fourier spectra
for detecting the frequency components contributing to the
pulse wave, (ii) the maximum amplitude spectra for detecting
the amplitude values of dominant components, (iii) the double
integrated displacement profile for identifying the displacement
profile corresponding to the acceleration pulse, (iv) the deforma-
tion process of surface soil for detecting the seismic behavior, such
as the whipping mode and uniform layered mode, (v) the relative
displacement profile between the surface and the borehole for
suggesting the deterioration state as the average shear strain, and
(vi) the use of phase curve for identifying the sway, the updown,
and the rocking modes. From above analytical results, the authors
attempt to investigate the occurrence mechanism of pulse wave
at AKTH04 (EW) as the collision between the observation house
and the side soil. Regarding the upward pulse at IWTH25 (UD),
the collision between the observation house and the below soil is
considered as the plausible occurrence mechanism of observed
large acceleration.
SEISMIC EVENT OF IWATEMIYAGI
NAIRIKU EARTHQUAKE IN 2008
The site map of AKTH04, IWTH25, and epicenter of the Iwate
Miyagi Nairiku earthquake are illustrated in Figure 2 together
with the fault plane and the GPS-displacement (GSI, 2016). The
maximum horizontal displacement of 1.533m was measured at
the GPS-based control point of Kurikoma-2. This is five times
larger than those at the nearby GPS-based control points. The
epicentral distance of IWTH25 is 3km and that of AKTH04 is
22 km. The seismic event of the IwateMiyagi Nairiku earth-
quake in 2008 was reported by The Headquarters for Earthquake
Research Promotion (2016).
Regarding horizontal components at ground surface, the peak
values at AKTH04 are larger than those at IWTH25. On the other
hand, in vertical components at ground surface, the peak value
at IWTH25 is larger than that at AKTH04. Since the epicentral
distance of AKTH04 is greater than that of IWTH25, the large
acceleration at AKTH04 may be caused by some amplification
mechanisms in the propagation process from the borehole to the
ground surface. At the borehole, the peak values at IWTH25 are
larger than those at AKTH04. This is considered to be mainly
related to the shorter epicentral distance. The amplification ratios
at IWTH25 are smaller than those at AKTH04. This may be
related to the amplification in the propagation process including
the interaction between the observation house and the surround-
ing soil.
2 February 2017|Volume 3|Article 13
Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

FIGURE 2 | Seismic event of the IwateMiyagi Nairiku earthquake in 2008.

ANALYSIS OF DOMINANT COMPONENT Non-Stationary Fourier Spectra of AKTH04

PROPERTY BY NON-STATIONARY In order to compare the property of dominant components
between the surface and the borehole, the non-stationary Fourier
FOURIER SPECTRA
spectra and the maximum amplitude spectra are illustrated with
The property of distribution of dominant components is analyzed acceleration profiles in Figure 3. The peak accelerations at the
by the non-stationary Fourier spectra defined by Eq.1: surface and the borehole are listed with the amplification ratio in
Table1. In the acceleration profiles of AKTH04 (EW), the peak
( ) ( )

F i , t j = g , t j exp ii d , (1) acceleration 24.5 m/s2 occurred as the single downward pulse

wave.
where
The property of dominant components was analyzed as
( )
g , t j = y ( ) ; t j t a 2 t j + t a 2 , follows:

g ( , t ) = 0 ; < t
j j t a 2 , t j + t a 2 < , (1) The peak components at the borehole arrived earlier than
y () ; seismic acceleration record, the peak components at the surface.
(2) The dominant components at the surface occurred in a nar-
tj ( j=1, 2,,M); j-th time, i (i=1, 2, , N/2); i-th frequency,
row frequency range, and those at the borehole occurred in
N; number of steps in FFT analysis,
a wide frequency range.
M=T/T; number of steps in sweep process,
(3) The maximum amplitudes of the peak components at the
T; duration time of seismic record,
surface are larger than those at the borehole.
T; sweep interval time.
ta; width of data window (s).
The analytical results suggest that the dominant components
The influence of data window width is shown in Supplementary at the surface were amplified in the propagation process from the
Material. borehole to the surface.
The analytical result can be illustrated by the contours with the
horizontal axis of time and the vertical axis of frequency. The con- Non-Stationary Fourier Spectra
tour level is normalized by the maximum value and is displayed in at IWTH25
percentage. The maximum amplitude spectra are introduced by Consider next the NS, EW, and UD components at IWTH25.
Eq.2 to evaluate the amplitudes of dominant component. Also, The acceleration profiles, the non-stationary Fourier spectra,
the peak amplitude (VPA) is defined by Eq.3. and the maximum amplitude spectra are compared in Figure4
A(i ) = max{F (i ; t j ); t j = t1 ,..., t M }, (2) between the surface and the borehole. In the acceleration profiles
of IWTH25 (UD), the peak acceleration 38.7m/s2 occurred as the
VPA = max {A ( i ) ; i = 1,, N 2)}. (3) maximum pulses of several upward pulses.

Frontiers in Built Environment | www.frontiersin.org 3 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

15 30 15
Acceleration (m/s2)

Acceleration (m/s2)
AKTH04 (NS) surface AKTH04 (EW) surface

Acceleration (m/s 2)
10 20 AKTH04 (UD) surface
borehole borehole 10
borehole
5 10 5
0 0 0
-5 -10 -5
-10 -20 -10
-15 -30 -15
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Time (s) Time (s) Time (s)
20 20 20
AKTH04NS surface AKTH04EW surface AKTH04UD Surface
Data-windw width=0.4s Data-windw width=0.4s Data-windw width=0.4s 100
FFT time=5.12s FFT time=5.12s FFT time=5.12s
15 15 15 90
80

Frequency (Hz)

Frequency (Hz)
Frequency (Hz)

70
60
10 10 10 50
40
30
5 5 5 20
10
0
0 0 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Time (s) Time (s) Time (s)
12.2s surface 11.6s 11.6s
8.6s 8.6s 10.0s
20
3.6s 3.0s 1.6s
AKTH04NS borehole 20 20
AKTH04EW borehole AKTH04UD Borehole
Data-windw width=0.4s Data-windw width=0.4s Data-windw width=0.4s
FFT time=5.12s 100
FFT time=5.12s FFT time=5.12s 90
15 15 15
Frequency (Hz)

80
Frequency (Hz)

Frequency (Hz)
70
60
10 10 10 50
40
30
5 5 5 20
10
0
0 0 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Time (s) Time (s) Time (s)
borehole
6 6 6
Maximum amplitude (m/s2)

Maximum amplitude (m/s2)

Maximum amplitude (m/s2)

AKTH04 (NS) surface AKTH04 (EW) surface AKTH04 (UD) surface
borehole borehole borehole
4 4 4

Peak 3.3Hz; 4.4m/s2 Peak 6.1Hz; 2.4m/s2

2 Peak 3.1Hz; 3.5m/s2 2 2

Peak 1.95Hz; 0.60m/s2 Peak 7.0Hz; 0.68m/s2 Peak 0.2Hz; 0.38m/s2

0 0 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Frequency (Hz) Frequency (Hz) Frequency (Hz)

NS component EW component UD component

FIGURE 3 | Acceleration profiles, non-stationary Fourier spectra, and maximum amplitude spectra of AKTH04.

The property of dominant components was analyzed as TABLE 1 | Peak accelerations and amplification ratio (surface/borehole).
follows: Site Epicentral Peak acceleration Amplification
distance (m/s2) ratio (surface/
(1) The dominant components at the surface in the NS and (km) borehole)

EW components scatter in the frequency range of 10Hz or NS EW UD NS EW UD

smaller with the amplitude values of 2.0m/s2 or larger.
(2) The dominant components at the surface in the UD compo- AKTH04 Surface 22 13.2 24.5 10.9 5.71 7.18 5.89
Borehole 2.31 3.41 1.85
nent scatter in the wide frequency range of 20Hz or smaller IWTH25 Surface 3 11.4 14.3 38.7 1.10 1.91 5.68
with the amplitude of 3.0m/s2 or larger. Borehole 10.4 7.48 6.81

The analytical results indicate that the large accelerations at the

borehole). The property of dominant components was analyzed
surface in the UD component were caused by the superposition of
as follows:
several higher frequency components from 5 to 17Hz.
The peak and average values of dominant components are (1) At the surface at AKTH04, the peak maximum amplitude
listed in Table 2 with the average amplification ratio (surface/ occurred in the EW component at the frequency of 3.3Hz

Frontiers in Built Environment | www.frontiersin.org 4 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

20 20 40
Surface Surface Surface
Acceleration (m/s2)

Acceleration (m/s2)

Acceleration (m/s2)
IWTH25 (NS) IWTH25 (EW) 30
Borehole Borehole Borehole
10 10 20
10
0 0 0
-10
-10 -10 -20
-30
IWTH25 (UD)
-20 -20 -40
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Time (s) Time (s) Time (s)
20 20 20
IWTH25NS Surface IWTH25EW Surface
Data-windw width=0.4s Data-windw width=0.4s
FFT time=5.12s FFT time=5.12s 100
15 15 15 90
Frequency (Hz)

Frequency (Hz)
80

Frequency (Hz)
70
10 60
10 10 50
40
30
5 5 5 20
IWTH25UD Surface 10
Data-windw width=0.4s 0
0 FFT time=5.12s
0 0
0 5 10 15 20 0 5 10 15 20
6.0s Time (s)
5.0s 7.4s
Time (s) 2.4s
0
4.2s
5 10
Time (s)
15 20

3.0s surface
20 3.0s 20 2.4s 20 1.8s
IWTH25NS Borehole IWTH25EW Borehole IWTH25UD Borehole
Data-windw width=0.4s Data-windw width=0.4s Data-windw width=0.4s
FFT time=5.12s FFT time=5.12s 100
FFT time=5.12s 90
15 15 15
80
Frequency (Hz)

Frequency (Hz)
Frequency (Hz)

70
60
10 10 10 50
40
30
5 5 5 20
10
0
0 0 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Time (s) Time (s) Time (s)

borehole
6 6 6
IWTH25 (NS) Surface IWTH25 (EW) Surface IWTH25 (UD) Surface
Maximum amplitude (m/s 2)

Maximum amplitude (m/s 2)

Maximum amplitude (m/s 2)

5 Borehole 5 Data-window width;0.4s Borehole 5 Borehole

Data-window width;0.4s Data-window width;0.4s
4 FFT analytical time; 5.12s
Peak 10.9Hz4.9m/s2
4 FFT analytical time; 5.12s 4 FFT analytical time; 5.12s
3 Peak 2.9Hz2.4m/s2 3 Peak 7.4Hz2.7m/s2 3
2 2 2
1 Peak 0.39Hz1.08m/s2 1 1
Peak 4.5Hz1.26m/s2 Peak 0.39Hz1.0m/s2
0 0 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Frequency Frequency Frequency

NS component EW component UD component

FIGURE 4 | Acceleration profiles, non-stationary Fourier spectra, and maximum amplitude spectra of IWTH25.

with the amplitude 4.4 m/s2, and the average value is with the amplitude of 1.26 m/s2, and the average value is
1.65m/s2. 0.88m/s2.
(2) At the surface at IWTH25, the peak maximum amplitude
occurred in the UD component at the frequency of 10.9Hz The VPA and the average VPA in the borehole at IWTH25 are
with the amplitude of 4.89 m/s2, and the average value is almost two times larger than those at AKTH04. This may be
2.80m/s2. related to the difference of epicentral distance, i.e., IWTH25
( 3) At the borehole at AKTH04, the peak maximum amplitude (3km) and AKTH04 (22km). The dominant components in the
occurred in the EW component at the frequency of 7.0Hz borehole are considered to be related to the radiated components
with the amplitude of 0.68 m/s2, and the average value is at the seismic asperity.
0.38m/s2. On the other hand, the peak value in the horizontal compo-
(4) At the borehole at IWTH25, the peak maximum amplitude nents at the surface at AKTH04 is larger than those at IWTH25.
occurred in the EW component at the frequency of 4.5Hz This may be related to the amplification property of the surface

Frontiers in Built Environment | www.frontiersin.org 5 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

TABLE 2 | Amplitude of dominant components. Comparing the time interval from peak-to-peak, the time
Site VPA (m/s2) Average VPA (m/ Amplification interval at AKTH04 is shorter than that at IWTH25. This sug-
s2) ratio; average gests that the frequency in the surface soil at AKTH04 is higher
VPA (surface/ than that at IWTH25. As for the amplification from the borehole
borehole) to the surface, the amplification ratio at AKTH04 is larger than
NS EW UD NS EW UD NS EW UD that at IWTH25. Comparing the deformation processes in the
surface soil between AKTH04 and IWTH25, the whipping mode
AKTH04 Surface 3.53 4.39 2.41 1.06 1.65 1.24 3.7 4.3 6.5 is predominant in the surface soil at AKTH04. On the other hand,
Borehole 0.60 0.68 0.38 0.29 0.38 0.19
the uniform layered mode is predominant in the surface soil at
IWTH25 Surface 2.38 2.65 4.89 1.41 1.53 2.80 1.8 1.7 3.8 IWTH25. These properties are consistent with the property of
Borehole 1.13 1.26 1.03 0.80 0.88 0.74
VS profile.
In order to evaluate the seismic behavior of the surface soil, the
authors introduce the relative displacement between the surface
soil. The peak value in the UD component at the surface at and the borehole. Using the relative displacement, the shear
IWTH25 is significantly amplified. strain profile can be calculated by dividing the relative displace-
ment by the distance from the borehole to the surface. Assuming
ANALYSIS OF DOUBLE INTEGRATED the soil condition as the gravel soil, the maximum shear strain
at AKTH04 and IWTH25 are plotted in G relation (see
DISPLACEMENT PROFILES
Figure6). The soil state during the seismic event is estimated to
In order to identify the occurrence mechanism of pulse waves, the be in a deteriorated state (the ratio of stiffness is 0.6 or smaller).
double integrated displacement profile has been analyzed. Since Many seismic records in KiK-net with the peak acceleration of
the pulse wave is considered to be composed of high frequency 10m/s2 or larger have been analyzed. The deterioration ratio of
components, the band-pass filter, as shown Figure5, is adopted 0.6 or smaller was approximately average. Therefore, it may be
to eliminate the low frequency components. Four kinds of cutoff concluded that the derived deterioration ratios have less influence
range are set in the lower frequency range as shown in Table3. on the large acceleration.
The difference in displacement profiles by the band-pass
filter has been investigated. Three kinds of double integrated Amplification in Surface Soil
displacement profiles at AKTH04 and IWTH25 are illustrated in For the AKTH04 (EW), the distribution property of dominant
Figure5. One is the double integrated displacement profiles of components in the displacement profiles at the surface and the
the original seismic record, and the others are those of Case-1 borehole is compared with that in the relative displacement
and Case-2. Regarding AKTH04 (EW), the profile of Case-1 has profile by using the non-stationary Fourier spectra as shown in
a smaller difference from that of Case-2. For IWTH25 (UD), Figure7.
the profile of Case-1 has an upward drift component and that of The property of dominant components can be evaluated as
Case-2 has no drift component. Therefore, the double integrated follows:
displacement profiles of Case-2 have been adopted to evaluate the
seismic behavior at the surface and the borehole. (1) The dominant components at the borehole occur at the time
of 8.6s. This is earlier than the occurrence time of 10.2s at
Seismic Behavior of Surface Soil the surface.
The displacement profiles (Case-2) at AKTH04 (EW) and (2) The frequency of peak dominant component at the borehole
IWTH25 (EW) are illustrated in Figure6. The shaded parts are is 1.1Hz. This is lower than the frequency of 1.9Hz at the
illustrated as the deformation process of the surface soil with the surface.
interval time of 0.05s (AKTH04) and 0.1s (IWTH25). The maxi- (3) The peak maximum amplitude at the borehole is 0.014m.
mum peak-to-peak displacement at AKTH04 (EW) occurs in the This is smaller than that of 0.029m at the surface.
interval from 9.85 to 10.05s. The amplitude is 0.077m at the sur- (4) The distribution property of dominant components in the
face and 0.0128m at the borehole, and the amplification ratio is relative displacement profile is almost similar to that of the
6.0. The maximum peak-to-peak displacement at IWTH25 (EW) displacement profile at the surface.
occurs in the interval from 4.6 to 5.1s. The amplitude is 0.10m (5) The peak component at the relative displacement is detected
at the surface and 0.031m at the borehole, and the amplification at the frequency of 2.4Hz with the amplitude of 0.025m.
ratio is 3.3. This is 86% of the peak value at the surface.
This vibration property is consistent with the shear wave
velocity (VS) in the surface soil as shown in Figure6. Comparing It may be concluded that the large acceleration at the surface
the VS up to 100m depth, the VS at AKTH04 is larger than that is related to the local mode in the propagation process from the
at IWTHH25. The seismic record at the surface is estimated to borehole to the surface.
be influenced by the top soil. On the other hand, the shear wave For IWTH25 (UD), the dominant components in the accel-
velocity at the borehole is 1,500m/s at AKTH04 and 1,810m/s at eration profile and the relative displacement profile have been
IWTH25. These satisfy the engineering bedrock condition of the investigated in the interval from 3.5 to 4.5s of Case-2, Case-3,
shear wave velocity 700m/s or larger. and Case-4, and those are shown in Figure8.

Frontiers in Built Environment | www.frontiersin.org 6 February 2017|Volume 3|Article 13

Kamagata and Takewaki
FIGURE 5 | Double integrated displacement profiles.
TABLE 3 | List of cutoff range.
fL1 (Hz) fL2 (Hz)
Case-1 0.2 0.5
Case-2 0.5 1.0
Case-3 5.0 6.0
Case-4 7.0 8.0
In the acceleration profile, the pulse wave of Case-2 has an
asymmetric shape (upward), and those of Case-3 and Case-4 are
close to a symmetric shape. In the relative displacement profile
of Case-2, a low frequency component has been observed, and
the peaks are not synchronized with the peaks in the acceleration
profile. In the relative displacement of Case-4, the phase is almost
coincident with the acceleration in the opposite direction.
The distribution property of the dominant components in the
non-stationary Fourier spectra can be evaluated as follows:
(1) In the acceleration profiles, the dominant components scat-
ter in the frequency range higher than 10Hz. This behavior
is the same independent of the applied filter.
(2) The peak amplitude is detected at 11.2Hz with the amplitude
5.2m/s2.
(3) In the relative displacement profiles, the dominant compo-
nents differ in Case-2, Case-3, and Case-4. This is related to
the frequency components smaller than 10Hz.
(4) In the relative displacement profile of Case-2, the dominant
component occurs at 11.1Hz with the amplitude value of
0.0014m. This corresponds to the dominant component of
Frontiers in Built Environment | www.frontiersin.org
Occurrence Mechanism of Large Acceleration
11.2Hz with the amplitude of 5.2m/s2 in the acceleration
profile.
The above analytical results suggest that the acceleration
profile of Case-4 includes the pulse components in the seismic
record. The double integrated displacement profile of Case-4 also
includes the components of pulse wave in the seismic record. From
the relation of VA = 2VD between the acceleration value (VA) and
the displacement value (VD), the peak displacement 0.0014m in
the successive pulses with the frequency of 11.1Hz corresponds
to the acceleration 6.8m/s2. This is 18% of the peak amplitude
38.7m/s2. It may be concluded that pulse wave is composed of
various frequency components.
Considering the layout of the observation house (see
Figure15), the vertical displacement dv of 0.0014m corresponds
to the rotation angle 2.8/1,000 (=dv/w; w=0.5m; w: the length
from the pit to the edge at the long side) or 1/1,000 (w=1.3m; the
length from the pit to the edge at the short side). The estimated
small inclination of the observation house would not be observed
after the earthquake.
OCCURRENCE MECHANISM OF LARGE
ACCELERATION VALUES
Consider the occurrence mechanism of large acceleration in
the seismic record of IWTH25 (UD). Many researchers have
investigated the cause of upward pulse waves. They assumed an
original occurrence mechanism from the upward pulse shape.
The authors analyzed the seismic records and produced additional
7 February 2017|Volume 3|Article 13
Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

0.06 0.06
AKTH04 (EW) surface surface
Displacement (m)

0.04 0.04

Displacement(m)
Case-2 borehole borehole
0.02 0.02
0 0
-0.02 -0.02
-0.04 -0.04
-0.06 -0.06 IWTH25 (EW) Case-2
0 5 10 15 20 0 5 10 15 20 0
Time (s) Time (s)

AKTH04 EW) IWTH25 (EW)

0.072m 0.077m 0.063m -100

Depth (m)
0 AKTH04 (EW) 0 AKTH04 (EW) 0 AKTH04 (EW)
9.65s 9.85s 10.05s
9.70s 9.90s 10.10s
9.75s 9.95s 10.15s -200
9.80s 10.0s 10.20s
Depth (m)

Depth (m)

Depth (m)
-50 9.85s -50 10.05s -50 10.25s
10.30s AKTH04

Amp. Ratio; 31.3 Amp. Ratio; 6.0 Amp. Ratio; 2.5 IWTH25
-300
0 500 1000 1500 2000
0.0023m 0.0128m 0.0250m Velocity of Shear wave (m/s)
-100 -100 -100
-0.06 -0.03 0 0.03 0.06 -0.06 -0.03 0 0.03 0.06 -0.06 -0.03 0 0.03 0.06
Deformation (m) Deformation (m) Deformation (m)
Profile of VS
AKTH04 (EW) from 9.65s to 10.3s 1

0.035m 0.054m 0.100m 0.8

0 IWTH25 (EW) 0 IWTH25 (EW) 0 IWTH25 (EW) Alluvial clay
3.5s 4.1s 4.6s Diluvial clay
0.6
3.6s 4.2s Silt

G/Go
4.7s
-100 3.7s -100 4.3s
4.4s -100 4.8s Alluvial sand
Depth (m)

3.8s
Depth (m)

4.9s 0.4
Depth (m)

3.9s 4.5s Diluvial sand

5.0s
4.0s 4.6s 5.1s 0.2
Gravelly soil
-200 -200
4.1s -200 AKTH04
Amp. Ratio; 1.7 Amp. Ratio; 2.6 Amp. Ratio; 3.3 IWTH25
0
0.022m
-7 -6 -5 -4 -3 -2 -1
-300 0.022m -300 -300 0.031m 10 10 10 10 10 10 10
-0.06 -0.03 0 0.03 0.06 -0.06 -0.03 0 0.03 0.06 -0.06 -0.03 0 0.03 0.06 Shear strain
Displacement (m) Displacement (m) Displacement (m)

IWTH25 (EW) from 3.5s to 5.1s Deterioration ration of surface soil

FIGURE 6 | Deformation process of surface soil of AKTH04 (EW) and IWTH25 (EW).

information from the seismic records. During the IwateMiyagi Both pulse waves in the acceleration profiles in AKTH04 (EW)
Nairiku earthquake, the large acceleration of 24.5 m/s2 was and IWTH25 (UD) synchronize with the pulse waves in the
recorded at AKTH04 (EW). The epicentral distance of AKTH04 relative displacement profiles. This suggests that the acceleration
is 22km. This is longer than that of IWTH25 (2km). The differ- 24.5m/s2 at AKTH04 (EW) is related to the displacement with the
ence of AKTH04 and IWTH25 is due to the direction, i.e., the amplitude of 0.03m, and the acceleration 38.7m/s2 at IWTH25
horizontal component at AKTH04 and the vertical component (UD) is related to the displacement with the amplitude of 0.02m
at IWTH25. Apparently, the previous research on IWTH25 (UD) or less.
cannot explain the occurrence mechanism of AKTH04 (EW). In
this section, the occurrence mechanisms of large acceleration in Seismic Records at AKTH04
the horizontal component and the vertical component will be The acceleration and the relative displacement of EW and UD
investigated. profiles are illustrated in Figure10. The occurrence point of the
Let us investigate the pulse waves that caused the peak peak acceleration is marked by the red-star in the profile and
acceleration, i.e., from 11 to 12s at AKTH04 (EW) and from 3.5 the phase curve. In the profiles of Case-2, the peak acceleration
to 4.5s at IWTH25 (UD). They are compared with the relative occurs in the process toward the positive reversal point in the
displacement profile in Figure9. relative displacement profile. In the profiles of Case-4, the peak
At AKTH04 (EW), the peak acceleration of 24.5m/s2 occurred acceleration occurs at the positive reversal point in relative dis-
as the single downward pulse wave at 11.57s. This corresponds placement profile.
to the positive terminal point of cyclic movement in the relative Using the EW and the UD relative displacement profiles, the
displacement profile of Case-2 and Case-3. At IWTH25 (UD), the phase curve for the station AKTH04 is illustrated in Figure12.
multiple upward pulse waves in the acceleration profile occurred The orbit of 8-shape was detected in the phase curve of Case-2.
from 3.5 to 4.5s. This corresponds to the negative reversal points This was identified as the coupling of sway and rocking modes. In
of Case-2 and Case-4 in relative displacement profiles. the phase curve of Case-3, the elliptical orbit was detected. This

Frontiers in Built Environment | www.frontiersin.org 8 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

0.06

Relative displacement(m)
0.06
surface
0.04

Displacement(m)
borehole 0.04
0.02 0.02
0 0
-0.02 -0.02
-0.04 -0.04
-0.06 AKTH04 (EW) Case-2 -0.06 AKTH04 (EW) Case-2
0 5 10 15 20 0 5 10 15 20
Time (s) Time (s)

5 5
AKTH04EW(surface) AKTH04EW
Case-2; Displacement Case-2; Relative displacement
Data-windw width=1.0s 100
4 4 Data-windw width=1.0s 90
FFT time=20.48s FFT time=20.48s
80
Frequency (Hz)

Frequency (Hz)
3 3 70
60
50
2 2 40
30
20
1 1 10
0
0 0
10.0s 11.6s
0 5 1010.2s 15 20 0 5 10 15 20
Time (s) 1.6s
Time (s)
Displacement at surface Relative displacement
8.6s 1.6s
5
AKTH04EW(borehole)
Case-2; Displacement 0.04
4 Data-windw width=1.0s AKTH04 (EW) Case-2 Displacement at surface
FFT time=20.48s
Maximum amplitude (m)

Data window width=1.0s Displacement at borehole

Frequency (Hz)

0.03 1.9Hz; 0.029m Relative displacement

3

0.02
2 2.4Hz; 0.025m
1 0.01
1.1Hz; 0.014m
0 0
0
8.6s 10 10.6s15
5 20 0 1 2 3 4 5
2.0s
Time (s) Frequency (Hz)
Displacement at borehole Maximum amplitude spectra

FIGURE 7 | Displacement and relative displacement profiles of AKTH04 (EW).

was identified as the coupling of the same frequency mode with the frequency of fR. The relation of fR=2fV was derived
the phase delay between the EW and UD components. from the non-stationary Fourier spectra shown in Figure11.
The dominant components in the relative displacement profile The amplitudes of horizontal and vertical movements were
in the EW and UD components were analyzed by the non-sta- determined from the relation of the relative displacement
tionary Fourier spectra and are shown in Figure11. The coupling profiles of EW and UD components.
modes of 2.8 and 6.2Hz were detected in the UD component, and The sway mode in the horizontal displacement:
the single mode of 3.1Hz was detected in the EW component.
DH = sin ( f H t ) ; f H = 3 Hz. (4)
The occurrence process of the pulse wave can be explained as
follows: The rocking mode in the vertical displacement:

(1) In the phase curve of Case-2 in Figure10, an 8-shape orbit DR = 0.5 sin ( f R t ) ; f R = 6 Hz. (5)
was detected from 11.0 to 11.4 s. This was identified by
the sway mode of sinusoidal wave with the frequency of The updown mode and rocking mode in vertical displacement:
fH (3Hz) and the rocking mode of a sinusoidal wave with DV = 0.3 sin ( f V t ) + 0.3 sin ( f R t ) ; f V = 3 Hz, f R = 6 Hz. (6)
the frequency of fR (6Hz). The coupling of both sinusoidal
waves was plotted as the symmetric 8-shape orbit. As shown (2) In the phase curve of Case-3 shown in Figure10, an elliptical
in Figure11, the vertical movement was composed of dual orbit was detected from 11.4 to 11.56s. This was identified
frequency components. This was identified by the updown by the coupling of the horizontal and the vertical modes of
mode with the frequency of fV and the rocking mode with the same frequency with the phase delay.

Frontiers in Built Environment | www.frontiersin.org 9 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

40 0.02 40 0.01 40 0.006

Acceleration Surface Acceleration Surface Acceleration Surface

Relative displacement (m)

Relative displacement (m)

Relative displacement (m)

Relative displacement Relative displacement Relative displacement
Acceleration (m/s2)

Acceleration (m/s2)

Acceleration (m/s2)
20 0.01 20 0.005 20 0.003

0 0 0 0 0 0

-20 -0.01 -20 -0.005 -20 -0.003

IWTH25 (UD) Case-2 IWTH25 (UD) Case-3 IWTH25 (UD) Case-4

-40 -0.02 -40 -0.01 -40 -0.006
3. 5 4 4.5 3. 5 4 4.5 3. 5 4 4.5

20 Time (s) 20 Time (s) 20 Time (s)

100
15 15 15 90
80

Frequency (Hz)
Frequency (Hz)

Frequency (Hz)
70
60
10 10 10 50
40
30
5 5 IWTH25UD Surface Case-3 5 IWTH25UD Surface Case-4 20
IWTH25UD Surface Case-2 Data-windw width=0.4s Data-windw width=0.4s 10
Data-windw width=0.4s FFT time=5.12s FFT time=5.12s 0
FFT time=5.12s
0 0 0
3.5 3.7 3.9 4.1 4.3 4.5 3.5 3.7 3.9 4.1 4.3 4.5 3.5 3.7 3.9 4.1 4.3 4.5
Time (s) Time (s) Time (s)
Acceleration profile
20 20 20
IWTH25UD Case-2 IWTH25UD Case-3 IWTH25UD Case-4
Relative displacement Relative displacement Relative displacement
Data-windw width=0.4s Data-windw width=0.4s Data-windw width=0.4s 100
15 FFT time=5.12s 15 FFT time=5.12s 15 FFT time=5.12s 90
Frequency (Hz)

80
Frequency (Hz)

Frequency (Hz)
70
60
10 10 10 50
40
30
5 5 5 20
10
0
0 0 0
3.5 3.7 3.9 4.1 4.3 4.5 3.5 3.7 3.9 4.1 4.3 4.5 3.5 3.7 3.9 4.1 4.3 4.5
Time (s) Time (s) Time (s)
Relative displacement profile
Case-3 Case-4
Case-2
10 0.006
IWTH25 (UD)
Maximum amplitude (m/s 2)

Case-2 IWTH25 (UD)

Maximum amplitude (m)

Case-3 Relative displacement 0.005 Relative displacement

8
Case-4 from 3.5s to 4.5s from 3.5s to 4.5s
0.004 Case-2
6 11.2Hz; 5.2m/s2 Case-3
0.003 Case-4
4
0.002 11.1Hz; 0.0014m
2
0.001

0 0 10.5Hz; 0.001m
0 5 10 15 20 0 5 10 15 20
Frequency (Hz) Frequency (Hz)

Acceleration profiles Relative displacement profiles

FIGURE 8 | Comparison of cutoff frequency range in band-pass filter of IWTH25 (UD).

30 0.06 40 0.02
Relative displacement AKTH04 (EW) Acceleration Surface IWTH25 (UD)
Relative displacement (m)

Relative displacement (m)

Case-2
Acceleration (m/s2)

Acceleration (m/s2)

15 Case-3 0.03 20 0.01

0 0 0 0

-15 -0.03
11.3s -20 Relative displacement
Case-2
-0.01

-30
Acceleration Surface 11.57s -0.06 -40 Case-4 4.09s -0.02
11 11.5 12 3.5 4 4.5
Time (s) Time (s)

AKTH04 (EW) IWTH25 (UD)

FIGURE 9 | Pulse waves in acceleration and relative displacement profiles of AKTH04 (EW) and IWTH25 (UD).

Frontiers in Built Environment | www.frontiersin.org 10 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

30 0.06 20 0.006
AKTH04 (EW) Case-2 Acceleration Surface AKTH04 (EW) Case-3 Acceleration Surface

Relative displacement (m)

Relative displacement (m)

Acceleration (m/s2)

Acceleration (m/s2)
15 0.03 10 0.003

0 0 0 0

-15 Relative displacement -0.03 -10 Relative displacement -0.003

EW EW
-30 UD -0.06 -20 UD -0.006
11 11.2 11.4 11.6 11.8 12 11 11.2 11.4 11.6 11.8 12
Time (s) Time (s)
0.015 0.004
AKTH04 Case-2 AKTH04 Case-3 11.40s - 11.56s
Relative displacement (UD) (m)

Relative displacement (UD) (m)

11.56s - 11.63s
0.01
0.002
0.005

0 0

-0.005
-0.002
-0.01
11.0s - 11.4s
11.4s - 11.8s
-0.015 -0.004
-0.04 -0.02 0 0.02 0.04 -0.004 -0.002 0 0.002 0.004
Relative displacement (EW) (m) Relative displacement (EW) (m)

Case-2 Case-3

FIGURE 10 | Phase curves correspondent to the pulse wave of AKTH04.

The sway mode in horizontal displacement: Seismic Records at IWTH25

DH = sin ( f H t ) ; f H = 6 Hz. (7)
The updown mode in vertical displacement:
DR = 0.5 sin ( f V t
The identified displacement profiles and the phase curves are
illustrated in Figure12.
The occurrence mechanism of the pulse wave can be explained
as follows (see Figure12):
(1) Coupling of the sway mode to the left direction and the
upward mode with the clockwise rocking mode.
(2) Coupling of the sway mode to the right direction and the
downward mode with the counter-clockwise rocking mode.
(3) Continuous movement from the stage (2) of the coupling
of the sway mode to the right direction and the downward
mode with the counter-clockwise rocking mode.
(4) Coupling of the sway mode to the left direction and the
upward mode with the clockwise rocking mode.
The pulse wave at 11.57s occurs at the terminal point of the
process (3). Some irregular movement like the collision between
the base-mat of the observation house and the surrounding soil
can be estimated as the occurrence mechanism.
The relative displacement profile at IWTH25 (UD) is illustrated
together with the acceleration profile in the entire profile and the
zoomed profile from 3.4 to 4.4s (hatched in the entire profile) (see
Figure13). In the zoomed profile, the upward pulse waves of the
2 ) ; f V = 6 Hz. (8)
acceleration profile numbering from P1 to P4 are focused upon.
They occurred simultaneously at the downward reversal points in
the relative displacement profile as shown in the zoomed profile.
Furthermore, the authors focused on the additional upward pulse
waves numbering from Q1 to Q3, which occurred in each interval
of numbered pulses from P1 to P4, and they occurred at the inflec-
tion point of the relative displacement profile.
The relative displacement profiles are illustrated as the phase
curves in two kinds of vertical planes with the axes of the NS
and UD components and the EW and UD components (see
Figure 13). The seismic behavior of the observation house is
summarized as follows:
(1) In the phase curve of the EW and UD components, the
reversal point PA-EW is the positive maximum amplitude and
the reversal point PB-EW is the negative maximum amplitude
in the relative displacement profile of the EW component.
(2) The reversal points P2 and P3 occur simultaneously with the
reversal points PA-EW and PB-EW in the relative displacement
profile of the EW component.

Frontiers in Built Environment | www.frontiersin.org 11 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

Relative displacement(m)

Relative displacement(m)
0.06 0.02
AKTH04 (EW) Case-2 AKTH04 (UD) Case-2
0.03 0.01

0 0

-0.03 -0.01

-0.06 -0.02
11 11.2 11.4 11.6 11.8 12 11 11.2 11.4 11.6 11.8 12
Time (s) Time (s)
10 10
AKTH04(EW) C-2 AKTH04(UD) C-2
Relative displacement Relative displacement 100
8 Data-windw width=0.4s 8 Data-windw width=0.4s 90
FFT time=10.24s FFT time=10.24s 80
Frequency (Hz)

Frequency (Hz)
6 6 70
6.2Hz 60
50
4 4 40
3.1Hz 30
20
2 2 2.8Hz 10
0
0 0
11 11.2 11.4 11.6 11.8 12 11 11.2 11.4 11.6 11.8 12
Time (s) Time (s)

0.02
AKTH04 (EW) Case-2 EW
Maximum amplitude (m)

Relative displacement UD
Data-window width;0.4s
3.1Hz; 0.0128m FFT analytical time; 10.24s
0.01

2.8Hz; 0.0025m
6.2Hz; 0.0019m
0
0 2 4 6 8 10
Frequency

FIGURE 11 | Non-stationary Fourier spectra and maximum amplitude spectra of one of the horizontal (EW) and vertical components of AKTH04.

(3) In the phase curve of the NS and UD components, the (3) The components composing the pulse wave of 38.7m/s2 are
reversal point PA-NS is the negative maximum amplitude and analyzed by the amplitude spectra at the time of 4.1s. Three
the reversal point PB-NS is the positive maximum amplitude main components are 5.5Hz (1.9m/s2), 10.9Hz (5.3m/s2),
in the relative displacement profile of the NS component. and 16.8Hz (1.8m/s2).
(4) The reversal points P1 and P4 occur simultaneously with the (4) The dominant component of 10.7Hz (1.6m/s2) is detected
reversal points PA-NS and PB-NS in the relative displacement in the amplitude spectra of the EW component.
profile of the NS component.
(5) The typical additional pulses Q1, Q2, and Q3 occur at the It was concluded that the analytical findings result from the
interval of P1, P2, P3, and P4. non-linear interaction between the observation house and the
surrounding soil (see Figure15). In more detail, the following
The property of the pulse waves in NS, EW, and UD compo- points were obtained.
nents were analyzed by the non-stationary Fourier spectra, the
maximum amplitude spectra, and the amplitude spectra at 4.1s (1) The upward pulse wave was caused by the collision

as shown in Figure14. between the base-mat of the observation house and the
The dominant components are summarized as follows: soil beneath it.
(2) The additional upward pulse wave was caused by the colli-
(1) The amplitude of dominant components in the UD compo- sion between the base-mat of observation house and the soil
nents exceeds 2.0m/s2 in the frequency range from 3.5 to beneath it.
18Hz. (3) The rocking mode can be estimated numerically as the
(2) The peak amplitude occurred at the frequency of 10.9 Hz inclination of the base-mat from 2.2/1,000 to 0.85/1,000.
with the amplitude of 5.3m/s2 in the UD component. The peak displacement corresponding to 10.9Hz (5.3m/s2)

Frontiers in Built Environment | www.frontiersin.org 12 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

1.5 1.5 1.5

Rocking mode 3Hz Sway mode EW
1
Displacement(m)

Sway mode 6Hz 1 1

Displacement(m)

Displacement(m)
Updown and Rocking mode UD
0.5 0.5 0.5
0 0 0
-0.5 -0.5 -0.5
-1 -1 -1
-1.5 Sway and rocking mode Sway, updown and rocking mode Sway and updown mode
-1.5 -1.5
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Time (s) Time (s) Time (s)

1.5 1.5 1.5

8-shape Orbit-I 8-shape Orbit-II elliptical Orbit

1 1 1
Vertical movement

Vertical movement

Vertical movement
0.5 0.5 0.5

0 0 0

-0.5 -0.5 -0.5

-1 -1 -1

-1.5 -1.5 -1.5

-1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5
Horizontal movement Horizontal movement Horizontal movement

Sway mode(3Hz) and Sway mode (3Hz) and Sway mode (6Hz) and
rocking mode (6Hz) Updown mode(3Hz), rocking mode (6Hz) Updown mode (6Hz)

(2) Sway and (3) Sway and

Downward and rocking mode Upward by rocking mode

(1) Sway and (4) Sway and

Upward and rocking mode Downward by rocking mode

FIGURE 12 | Identified displacement profiles and phase curves of AKTH04.

is 0.0011m. The inclination can be derived by dividing the From the double integrated displacement profile, useful
vertical displacement by the length from the edge (0.5 and information on the seismic behavior was derived. First, the
1.3m). frequency components that do not cause the large accelera-
tions were eliminated by the band-pass filter. The deformation
CONCLUSION process in the surface soil was illustrated using the displace-
ment at the surface and the borehole. The whipping mode
During the IwateMiyagi Nairiku earthquake in 2008, large at AKTH04 and the uniform layered mode at IWTH25 were
accelerations were measured at the KiK-net stations, i.e., 38.9m/ detected. These phenomena are consistent with the profile
s2 in IWTH25 (UD) and 24.5m/s2 in AKTH04 (EW). The occur- of the shear wave velocity. Second, the relative displacement
rence mechanisms of such large acceleration have been explained between the surface and the borehole was introduced, and
by using several analytical results, such as the non-stationary it was made clear that the relative displacement includes
Fourier spectra and the double integrated displacement profiles. the component of large acceleration. Furthermore, the shear
The frequency components contributing to the pulse wave strain profile was obtained by dividing the relative displace-
have been obtained from the non-stationary Fourier spectra. The ment by the interval distance between the surface and the
difference in the frequency and the occurrence time of the maxi- borehole. The deterioration state of the surface soil was esti-
mum amplitude between the surface and the borehole indicates mated by plotting the maximum shear strain on the G
that the large acceleration was excited in the propagation process relation.
from the borehole to the surface. The frequency components The pulse wave in the relative displacement profile was
related to the large acceleration have been identified at the same detected. This corresponds to the pulse wave in the acceleration
time. profile. The phase curve was drawn by the relative displacement

Frontiers in Built Environment | www.frontiersin.org 13 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

40 0.03
IWTH25 (UD)

Relative displacement (m)

Acceleration (m/s2)
20 0.015

0 0

-20 -0.015
Acceleration
Relative displacement (Case-2)
-40 -0.03
0 5 10 15 20
Time (s)

Entire profile
40 0.03 0.04
IWTH25 (UD) P3

Relative displacement (m)

IWTH25
Acceleration (m/s2)

Relative displacement (m)

P2
20 P1 0.015 PA-EW
Q2 P4 0.02 PB-NS
Q1 Q3
0 0 0

P1 PA-NS
-20
Acceleration P4 -0.015 -0.02 NS P1 P4
P2 P EW
Relative displacement (Case-2) 3 UD
P2 P3 PB-EW
-40 -0.03 -0.04
3.4 3.6 3.8 4 4.2 4.4 3.4 3.6 3.8 4 4.2 4.4
Time (s) Time (s)
Reversal point
Inflection point Zoomed profile

0.02 0.02
Mainshock IWTH25 Mainshock IWTH25 3.4-3.6s
Relative displacement-UD (m)

0.015
Relative displacement-UD (m)

0.015 3.6-3.9s
3.9-4.2s
0.01 0.01 4.2-4.4s

0.005 end start 0.005 start end

0 0

-0.005 -0.005
PA-NS
P1 PB-NS P4 P1
-0.01 3.4-3.6s P4 -0.01 PB-EW PA-EW
3.6-3.9s
-0.015 3.9-4.2s -0.015
4.2-4.4s P2 P3 P2
-0.02 P3 -0.02
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04

Relative displacement-NS (m) Relative displacement-EW(m)

Vertical plane of NS and UD component Vertical plane of EW and UD component

FIGURE 13 | Comparison of acceleration and relative displacement profiles of IWTH25.

of the UD and EW components, in which an 8-shape orbit was Regarding the large acceleration at IWTH25 (UD), the syn-
detected. This was identified by the sway, the updown, and the chronized pulse waves were detected between the acceleration
rocking modes, and they were modeled by the sinusoidal waves. and the relative displacement profiles. Furthermore, the syn-
The pulse wave at AKTH04 (EW) occurred at the typical point chronized reversal points were found among the NS, the EW, and
in the phase curve. By using these results, it was concluded the UD components. From the singular orbit in the movement
that the large accelerations were caused by a singular mode, of updown direction, the occurrence mechanism of large accel-
such as the collision between the observation house and the eration was explained as the collision between the observation
side soil. house and the soil beneath it. Also, the additional pulse wave was

Frontiers in Built Environment | www.frontiersin.org 14 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

10 10 40
Acceleration (m/s2)

Acceleration (m/s2)
IWTH25 (NS) IWTH25 (EW)

Acceleration (m/s2)
IWTH25 (UD)
5 Surface 5 Surface 20 Surface

0 0 0

-5 -5 -20

-10 -10 -40

3 4 5 3 4 5 3 4 5
Time (s) Time (s) Time (s)
20 20 20
IWTH25NS(surface) IWTH25EW(surface)
Data-windw width=0.4s Data-windw width=0.4s 100
FFT time=5.12s FFT time=5.12s
15 15 15 90
80

Frequency (Hz)
Frequency (Hz)
Frequency (Hz)

70
10.7Hz 10.9Hz 60
10 10 10 50
40
30
5.7Hz 20
5 5 5
IWTH25UD(surface) 10
Data-windw width=0.4s 0
FFT time=5.12s
0 0 0
3 4 4.1s 5 3 4
4.1s 5 3 4 4.1s 5
Time (s) Time (s) Time (s)

NS component EW component UD component

10 10
IWTH25 Surface NS IWTH25 Surface at 4.1s NS

Maximum amplitude (m/s2)

Maximum amplitude (m/s 2)

Acceleration EW Acceleration EW
UD UD
Data-window width;0.4s Data-window width;0.4s
FFT analytical time; 5.12s FFT analytical time; 5.12s
5
10.9Hz; 5.3m/s2 5 10.9Hz; 5.3m/s2
5.7Hz; 3.5m/s2
10.7Hz; 1.6m/s2 5.5Hz; 1.9m/s210.7Hz; 1.6m/s2 16.8Hz; 1.8m/s2

0 0
0 5 10 15 20 0 5 10 15 20

Frequency (Hz) Frequency (Hz)

Maximum amplitude spectra of entire profile Amplitude spectra at 4.1s

FIGURE 14 | Dominant components of NS, EW, and UD components of IWTH25.

North
35 3m 2m

0.36m House base-mat

0.45m 0.4m
2m Seismometer
1. 3m Pit base-mat
Pit-1
0.5m Pit-2 Section of observation house

Layout of observation house

Sway and
Collision between Collision between base-mat and
Rocking
base-mat and side soil soil beneath it
Sway to left direction Sway to right direction

Rocking

Collision between pit-mat and soil beneath it

FIGURE 15 | Layout of observation house and occurrence mechanism by rocking mode of base-mat of IWTH25.

Frontiers in Built Environment | www.frontiersin.org 15 February 2017|Volume 3|Article 13

Kamagata and Takewaki
estimated to be caused by the collision between the pit-mat and
the soil beneath it.
AUTHOR CONTRIBUTIONS
SK carried out the theoretical and numerical analysis. IT dis-
cussed the theoretical and numerical analysis.
REFERENCES
Aoi, S., Kunugi, T., and Fujiwara, H. (2010). Trampoline effect in extreme ground
motion. Science 322, 727730. doi:10.1126/science.1163113
Geospatial Information Authority of Japan (GSI). (2016). GPS Stations Detect
Crustal Deformation Caused by the Iwate-Miyagi Nairiku Earthquake in 2008.
Available at: http://www.gsi.go.jp/ENGLISH/page_e30283.html
Kamagata, S., and Takewaki, I. (2013a). New insights into seismic behavior
of building and surrounding soil at Hamaoka nuclear power station
during Suruga Bay earthquake in 2009. Soil Dyn. Earthquake Eng. 53,
7391. doi:10.1016/j.soildyn.2013.06.007
Kamagata, S., and Takewaki, I. (2013b). Role of records during the 2011 off the
Pacific coast of Tohoku earthquake in seismic resistant design of nuclear power
station. Int. J.Earthquake Eng. Hazard Mitigation 1, 921.
Kinoshita, S. (2011). A stochastic approach for evaluating the nonlinear
dynamics of vertical motion recorded at the IWTH25 site for the 2008 Mw
6.9 Iwate-Miyagi inland earthquake. Bull. Seismol. Soc. Am. 101, 29552966.
doi:10.1785/0120100334
Kobayashi, G., Sugino, H., Tsutsumi, H., and Ebisawa, K. (2014). Guidance for
Evaluation of Seismic Ground Motion from Diffuse Seismicity. JNES-RE-Report
Series, JNES-RE-2013-2045. Tokyo: Japan Nuclear Energy Safety Organization.
Kojima, K., Kamagata, S., and Takewaki, I. (2014). A new interpretation of large
amplitude earthquake acceleration from non-linear local soil-structure inter-
action. Nucl. Eng. Design 273, 271287. doi:10.1016/j.nucengdes.2014.03.023
Mogi, H., Shrestha, S. M., Kawakami, H., and Kawamura, J.(2013). Nonlinear
soil behavior examined on the basis of propagation time observed at the
KiK-net Ichinoseki-Nishi vertical array. Bull. Seismol. Soc. Am. 103, 180195.
doi:10.1785/0120120064
National Research Institute for Earth Science and Disaster Prevention (NIED).
(2016). Strong-Motion Seismograph Networks (K-NET, KiK-net). Available at:
http://www.kyoshin.bosai.go.jp/
Frontiers in Built Environment | www.frontiersin.org
Occurrence Mechanism of Large Acceleration
ACKNOWLEDGMENTS
The authors used the seismic records of K-NET and KiK-net
provided by the National Research Institute for Earth Science
and Disaster Prevention. Part of the present work is based on
the discussion made in the JNES (Japan Nuclear Energy Safety
Organization) committee.
Ohmachi, T., INOUE, S., Mizuno, K., and Yamada, M. (2011). Estimated cause of
extreme acceleration records at the KiK-net IWTH25 station during the 2008
Iwate-Miyagi Nairiku earthquake, Japan. Trans. Jpn. Assoc. Earthquake Eng. 11,
3247.
The Headquarters for Earthquake Research Promotion. (2016). Monthly Reports
on Evaluation of Seismic Activities in Japan; Evaluation of Seismic Activities
for 2008 (Jun 12 2008). Available at: http://www.jishin.go.jp/main/chousa/
08jun_iwate_miyagi2/p07.htm
The World Nuclear Association. (2016). Nuclear Power Plants and Earthquakes.
Available at: http://www.world-nuclear.org/information-library/safety-and-
security/safety-of-plants/nuclear-power-plants-and-earthquakes.aspx
Tobita, T., Iai, S., and Iwata, T. (2010). Numerical analysis of near-field asym-
metric vertical motion. Bull. Seismol. Soc. Am. 100, 14561469. doi:10.1785/
0120090301
Yamada, M., Mori, J., and Heaton, T. (2009). The slapdown phase in high-acceler-
ation records of large earthquakes. Seismol. Res. Lett. 80, 559564. doi:10.1785/
gssrl.80.4.559
Yamada, M., Mori, J., and Ohmi, S. (2010). Temporal changes of subsurface veloci-
ties during strong shaking as seen from seismic interferometry. J.Geophys. Res.
115, B03302. doi:10.1029/2009JB006567
Conflict of Interest Statement: The authors declare that the research was con-
ducted in the absence of any commercial or financial relationships that could be
construed as a potential conflict of interest.
Copyright 2017 Kamagata and Takewaki. This is an open-access article distributed
under the terms of the Creative Commons Attribution License (CC BY). The use,
distribution or reproduction in other forums is permitted, provided the original
author(s) or licensor are credited and that the original publication in this journal
is cited, in accordance with accepted academic practice. No use, distribution or
reproduction is permitted which does not comply with these terms.
16 February 2017|Volume 3|Article 13
Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

APPENDIX ( 2) Three cycles of sinusoidal wave (see FigureA2).

The maximum amplitude spectra suggest an important con-
Accuracy of Non-Stationary Fourier dition on the data window width. The data window width
Spectra longer than the one-cycle dominant component may mislead
The parameters in non-stationary Fourier spectra are determined the maximum amplitude of the dominant component. As for
based on the analytical results using sinusoidal waves. the cyclic wave, the data window width should be selected
appropriately depending on the frequency component in the
( 1) One-cycle sinusoidal wave (see FigureA1). seismic record. Parametric analysis may be necessary for the
The one-cycle sinusoidal wave with the frequency of 1.0 Hz data window width.
was analyzed with three kinds of data window width, such (3) The continuously modulating sinusoidal waves (see

as 1.0, 1.5, and 2.0 s. The data window width longer than the FigureA3).
duration of dominant component may mislead the occur- The non-stationary property of seismic records can be mod-
rence time of the dominant component but has no influence eled by using the continuously modulating sinusoidal wave
on the maximum amplitude spectra. (from 1 to 9 Hz) expressed by Eq. A1.
In the analysis of seismic records, the dominant frequency
can be estimated from the shape of seismic record, i.e. a pulse
{
y ( t ) = sin 2 (1 + 8.0 t TN ) t ; TN = 10 s (A1) }
wave, a cyclic wave. The suitable data window should be where TN is the duration.
selected based on some case studies of data window width.
The data window width longer than the dominant com- The frequency components with a short duration time can
ponent may mislead the occurrence time of the dominant be analyzed correctly. The non-stationary Fourier spectra are
component. applicable to seismic records.

2 2 2
Acceleration (m/s 2)

Acceleration (m/s 2)

Acceleration (m/s 2)
Tj=0.5s SIN 1cycle Tj=1.0s SIN 1cycle Tj=1.0s SIN 1cycle
1 1 1
Tj=1.0s Tj=1.5s Tj=1.5s
0 0 0
Tj=1.5s Tj=2.0s Tj=2.0s Tj=2.5s Tj=2.0s Tj=2.5s
-1 -1 -1

-2
Tj=2.5s -2
Tj=3.0s -2 Tj=3.0s
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Time (s) Time (s) Time (s)
5 5 5
SIN(1.0Hz)1cycles SIN(1.0Hz)1cycles SIN(1.0Hz)1cycles
Data-window width-2.0s Data-window width-1.5s Data-window width-1.0s 100
4 FFT analytical time =20.48s 4 FFT analytical time =20.48s 4 FFT analytical time =20.48s 90
Frequency (Hz)

Frequency (Hz)

Frequency (Hz)

80
70
3 3 3 60
50
2 2 2 40
30
20
1 1 1 10
0
0 0 0
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Time (s) Time (s) Time (s)
Data window width=2.0s Data window width=1.5s Data window width=1.0s

1.5
Maximum amplitude (m/s 2)

Data-window width
2.0s
1.5s
1 1.0s

0.5

0
0 1 2 3 4 5
Frequency (Hz)

Maximum amplitude spectra

FIGURE A1 | Non-stationary Fourier spectra of one-cycle sinusoidal wave.

Frontiers in Built Environment | www.frontiersin.org 17 February 2017|Volume 3|Article 13

Kamagata and Takewaki Occurrence Mechanism of Large Acceleration

Acceleration (m/s 2)
SIN 3cycles
1

-1

-2
0 1 2 3 4 5
Time (s)
5 5 5
SIN(1.0Hz)1cycles SIN(1.0Hz)1cycles SIN(1.0Hz)1cycles
Data-window width-2.0s Data-window width-1.5s Data-window width-1.0s 100
4 FFT analytical time =20.48s 4 FFT analytical time =20.48s 4 FFT analytical time =20.48s 90
Frequency (Hz)

Frequency (Hz)
Frequency (Hz)
80
70
3 3 3 60
50
2 2 2 40
30
20
1 1 1 10
0
0 0 0
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Time (s) Time (s) Time (s)
Data window width=2.0s Data window width=1.5s Data window width=1.0s
3
Data window width
Acceleration amplitude

2.0s
2 1.5s
1.0s
(m/s2)

0
0 1 2 3 4 5
Frequency (Hz)
Maximum amplitude spectra

FIGURE A2 | Non-stationary Fourier spectra of three-cycle sinusoidal wave.

1.5
Acceleration (m/s2)

1
0.5
0
-0.5
-1
1.5
-1.5
Maximum amplitude (m/s2)

Sweep wave
0 2 4 6 8 10
Time (s)
10 1
Sweep(1-9Hz)
FFT analytical time=5.12s
8 Data-window width=1.0s
0.5
Frequency (Hz)

6
0
4 0 2 4 6 8 10
Frequency (Hz)
2

0
0 2 4 6 8 10
Time (s)

FIGURE A3 | Non-stationary Fourier spectra of continuously modulating sinusoidal wave.

Frontiers in Built Environment | www.frontiersin.org 18 February 2017|Volume 3|Article 13