Seismological Research Letters
V
olume 80, Number 4
J
uly/August 2009
5
59
doi: 10.1785/gssrl.80.4.559
INTRODUCTION
The 2008 Iwate-Miyagi Nairiku earthquake (
M
w
6.9,
M
jma
7.2 )
produced strong shaking throughout northern Honshu, Japan,
with severe damage to buildings and extensive landslides.
The shallow event occurred in southwestern Iwate Prefecture
(39.03°N, 140.88°E, depth 8 km) on 13 June 2008 at 23:43:45
GMT (Japan Meteorological Agency 2008). This earthquake
produced relatively high-frequency ground motions, which
resulted in large values of peak ground acceleration (PGA).
The surface accelerometer of the station IWTH25 of KiK-net,
located 3 km southwest of the epicenter, produced one of the
largest strong-motion values of PGA (4,278 cm/s
2
for the vec-
tor sum of the three components) ever recorded (
http://www.
kik.bosai.go.jp/kik/index_en.shtml).
The new accelerometers installed in KiK-net last year have
a recording range up to 4,000 cm/s
2
, which made it possible
to record such large ground motions near the source (
http://
www.kik.bosai.go.jp/kik/index_en.shtml).The
sampling rate of
the record of IWTH25 is 100 Hz (
http://www.kik.bosai.go.jp/
kik/index_en.shtml).
The surface acceleration record at station IWTH25 shows
an asymmetric amplification in the vertical components (Aoi
et
al.
2008). The upward vertical acceleration is much larger than
the downward direction, although in the borehole record at a
depth of 260 m at the same site, the upward and downward
accelerations have symmetric amplitudes (Figure 1). On the
other hand, the horizontal components do not show this asym-
metric effect. This difference between the surface and borehole
recordings for the vertical component implies a strong nonlin-
ear amplification. In this paper, we will analyze these records
and propose a mechanism to produce the large vertical accel-
erations. The predominance of large upward acceleration spikes
is not unique to the Iwate-Miyagi Nairiku earthquake, so our
proposed mechanism may be applicable to a number of large
vertical acceleration records.
THE 2008 IWATE-MIYAGI NAIRIKU EARTHQUAKE
The KiK-net station IWTH25 (operated by the National
Research Institute for Earth Science and Disaster Prevention)
has accelerometers on the surface and in a borehole at a depth
of 260 m. In Figure 1, the red and black lines show the bore-
hole and ground surface records, respectively. The amplifica-
tion of the vertical acceleration is much larger than that of the
horizontal acceleration, and many of the upward peaks of the
vertical acceleration are much larger than the downward peaks.
However, the velocity and displacement waveforms (time-
domain integration of acceleration) are quite similar for the
borehole and ground surface data. These data indicate that the
large amplitude high-frequency accelerations are due to near-
surface effects and are not coming from the earthquake source.
The frequency-dependent amplifications in the near surface are
somewhat different for the mainshock and a large aftershock
(
M
j
5.6). On the horizontal component there appears to be
larger amplification for the aftershock in the 10 to 20 Hz range,
while for the vertical component there is a larger amplification
for the mainshock in the 10 to 20 Hz range (Figure 2).
Taking a closer look at the acceleration record, Figure
3 shows the borehole and surface accelerations focused on
the time of the large amplitudes. The borehole acceleration
is symmetric, but the surface acceleration is asymmetric in
both amplitude and frequency. The positive pulses (dark-gray-
colored sections) are narrow with large amplitude, while the
negative pulses (light-gray-colored sections) are broader with
smaller amplitude. The areas of the upward and downward
pulses are the same, which explains why integrating the acceler
-
ation records gives similar velocity records for the borehole and
surface records. Therefore, the borehole record is regarded as
the input ground motion, and the surface record is a combina-
tion of this input motion and the high-frequency near-surface
response. Note that the 100-Hz sampling might not be high
enough to record the actual high-frequency accelerations asso-
ciated with the impact of a separated layer at depth, although
the near-surface attenuation will also damp the motions.
t
he
s
l
apdown
p
h
ase in High-acceleration
r
e
cords of
l
a
rge
e
a
rthquakes
Masumi Yamada, Jim Mori,
and Thomas Heaton
Masumi Yamada,
1
Jim Mori,
2
and Thomas Heaton
3
1.
P
ioneering Research Unit for Next Generation, Kyoto University,
Japan
2.
D
isaster Prevention Research Institute, Kyoto University, Japan
3.
C
alifornia Institute of Technology, Pasadena, California
560
S
eismological Research Letters
V
olume 80, Number 4
J
uly/August 2009
Yamada, Mori and Heaton
12
Figures and Tables
Figure 1: Three-component acceleration, velocity, and displacement records at station IWTH25 for the
2008 Iwate-Miyagi Nairiku earthquake. The red and black lines are records for the borehole and surface,
respectively.
▲
F
igure 1.
Three-component acceleration, velocity, and displacement records at station IW TH25 for the 2008 Iwate-Miyagi Nairiku
earthquake. The red and black lines are records for the borehole and surface, respectively.
Yamada, Mori and Heaton
13
EW
NS
UD
EW
NS
UD
Mainshock
Aftershock
10
100
1000
10
100
1000
10
100
1000
10
100
1
0.1
10
100
1
0.1
10
100
1
0.1
0.1 1 10
0.1 1 10
Surface
Borehole
Surface
Borehole
Surface
Borehole
Surface
Borehole
Surface
Borehole
Surface
Borehole
Figure 2: Acceleration amplitude spectra at the station IWTH25 in the EW, NS, and UD components from
the top. The black and gray lines show the borehole and ground surface records, respectively. The left
column is for the mainshock, and the right column is for the largest aftershock (
M
j
5.6).
Station
Earthquake
Date
Acc+
Acc-
Soil condition
Reference
gazli
Gazli
1976/5/17
1310 1040
tertiary sedimentary
rock
COSMOS VDC
site1
Nahanni
1985/12/23
2309
631 bedrock
COSMOS VDC
IWTH04
Miyagi
2003/5/26
1280
480 Vs30: 456m/s
KiK-net
TTN034
Chengkung
2003/12/10
1866 1157 Class D
CWB
041
Chuetsu
2004/10/23
1059
815 Vs30: 641m/s
Kubo et al. (2003)
AKTH04 Iwate-Miyagi 2008/6/14
1094
847 Vs30: 459m/s
KiK-net
IWTH25
Iwate-Miyagi 2008/6/14
3866 1703 Vs30: 526m/s
KiK-net
Table 1: List of the records of which the vertical acceleration exceeds 1000 cm/s
2
. The columns show
the station ID, earthquake, date of the earthquake, upward PGA (cm/s
2
), downward PGA (cm/s
2
), and soil
condition. The soil condition is cited from the COSMOS Virtual Data Center website, KiK-net website,
personal communication with CWB, and Kubo et al. (2003). The VS30 is the average shear wave velocity
for the surface 30m. Class D corresponds to the Vs30 is equal to 180-360 m/s.
▲
F
igure 2.
Acceleration amplitude spectra at station IW TH25 in the EW, NS, and UD components from the top. The black and gray
lines show the borehole and ground surface records, respectively. The left column is for the mainshock, and the right column is for the
largest aftershock (
M
j
5.6).
Seismological Research Letters
V
olume 80, Number 4
J
uly/August 2009
5
61
LARGE VERTICAL ACCELERATIONS FOR OTHER
EARTHQUAKES
Figure 4 shows the available strong-motion records that have
vertical accelerations over 1 g. Table 1 shows the PGA values
and soil conditions at the stations. Most of the records are on
stiff soil. One can see that all the records tend to have larger
upward accelerations than downward. Also, the vertical com-
ponent tends to be larger than the horizontal component. For
example, the record of the 1976 Gazli earthquake (
M
w
6.3)
shows a vertical PGA (1,310 cm/s
2
) that is about twice as large as
the horizontal PGA (729 cm/s
2
). The vertical PGA (2,321 cm/
s
2
) of site1 for the 1985 Nahanni earthquake (
M
w
6.4) is also
twice as large as the horizontal PGA (1,338 cm/s
2
).
Th
e asymmetric amplification of the vertical acceleration
is also very obvious for a number of the vertical acceleration
records from around the world (see Figure 4). Figure 5(A)
shows the relationship between the upward and downward
PGA of the vertical records that have amplitudes greater than
500 cm/s
2
. The U/D ratio (upward-to-downward peak accel
-
eration ratio) is close to 1 if the vertical PGA is less than 1 g
(980 cm/s
2
), and the U/D ratio is significantly larger if the
vertical PGA is greater than 1 g. For records that have accel
-
erations greater than1 g, the positive amplitudes of the vertical
acceleration are larger than the negative amplitudes, and the
downward accelerations seem to have a lower bound of about
1 g (Figure 4).
SLAPDOWN PHASE
Large upward spikes in acceleration are observed in near-field
observations of nuclear explosions (Eisler and Chilton 1964;
Chilton
et al.
1966), which may be analogous to these strong-
motion records. There is a substantial body of work on spall
that comes from explosion seismology beginning in the 1960s
(Eisler
et al.
1966; Day
et al.
1983; Viecelli 1973; Springer
1974; Day and McLaughlin 1991). In the process of nuclear
explosions, an upper soil layer separates (spalls) and is flung
upward due to large tensile stress from extremely large accel
-
erations caused by the explosion (on the order of several tens
of g). Then, the layer free flies to the ground with a downward
acceleration controlled by gravity. When the returning layer
hits the original separated surface, a large upward spike in
acceleration is produced (slapdown phase). Figure 6 (left) shows
the particle acceleration and velocity from a surface instrument
during a nuclear explosion (Perret 1972). The velocity was digi
-
tized from the original figure in the paper and differentiated to
acceleration. In the acceleration record, the first upward spike is
the direct shock from the nuclear explosion, then the extended
acceleration at negative 1 g is due to free flight, and the second
upward spike is the slapdown phase. Figure 5 (B) shows the
comparison of the U/D ratio of vertical accelerations for large
earthquakes and nuclear explosions (Perret 1973.) The records
for the nuclear explosions also have characteristics similar to
records for large earthquakes, as the positive amplitudes are
larger than the negative amplitudes and the downward accel-
erations have a lower bound of about 1 g.
Figure 6 (middle) and Figure 6 (right) are earthquake
strong-motion records which we interpret in the same way,
assuming that there is a near-surface soil layer separated from a
sublayer. The relatively long-period negative acceleration is asso-
ciated with the free flight of the near-surface layer that has been
flung upward by large vertical accelerations. When the surface
layer returns and hits the sublayer, the positive sharp spike in
Yamada, Mori and Heaton
14
Figure 3: Expanded view of acceleration waveforms of station IWTH25. Top is the surface record and
b
ottomistheboreholerecord. TherightfigureshowsP-andS-wavevelocitystructuresatthestation.
▲
F
igure 3.
Expanded view of acceleration waveforms of station IW TH25. Top is the surface record and bottom is the borehole record.
The right figure shows
P
- and
S
-wave velocity structures at the station.
562
S
eismological Research Letters
V
olume 80, Number 4
J
uly/August 2009
acceleration is produced. One difference from the explosion
record is that the input ground motions from the earthquake
are not a single pulse, as in the explosion. This interpretation is
somewhat similar to the explanation of (Aoi
et al.
2008), which
uses a model of a mass bouncing on a trampoline. Both of these
interpretations invoke a free flight of the near-surface layer to
explain the negative 1-g accelerations.
In Figure 4, the large upward spikes in the records at site1
for 1985 Nahanni earthquake, station IWTH04 for the 2003
Miyagi-ken Hokubu-oki earthquake, and station TTN034 for
the 2003 Chengkung earthquake appear to show clear indi
-
vidual slapdown phases. The records at stations IWTH25 and
AKTH04 for the 2008 Iwate-Miyagi Nairiku earthquake and
station gazli for the 1976 Gazli earthquake show more com
-
plicated waveforms that may include multiple slapdown phases
along with the input ground motion. All stations in Figure 4
are considered to be on relatively stiff sites, so the slapdown
phase is likely to be generated by a brittle fracture in the sub-
surface.
There are several reasons why the slapdown phases are not
seen in the borehole record of the 2008 Iwate-Miyagi Nairiku
earthquake. Note that similar to acceleration waveforms from
nuclear explosions, the slapdown phase was not clearly seen on
a borehole accelerometer (Eisler
et al.
1966). The downhole
accelerometer is four to eight times as far as the surface accel-
erometer, assuming the separation of the surface layer is 30
to 60 m (this assumption is explained in the next section), so
that the high-frequency waves may largely be attenuated. Also,
the shallow lower-velocity material tends to amplify the waves
more than the harder layers at depth. Probably most impor-
tant, although the stress at the separated boundary is the same
in the upward and downward directions, is that the accelera-
tion of the upper separated layer will be larger than that of the
basement rock, which is fixed to the rest of the Earth. This is
somewhat analogous to the larger ground motions seen for the
hanging wall of thrust faults compared to the foot wall.
SEPARATION OF THE NEAR-SURFACE LAYER
Using this model of a spalling near-surface layer, we can esti
-
mate the dimensions of the thickness of the layer and the
amount it separates from the sublayer. Eisler and Chilton
(1964) show that the thickness of the spalled layer and the spall
gap (vertical displacement of the surface layer from the sepa-
Yamada, Mori and Heaton
15
2004 Chuetsu, 041
2008 Iwate, AKTH04
1976 Gazli, gazli
2003 Miyagi, IWTH04
2008 Iwate, IWTH25
1985 Nahanni, site1
2003 Chengkung, TTN034
Figure 4: Acceleration records on stiff sites that have vertical accelerations over 1g.
▲
F
igure 4.
Acceleration records on stiff sites that have vertical
accelerations over 1 g.
TABLE 1
List of the records in which the vertical acceleration exceeds 1,000 cm/s
2
. The columns show the station ID, earthquake,
date of the earthquake, upward PGA (cm/s
2
), downward PGA (cm/s
2
), and soil condition. The soil condition is cited from the
COSMOS Virtual Data Center Web site,
http://db.cosmos-eq.org/scripts/default.plx;
KiK-net website,
http://www.kik.bosai.go.jp/
kik/index_en.shtml;
personal communication with Central Weather Bureau in Taiwan; and Kubo
et al.
2003.The VS30 is the
average shear wave velocity for the surface 30 m. Class D corresponds to the Vs30
=
180 –360 m/s.
Station
Earthquake
Date
Acc+
Acc–
Soil Condition
Reference
gazli
Gazli
1976/5/17
1310
1040
3.5cm tertiary sedimentary rock
COSMOS VDC
site1
Nahanni
1985/12 /23
2309
631
Bedrock
COSMOS VDC
IW TH04 Miyagi
2003/5/26
1280
480
Vs30: 456m/s
KiK-net
T TN034
Chengkung
2003/12 /10
1866
1157
Class D
CWB
041
Chuetsu
2004/10/23
1059
815
Vs30: 641m/s
Kubo
et al.
2003
AK TH04 Iwate-Miyagi
2008/6/14
1094
847
Vs30: 459m/s
KiK-net
IW TH25 Iwate-Miyagi
2008/6/14
3866
1703
Vs30: 526m/s
KiK-net
Seismological Research Letters
V
olume 80, Number 4
J
uly/August 2009
5
63
ration surface) can be computed from the acceleration data.
When spalling occurs, surface acceleration records consist of a
direct pulse and slapdown phase separated by a period of con-
stant negative gravity acceleration (Eisler and Chilton 1964;
Eisler 1967). Assuming the separation surface does not move
and the spalled layer returns to the original level,
S
g
t
g
max
=
Δ
( )
,
8
2
(
1)
where
S
max
is the maximum spall gap,
g
is a gravity acceleration,
and ∆
t
g
is the total free-flight time. The slapdown phase is gen-
erated while the compressional wave travels from the surface
and reflects back from the now-closed spall gap. Again, assum-
ing no seismic wave follows the slapdown phase, the thickness
of the spalled layer (
d
) is computed from the duration of the
slapdown phase (∆
t
s
) and the
P
-wave velocity (
ν
p
):
d
t
v
s
p
= .
2
(
2)
If we assume the seismic ground motion can be treated as
a sequence of multiple slapdown phases, the thickness of the
layer and the separation gap can be computed from the records.
In Figure 3, the duration of ∆
t
g
varies for the pulses, and the
duration of ∆
t
s
is almost constant. This observation implies
that the spall gap changes depending on the pulse, but the
thickness of the spalled layer is constant. By Equations 1 and
2, the amount of separation of the layer (
S
max
) is about 1 to
12 mm, and the thickness of the layer (
d
) is about 46 to 58 m
Yamada, Mori and Heaton
16
Figure 5: Relationship between the upward and downward PGA for large acceleration records.
slapdown
free-flight
Nuclear Explosion
site1 (1985 Nahanni)
IWTH25 (2008 Iwate)
Time (second)
Time (second)
Time (second)
Acceleration
Velocity
8.5
9.5
9.0
4.0
5.0
4.5
Figure 6: Slapdown phase observed in the record of the nuclear explosion on Perret (1972), site1 during the
1985 Nahanni earthquake, and IWTH25 during the 2008 Iwate-Miyagi Nairiku earthquake.
Yamada, Mori and Heaton
16
Figure 5: Relationship between the upward and downward PGA for large acceleration records.
slapdown
free-flight
Nuclear Explosion
site1 (1985 Nahanni)
IWTH25 (2008 Iwate)
Time (second)
Time (second)
Time (second)
Acceleration
Velocity
8.5
9.5
9.0
4.0
5.0
4.5
Figure 6: Slapdown phase observed in the record of the nuclear explosion on Perret (1972), site1 during the
1985 Nahanni earthquake, and IWTH25 during the 2008 Iwate-Miyagi Nairiku earthquake.
▲
F
igure 5.
Relationship between the upward and downward PGA for large acceleration records.
▲
F
igure 6.
Slapdown phase observed in the record of the nuclear explosion from Perret (1972), at site 1 during the 1985 Nahanni
earthquake, and at IW TH25 during the 2008 Iwate-Miyagi Nairiku earthquake.
564
S
eismological Research Letters
V
olume 80, Number 4
J
uly/August 2009
for the acceleration record at station IWTH25 in Figure 4. The
layer thickness is consistent with the velocity profile for this
station, which shows sand and sandy clay layers (
http://www.
kik.bosai.go.jp/kik/index_en.shtml).
In the record at site1 for
the Nahanni earthquake in Figure 6, the separation distance is
about 2 to 8 mm and the layer thickness is about 30 m, assum
-
ing
ν
p
=
1 km/s. For these examples, the range of the separation
gap is roughly less than 1 cm, and the range of the thickness of
the near-surface layers is 30 to 60 m. These numbers seem to be
reasonable values for the mechanism to produce the asymmet-
ric acceleration records.
CONCLUSIONS
We analyzed the 4-g record of the 2008 Iwate-Miyagi Nairiku
earthquake and provide an explanation for the asymmetric
amplification in the vertical acceleration. We interpret the
large upward spikes in acceleration as slapdown phases, which
are also typically observed in near-field recordings of nuclear
explosion tests. The large upward acceleration is produced
when a near-surface layer separates from the sublayer and then
returns, striking the separation surface. This effect is seen in a
number of strong-motion records that have larger upward than
downward accelerations. If we assume the near-surface layer
returns to the original level, the separation gap is roughly 1 to
12 mm, and the thickness of the layer that is flung upward is
30 to 60 m.
ACKNOWLEDGMENTS
The authors acknowledge the National Research Institute for
Earth Science and Disaster Prevention (NIED) and the Japan
Meteorological Agency for the use of strong motion data. We
thank Dr. Yih-Min Wu at National Taiwan University for
offering the record of the Chengkung earthquake and Dr. Luis
Rivera at Louis Pasteur University and Dr. Hiroyuki Goto at
Kyoto University for meaningful discussion. Reviewers pro
-
vided detailed comments that greatly improved this manuscript.
This research was funded by the Program for Improvement of
Research Environment for Young Researchers from Special
Coordination Funds for Promoting Science and Technology
(SCF) commissioned by the Ministry of Education, Culture,
Sports, Science and Technology (MEXT) of Japan.
REFERENCES
Aoi, S., T. Kunugi, and H. Fujiwara (2008). Trampoline effect in extreme
ground motion.
Science
322
, 727–729.
Chilton, F., J. Eisler, and H. Heuback (1966). Dynamics of spalling of
the Earth’s surface caused by underground explosions.
Journal of
Geophysical Research
71
, 5,911–5,919.
Day, S., and K. McLaughlin (1991). Seismic source representations for
spall.
Bulletin of the Seismological Society of America
81
(1),191–201.
Day, S., N. Rimer, and J. Cherry (1983). Surface waves from underground
explosions with spall: Analysis of elastic and nonlinear source mod
-
els.
Bulletin of the Seismological Society of America
73
(1), 247–264.
Eisler, J. (1967). Near-surface spalling from a nuclear explosion in a salt
dome.
Journal of Geophysical Research
72
, 1,751–1,760.
Eisler, J., and F. Chilton (1964). Spalling of the Earth’s surface by under
-
ground nuclear explosions.
Journal of Geophysical Research
69
,
5,285–5,293.
Eisler, J., F. Chilton, and F. Sauer (1966). Multiple subsurface spalling
by underground nuclear explosions.
Journal of Geophysical Research
71
, 3,923–3,927.
Japan Meteorological Agency (2008). Iwate-Miyagi Nairiku earthquake
in 2008.
http://www.seisvol.kishou.go.jp/eq/2008_06_14_iwate-
miyagi/index.html.
Kubo, T., Y. Hisada, A. Shibayama, M. Ooi, M. Ishida, H. Fujiwara, and
K. Nakayama (2003). Development of digital maps of site ampli
-
fication factors in Japan, and their applications to early strong
motion estimations.
Journal of the Seismological Society of Japan
56
(1), 21–37.
Perret, W. (1972). Close-in ground motion from the MILROW and
CANNIKIN events.
Bulletin of the Seismological Society of America
62
(6), 1,489–1,504.
Perret, W. (1973).
Ground Motion in the Vicinity of the CANNIKIN
Nuclear Explosion
. Report SLA-73, 0043, Sandia Laboratories,
Albuquerque, NM.
Springer, D. (1974). Secondary sources of seismic waves from under
-
ground nuclear explosions.
Bulletin of the Seismological Society of
America
64
(3–1), 581–594.
Viecelli, J. (1973). Spallation and the generation of surface waves by an
underground explosion.
Journal of Geophysical Research
78 (
14),
2,475–2,487.
Pioneering Research Unit for Next Generation
Kyoto University
Uji, Gokasho, 611-0011, Japan
masumi@eqh.dpri.kyoto-u.ac.jp
(M. Y.)
Disaster Prevention Research Institute
Kyoto University
Uji, Gokasho, 611-0011, Japan
(J. M.)
California Institute of Technology
Pasadena, California, 91125, U.S.A.
(T. H.)