Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Mechanisms for Generation of Near-Fault Ground Motion Pulses for Dip-Slip Faulting
NATALIA POIATA,1 HIROE MIYAKE,2 and KAZUKI KOKETSU2
Abstract—We analyzed the seismological aspects of the near-
fault ground motion pulses and studied the main characteristics of
the rupture configuration that contribute to the pulse generation for
dip-slip faulting events by performing forward simulations in
broadband and low-frequency ranges for different rupture scenarios
of the 2009 L’Aquila, Italy (Mw 6.3) earthquake. The rupture
scenarios were based on the broadband source model determined
by Poiata et al. (Geophys J Int 191:224–242, 2012). Our analyses
demonstrated that ground motion pulses affect spectral character-
istics of the observed ground motions at longer periods, generating
significantly larger seismic demands on the structures than ordinary
records. The results of the rupture scenario simulations revealed the
rupture directivity effect, the radial rupture propagation toward the
site, and the focusing effect as the main mechanisms of the near-
fault ground motion pulse generation. The predominance of one of
these mechanisms depends on the location of the site relative to the
causative fault plane. The analysis also provides the main candidate
mechanisms for the worst-case rupture scenarios of pulse genera-
tion for the city of L’Aquila and, more generally, the hanging-wall
sites located above the area of large slip (strong motion generation
area).
Key words: Ground motion pulses, dip-slip faulting, direc-
tivity effect, focusing effect, 2009 L’Aquila earthquake, worst-case
rupture scenario.
1. Introduction
Seismic recordings of past earthquakes indicate
that near-fault ground motions could be significantly
different from those observed at larger distances from
seismic faults. The presence of strong coherent long-
period pulses recorded at some stations, correspond-
ing to a specific geometry of the fault-station
configuration, was reported as a distinctive charac-
teristic of the near-fault ground motions (e.g., Aki
1968; Somerville et al. 1997; Koketsu and Miyake
2008). The rapid development of strong motion net-
works allowed to capture long-period ground motion
pulses near seismic faults during large damaging
events like the 1999 Kocaeli (Mw 7.6) earthquake and
the 1999 Chi–Chi (Mw 7.6) earthquake, as well as
during smaller events such as the 1994 Northridge
(Mw 6.7), 1995 Kobe (Mw 6.9), and 2003 Bam (Mw
6.6) earthquakes. Most of these earthquakes resulted
in substantial material damage and loss of human
lives. The effect that near-fault ground motion pulses
can have on the engineered structures was first
revealed during the 1994 Northridge earthquake. It
was recognized that the considerable damage that
was observed as the result of this earthquake was
caused by large pulse-like ground motions recorded
in the near-source area (e.g., Heaton et al. 1995;
Strasser and Bommer 2009). The building codes that
existed during that period did not consider any cur-
rently known near-source effects. An example
illustrating the importance of including the near-
source effects into the building practice of critical
facilities was given by the experience of the Kashi-
wazaki-Kariwa nuclear power plant during the 2007
Chuetsu-oki (Mw 6.6) earthquake in Japan. This
reverse faulting event was the world’s first major
earthquake that occurred on a causative fault
extending beneath a nuclear power plant (Miyake
et al. 2010). The ground motions recorded by the
instruments inside the plant and characterized by
three seismic pulses were significantly stronger than
those that were anticipated at the time of its design.
The detailed analyses of the earthquake’s source
process by Miyake et al. (2010) determined that the
observed ground motion pulses corresponded to the
combined effect of the distribution of asperities (ar-
eas of large slip), rupture propagation, and the
S-wave radiation pattern. Fortunately, the plant
1 National Institute for Earth Physics, 12 Calugareni, 077125
Magurele, Ilfov, Romania. E-mail: [email protected] Earthquake Research Institute, University of Tokyo, 1-1-1
Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan.
Pure Appl. Geophys.
� 2017 Springer International Publishing
DOI 10.1007/s00024-017-1540-z Pure and Applied Geophysics
structures suffered only minor damage due to the
ground motion.
Long-period ground motion pulses recorded in
near-source regions are mainly attributed to the rup-
ture directivity effect. This phenomenon can be
observed in both strike-slip and dip-slip faulting (e.g.,
Koketsu et al. 2016). The rupture directivity effect is
observed when the rupture front propagates over the
earthquake source fault at high speed, typically
slightly less than the shear wave (S-wave) velocity of
the media. In this case, if a site is located in the
direction of rupture, most seismic energy of the wave
front will arrive in a single pulse of ground motion
(Fig. 1). The conditions contributing to the forward
directivity effect and the characteristics of the rupture
directivity pulses were identified and summarized by
Somerville et al. (1997) based on the near-fault
records of the 1992 Landers earthquake. The theo-
retical aspects of the physical mechanism describing
their generation were addressed in numerous studies
(e.g., Boore and Joyner 1978, 1989; Heaton 1990;
Joyner 1991; Miyatake 1998). It was also pointed out
that the forward directivity effect generates pulses
with dominant periods of C0.6 s, strongly affecting
the spectral content of the ground motions (e.g.,
Somerville et al. 1997; Somerville 2003; Koketsu and
Miyake 2008), and placing significant demands on
the structures near the earthquake source fault.
Moreover, Somerville (2003) indicated that the per-
iod of the pulse increases with the magnitude of the
event. Thus, even smaller earthquakes are able to
generate ground motions that could result in
increased seismic demand and have damaging effect
on structures in the near-fault areas.
The rupture directivity effect as described by
Somerville et al. (1997) explains well the fault-nor-
mal components observed during past strike-slip
faulting events (e.g., 1992 Landers). In case of dip-
slip faulting, however, both the near-fault ground
motion pulses and the hanging-wall effect (Abra-
hamson and Somerville 1996) are observed at some
stations. Inspection of 3-D S-wave radiation patterns
for dip-slip faulting (Fig. 2), as well as the near-fault
strong ground motion records from the stations on the
hanging-wall sites above the causative fault of the
2007 Chuetsu-oki (reverse faulting; Miyake et al.
2010), 2009 L’Aquila (normal faulting) and 2015
Gorkha (low-angle dip-slip faulting, Koketsu et al.
Forward directivity
Backward directivity
Direction of rupturepropagation
Wavefronts of S-wave
Site ASite B
Figure 1Snapshot of the S-wave fronts, illustrating the rupture directivity effect. Site A is located in the direction of the rupture propagation, while Site
B is located in the direction opposite to that of the rupture propagation. The propagating rupture is represented by a finite fault comprising four
point sources (white stars) placed at a distance R from each other; the rupture is assumed to start from the leftmost point and propagate toward
the right (black arrow) at a velocity Vr = 0.9Vs. The theoretically calculated synthetic waveforms from each of the four sources (black traces)
and the resulting total velocity waveforms (red traces) are shown for Sites A and B (marked as blue triangles)
N. Poiata et al. Pure Appl. Geophys.
Koketsu et al. 2016) earthquakes pointed toward a
more complex mechanism of pulse generation for
such events.
In this study, we use the observed near-field
ground motions of the 1992 Landers (Mw 7.3) strike-
slip and 2009 L’Aquila (Mw 6.3) normal faulting
earthquakes to demonstrate that the ground motion
pulses can significantly affect the shape of the
response spectra. This effect leads to an increase of
seismic demand on structures being potentially the
main cause of structural damage, such as those
observed during the analysed events. We also inves-
tigate the rupture configuration parameters that
contribute to the generation of the near-fault ground
motion pulses in dip-slip faulting events. This is
achieved by performing ground motion simulations
of the recordings at near-fault strong motion stations
in both broadband and low-frequency ranges using
the specific rupture scenarios based on the 2009
L’Aquila (Mw 6.3) earthquake source model. The
(a) (b)
(c)
δhanging wall
foot wall
X X
along strike rupture propagation
Site ASite B
Site A Site A
FN FN
Site A Site B Site A Site B
up-dip rupture propagation
(d) (e)
(f) (g)
up-dipalong strike
Figure 2a 3-D low-frequency and b high-frequency S-wave radiation patterns from a point source shear dislocation. The rectangles indicate the planes
of faulting and the arrows show the slip directions. c Schematic of a normal fault. The red star corresponds to the rupture starting point, the
up-dip and along-strike directions are shown by the full black arrows, and the slip directions by the open arrows. The blue triangles indicate
the location of the observation sites. Site A is located on the hanging-wall side and above the rupture plane, and Site B is located on the
footwall side of the fault. d Vertical section of the low-frequency radiation pattern of the up-dip rupture propagation for the normal fault (c).
The shaded area corresponds to the region of the maximum forward directivity effect, and the black arrow shows the orientation of the
directivity pulses observed in Site B. e Same as d, but for the high-frequency range. f Vertical section of the low-frequency radiation pattern of
the along-strike rupture propagation for the normal fault. c The slip direction is shown in circles next to the rupture starting point (red star).
g Same as f, but for the high-frequency range
Ground Motion Pulses for Dip-Slip Faulting
assumed scenarios are designed using the broadband
source model of Poiata et al. (2012). We further
analyzed the results of ground motion simulations
identifying the rupture configurations that are most
efficient in generating near-fault ground motion pul-
ses for the city of L’Aquila by comparing the
simulated and observed ground motions recorded at
the AQU strong motion station located inside the city.
1.1. Near-Fault Ground Motion Pulses
An earthquake results from shear dislocation
propagating along the fault. The 3-D low-frequency
radiation pattern of the S-wave generated by a point
shear dislocation on the horizontal plane is shown in
Fig. 2a (for a full mathematical description, see Aki
and Richards 1980). Moreover, high-quality wave-
form records from the stations at small and
intermediate distances from seismic sources indicated
that in the higher frequency range of 1–3 Hz, the
four-lobed radiation pattern of the S-wave is distorted
(e.g., Liu and Helmberger 1985; Takenaka et al.
2003; Takemura et al. 2009). The distortion results in
an isotopic distribution of S-wave amplitudes in all
directions for frequencies [3 Hz (Liu and Helm-
berger 1985). This phenomenon was attributed to the
scattering of the seismic waves by the heterogeneous
structure around the source region (‘‘heterogeneities
in the source-process’’; Liu and Helmberger 1985)
and along the propagation path of the seismic waves
(Takemura et al. 2009). Based on the results of these
studies and assuming some degree of simplification,
we represent the high-frequency radiation of S-waves
generated by a point shear dislocation as a sphere
(Fig. 2b). The limit between the low- and high-
frequency here is assumed at around 1–3 Hz, follow-
ing the results of Liu and Helmberger (1985).
Close to the earthquake fault, where the point-
source approximation is not valid, one has to consider
the fact that an earthquake is a shear dislocation that
propagates along a finite fault. Mathematical simpli-
fication of this is equivalent to a distribution (e.g.,
along a line) of the point sources over the fault, and
the propagation of the dislocation can then be
introduced by considering the appropriate time delays
(e.g., Kasahara 1981). This will result in an azimuth
dependence of the arrival times and summary
amplitudes of the waveforms observed at different
locations surrounding the ruptured fault, which is also
known as directivity (e.g., Aki and Richards 1980;
Lay and Wallace 1995). The effect of rupture
directivity on the S-wave point-source radiation
pattern symmetry can be described by the directivity
function (e.g., Aki and Richards 1980). The effects of
directivity on the S-wave radiation pattern in the low-
and high-frequency ranges of the seismic wave field
in the case of a dip-slip fault are shown in Fig. 2d, e.
The resulting elongation of the radiation pattern in
the direction of the rupture propagation corresponds
to the forward directivity effect discussed previously.
The forward rupture directivity effect at a given
site occurs under the following conditions (Somer-
ville et al. 1997): the rupture front must propagate
toward the site, the direction of the slip on the fault
must be aligned with the direction of the rupture, and
the rupture must propagate at a speed close to that of
the S-wave. These conditions are most easily fulfilled
in the case of strike-slip faulting, when the rupture
typically propagates horizontally along the strike, and
the slip on the fault is oriented along the direction of
the strike. Evidence for the directivity effects is seen
in the fault-normal components of the seismograms
recorded during the past strike-slip faulting events
(e.g., 1992 Landers earthquake; Fig. 3a) confirming
the above description. Following Somerville et al.
(1997), it is possible to show that the conditions for
the forward directivity effect are also satisfied in the
case of dip-slip faulting for both reverse and normal
faults (Fig. 2c). The alignments of the rupture
direction and of the slip direction along the dip of
the fault (Fig. 2d, e) generate a rupture directivity
effect at the stations located close to the surface trace
of the fault (Somerville et al. 1997; Somerville 2003).
Figure 2d, e illustrates this effect by showing low-
and high-frequency radiation patterns for an up-dip
rupture propagation in the case of a normal-faulting
event. The rupture directivity pulses in the case of
dip-slip faulting event are observed in the direction
normal to the dip of the fault and have components in
the vertical and horizontal directions normal to the
strike.
However, in case of dip-slip faulting events, the
fault plane normally forms an angle (d-dip angle)
with the horizontal plane (Fig. 2c). This geometrical
N. Poiata et al. Pure Appl. Geophys.
−150
0
150V
el. (
cm/s
)
0 5 10 15 20 25 30 35 40 45
Time (sec)
1992 Landers; station Joshua Tree; backward directivity
0
50
100
150
200
250
300
SV
(cm
/s),
(da
mpi
ng=
5%)
0 1 2 3 4 5 6 7 8 9 10
Period (sec)
observed
−150
0
150
Vel
. (cm
/s)
1992 Landers; station Lucerne; forward directivity
−150
0
150
Vel
. (cm
/s)
extracted pulse
−150
0
150
Vel
. (cm
/s)
0 5 10 15 20 25 30 35 40 45
Time (sec)
residual
0
50
100
150
200
250
300
SV
(cm
/s),
(da
mpi
ng=
5%)
0 1 2 3 4 5 6 7 8 9 10
Period (sec)
observed observedextracted pulseresidual
observed
(a) (b)
(c)
(d) (e)
13˚15'E 13˚30'E
42˚1
5'N
5 km
L’Aquila
Castelnuovo
Paganica
Onna
Mw 6.3
AQK
SP
SN
observedextracted pulseresidual
−50
−50
−50
0
50
Vel
. (cm
/s) 2009 L’Aquila; station AQK
0
Vel
. (cm
/s)
0
Vel
. (cm
/s)
0 5 10 15 20
Time (sec)
0
50
100
150
SV
(cm
/s),
(da
mpi
ng=
5%)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Period (sec)
extracted pulse
residual
observed
epicenter
rupture propagation
Lucerne
Joshua Tree
backward directivityregion
forward directivityregion
20 sec
Figure 3a Map of the Landers region indicating the causative faults of the 1992 Landers earthquake (modified from Somerville et al. 1997). The red
star shows the epicenter location and the black arrow the direction of rupture propagation. The fault-normal ground velocities for the Lucerne
forward directivity and the Joshua Tree backward directivity stations are also shown. b Left fault-normal component of the velocity record
(black trace) from Lucerne station recorded during the 1992 Landers earthquake. The extracted pulse (red trace) and the residual (gray trace)
waveforms are shown below. Right velocity response time-series spectra. c Left fault-normal component of the velocity record from the
Joshua Tree station recorded during the 1992 Landers earthquake. Right velocity response spectra of the corresponding time series. d Map of
the L’Aquila area indicating the location of the fault plane (black rectangle) ruptured during the 2009 L’Aquila earthquake. The red star
shows the location of the epicenter; the blue triangle shows the location of the near-fault hanging-wall station AQK. e Left strike normal
component of the velocity record (black) from the AQK station, recorded during the 2009 L’Aquila earthquake. The extracted pulse (red
trace) and the residual (gray trace) waveforms are shown below. Right velocity response spectra of the corresponding time series
Ground Motion Pulses for Dip-Slip Faulting
configuration generates an area on the free surface
that is located on the hanging-wall side of the fault
and above the ruptured fault plane. Considering the
3-D aspect of the S-wave radiation pattern (Fig. 2a,
b), a rupture propagating along the strike of a dip-slip
fault could affect the waveforms recorded at a station
on the hanging-wall and above the ruptured fault
plane. Figure 2f, g shows the low- and high-fre-
quency radiation patterns for along-strike rupture
propagation in case of a normal-faulting event. This
implies that pulse generation in a dip-slip faulting
event could be driven by a mechanism more complex
than the directivity effect, as it is described above.
Indeed, the record of pulse-like velocity ground
motions from near-fault stations located on the
hanging-wall side above the ruptured fault plane of
the recent 2009 L’Aquila (Mw 6.3) normal-faulting
event supports this idea. These stations were not
located near the surface trace of the fault, where the
forward directivity effect would be maximum for a
dip-slip fault. Furthermore, Kagawa (2009) made a
similar observation based on the record from the
Kashiwazaki-Kariwa Nuclear Power Plant during the
2007 Chuetsu-oki (Mw 6.6) reverse faulting
earthquake.
2. Methodology
2.1. Extraction of Long-Period Pulses
from the Observed Ground Motions
To demonstrate that near-fault ground motion
pulses can produce increased seismic demands on
structures, we illustrated their possible effect on the
spectral characteristics of the ground motions (implic-
itly their contribution to the seismic hazard). We
calculated the velocity response spectra of the fault-
normal components recorded at the two stations near
the causative fault of the 1992 Landers (Mw 7.3)
earthquake. The relative locations of the stations and
the fault are presented in Fig. 3a. For comparison, we
show both the record of the Lucerne station located in
the direction of the rupture propagation, which
recorded the corresponding long-period pulse (Fig. 3a,
b), and the backward directivity station of Joshua Tree
(Fig. 3a, c). These records are representative of the
effect of rupture directivity from a strike-slip fault,
illustrating both the forward and the backward direc-
tivity, on the near-fault ground motions (Somerville
et al. 1997). We decomposed the velocity record from
the Lucerne station into pulse and residual components
by applying the wavelet decomposition method of
Baker (2007) and Baker and Shahi (2007). Then, the
corresponding velocity response spectra were calcu-
lated (Fig. 3b) for all three resultant ground motions
and compared against each other. Same analysis for a
dip-slip faulting event is performed using the ground
motion of the 2009 L’Aquila (Mw 6.3) normal-faulting
event recorded at the AQK hanging-wall station. The
resulted decomposed and original observed velocity
response spectra for the two selected earthquakes are
discussed in Sect. 3.1.
2.2. Forward Simulation of Ground Motion Pulses
for Normal-Faulting Event
The 2009 L’Aquila, Italy (Mw 6.3) earthquake
was a first well-recorded normal-faulting event that
provided an unprecedented number of strong motion
observations in the near-fault area. The hypocenter of
the earthquake was located 2 km west of the city of
L’Aquila (e.g., Fig. 3d) at 42.3476�N, 13.3800�E and
a depth of 9.5 km (INGV 2009). In spite of its
moderate magnitude, this event caused extensive
damage to the cities and villages in the Abruzzo
region. Most of the damage was concentrated in the
city of L’Aquila, located close to the hypocentre, and
the villages of Castelnuovo, Onna, and Paganica
located further southeast (Fig. 3d). Thus, the 2009
L’Aquila earthquake offered an important case study
for the effects of damaging moderate-magnitude
earthquakes occurring in densely populated areas.
The observations of the 2009 L’Aquila mainshock
and its aftershocks recorded at the Italian strong
motion network (RAN), operated by the Civil Pro-
tection Department, and at MedNet station AQU have
been integrated into the ITalian ACcelerometric
Archive (ITACA). The high-quality digital acceler-
ation records of the mainshock included the near-
fault stations (at epicentral distances of 1.7–5.0 km)
on the hanging wall of the ruptured fault (Fig. 4a)
that recorded horizontal pick ground acceleration
(PGA) values of 0.35–0.65 g. The velocity records of
N. Poiata et al. Pure Appl. Geophys.
13˚24'E 13˚36'E
42˚1
2'N
42˚1
8'N
42˚2
4'N
5 km
AQGAQV
AQKAQU
GSA
L’Aquila
Onna
Paganica
Castelnuovo
0.0
0.2
0.4
0.6
0.8
1.0Slip (m)
Mw 6.3
(a) (b)EGFM modelscenario
w =
1.3
(km
)
l = 2.0 (km)(c)
(d)
(e)
(f)
Mw 4.6
Vp
Vs
Velocity (km/s)
0
5
10
15
20
25
30
Dep
th(k
m)
AQU
0
5
10
15
20
25
30
Dep
th(k
m)
Vp
Vs
Velocity (km/s)
GSA
Figure 4a Map of the L’Aquila area presenting the relative location of the surface projection of the rupture plane for the 2009 L’Aquila earthquake
(after Poiata et al. 2012) and the near-fault hanging-wall (AQG, AQV, AQU, and AQK) and footwall (GSA) stations. The red star indicates
the location of the epicenter reported by INGV. The red contoured black star indicates the location of the EGF event according to INGV. The
thick red line shows the location of the observed coseismic surface ruptures modified after Boncio et al. (2010), and the yellow squares mark
the cities that suffered the largest damage. The gray rectangle corresponds to the SMGA estimated from the EGF analysis of Poiata et al.
(2012). The areas green and blue rectangles correspond to the two rupture scenarios of the linear up-dip and the along-strike rupture
propagations used in the forward simulation analysis of the near-fault pulses. b–d Plane view of the fault settings for the rupture scenarios
used for the forward simulations. The rupture scenarios assumed different rupture starting points marked by gray stars. The white contoured
black star corresponds to the rupture starting point of the SMGA model (from Poiata et al. 2012). e SMGA model of Poiata et al. (2012). f 1-D
velocity models for the AQU and GSA stations used for low-frequency waveform modeling
Ground Motion Pulses for Dip-Slip Faulting
these near-fault stations showed clear short-duration
pulses with peak ground velocity (PGV) exceeding
30 cm/s (e.g., Fig. 3e).
The high socio-economic impact of the 2009
L’Aquila earthquake and the large number of high-
quality observational data produced by this event
attracted great interest from the geophysical commu-
nity. The process of the mainshock was studied by a
number of researchers providing a low-frequency
(\1.0 Hz) source model through the inversion of
geodetic and/or strong motion datasets (e.g., Atzori
et al. 2009; Walters et al. 2009; Cirella et al. 2009;
Poiata et al. 2012). These studies provided robust
general features of the 2009 L’Aquila mainshock
source model: a rupture on a southwest-dipping fault
of about 24 km by 16 km in size characterized by a
single main asperity located 8–10 km southeast of the
hypocentre (Fig. 4a). According to the broadband
source model of Poiata et al. (2012) that reproduced
the ground motions in the frequency range of
0.2–10 Hz, the earthquake ruptured a strong motion
generation area (SMGA), coincident with the area of
large slip (asperity) estimated from the low-frequency
source inversion and located beneath the highly
damaged city of Onna (Fig. 4a).
We made use of the available dataset of strong
motion recordings as well as the results of the
broadband source model (Fig. 4a) of Poiata et al.
(2012) to study the conditions that contributed to the
generation of the near-fault ground motion pulses
recorded at the stations located on the hanging wall of
the ruptured fault. The analysis consisted of perform-
ing forward simulations for different rupture
scenarios. The main objectives of our analysis were
as follows: (1) to compare the contributions of the
along-strike and up-dip rupture propagation to pulse
generation; (2) to determine the rupture configuration
(direction of rupture propagation and fault-station
geometry) that contributes positively to the pulse
generation; and (3) to identify the worst-case earth-
quake rupture scenario for the city of L’Aquila.
The assumed rupture scenarios were based on the
SMGA model of Poiata et al. (2012) illustrated in
Fig. 4 and estimated using the empirical Green’s
function (EGF) method (Irikura 1986). We used this
model to perform the forward simulations of the
waveforms for the near-fault strong motion stations in
both the broadband (0.2–10 Hz) and the low-fre-
quency (0.05–0.5 Hz) ranges. This option was
selected to account for the difference in the low-
and high-frequency radiation patterns discussed in the
previous section. The calculations were done for both
stations located on the hanging-wall side of the
causative fault above the ruptured fault plane (AQ*
stations, Fig. 4a) and the hanging-wall stations
located further away, as well as for the footwall
station GSA. This selection of stations should provide
more details on the spatial variability of the near-fault
ground motion pulses (fault-station geometry). Here,
we present only the results of an analysis on the
example AQU hanging-wall station and on the GSA
footwall station.
The broadband synthetic waveforms (0.2–10 Hz)
were calculated using the EGF method of Irikura
(1986). The method allows estimating the ground
motions of a target earthquake in broad frequency
range by using records of nearby smaller events,
which are treated as empirical Green’s functions
incorporating the properties of propagation path and
local site effects. This allows to bypass the high-
frequency ([1 Hz) limitation of the deterministic
approaches given by the uncertainties of the velocity
structure and complexity of source description. Based
on similarity relationship between the target and
small earthquakes and the omega-squared model of
the source spectra, the EGF method requires two
parameters N and C representing, respectively, the
ratios of the fault dimensions and the stress drop
between the events. Once these parameters are
estimated, the fault plane of the target event is
divided into N 9 N subfaults of sizes equivalent to
the rupture area of the element event, and the
synthetic waveforms for the target event are calcu-
lated as superposition of the recordings of the
element event scaled by the factor C and accounting
for the difference in the source-time functions
between the two events. The study of Poiata et al.
(2012) used Mw 4.6 aftershock that occurred on 7
April 2009 (21:34:29 UT, Fig. 4a) as the empirical
Green’s function (EGF) event for estimating the
parameters N and C by the spectral ratio fitting
method of Miyake et al. (1999), and providing the
model of SMGA of the 2009 L’Aquila earthquake
using the EGF method of Irikura (1986) discussed
N. Poiata et al. Pure Appl. Geophys.
previously. They provided a broadband (0.2–10 Hz)
source model for the 2009 L’Aquila earthquake
represented by a SMGA of 41.6 km2 (8.0 km in
length by 5.2 km in width) with a rise time of 0.8 s
and composed of 4 9 4 subfaults corresponding to
the SMGA of small (EGF) event (Fig. 4a, e). The
rupture starting point was located at the northwestern
part of SMGA implying a predominant southeast and
up-dip directions of rupture propagation (Fig. 4a, e).
The corresponding source parameters for small and
target events are summarized in Table 1. We use
these estimates and the same EGF event to perform
the forward simulations for the rupture scenarios
shown in Fig. 4b–d and compare them with the
SMGA model (Fig. 4e) of Poiata et al. (2012) as well
the observed records. For each of the rupture
scenarios, we assumed that the elementary subfault
corresponds to the fault size of the small event, and
the final source is composed by the superposition of
multiple elementary subfaults distributed according
to the designed geometry. The broadband synthetics
are then estimated by applying the EGF method
(Irikura 1986) for each of the scenarios. The assumed
source rupture scenarios were designed to allow
testing the contribution of the pure along-strike
(Fig. 4b) vs pure up-dip (Fig. 4c) rupture propagation
for the SMGA of the EGF model, the effect of the
different assumptions for the locations of the rupture
starting point (Fig. 4d) and the location of the rupture
area relative to the city of L’Aquila (Fig. 5a). For the
latter parameter, the AQU station was selected as
representative since no records were available from
the historical center of the city of L’Aquila (the most
damaged area of the city). The details of the assumed
rupture scenarios are summarized in Fig. 4.
The low-frequency (0.05–0.5 Hz) synthetic wave-
forms were determined using the extended reflectivity
method of Kohketsu (1985). The method allows a
deterministic calculation of the near-field seismo-
grams due to a finite fault (propagating rupture) in a
given 1-D velocity model. The finite fault is assumed
to be represented by a number of subfaults, each
corresponding to a point source with a given seismic
moment, mechanism, and rise time (s). The rupture
propagation is represented by a spherical rupture
front expanding at a speed Vr. For the calculation of
low-frequency synthetics, we used the same scenario
source models (based on SMGA of Poiata et al. 2012)
as for the broadband case (Fig. 4). We assumed that
each of the rupture fault’s subfaults represents the
EGF event (the Mw 4.6 aftershock on 7 April 2009,
21:34:29) with the size equivalent to that estimated
by Poiata et al. (2012) using the EGF method
(Table 1). The rise time for the ramp-type source-
time function and the seismic moment for each
subfault were estimated through low-frequency for-
ward modeling of the Mw 4.6 aftershock for the near-
fault hanging-wall stations shown in Fig. 4a assum-
ing that the source size of this event corresponds to
the subfault size. The results provided a seismic
moment M0 = 2.3 9 1015 Nm and a rise time
s = 1.0 s for the ramp-type source-time function.
The rupture propagation velocity was set to 2.9 km/s
(*0.85 VS), which is the same as that assumed for
the broadband modeling and derived from the EGF
analysis of Poiata et al. (2012). This value of lies in
the upper range of the commonly inferred earthquake
rupture velocities of 0.65–0.85 VS (e.g., Madariaga
1976). However, it is in good agreement with the near
shear wave speed of the rupture front propagation
Table 1
Source parameters of the EGF (small) event and the SMGA model used for the assumed rupture scenarios setup
C N EGF event
7 April 2009 21:34:29
SMGA (Poiata et al. 2012)
6 April 2009 01:32:40
Vr (km/s) Vs (km/s)
l (km) w (km) s (s) L (km) W (km) S (km)2 T (s)
3.5 4 2.0 1.3 0.2 8.0 5.2 41.6 0.8 2.9 3.5 Broadband model
NA NA 1.0 NA
M0 (Nm) Low-frequency model
2.3 9 1015
Ground Motion Pulses for Dip-Slip Faulting
13˚24'E 13˚36'E
42˚1
2'N
42˚1
8'N
42˚2
4'N
5 km
L’AquilaPaganica
Castelnuovo
0.0
0.2
0.4
0.6
0.8
1.0Slip(m)
AQU
Acc. (cm/s/s)
−1000
0
1000
AQU station, f = 0.2 - 10 Hz
0 5 10 15Time (sec)
Vel. (cm/s)
−80
0
80
0 5 10 15Time (sec)
Displ. (cm)
−20
0
20
0 5 10 15Time (sec)
Acc. (cm/s/s)
−1000
0
1000
0 5 10 15Time (sec)
Vel. (cm/s)
−80
0
80
0 5 10 15Time (sec)
Displ. (cm)
−20
0
20
0 5 10 15Time (sec)
EW
−0.35
0.00
0.35
0 5 10 15 20 25 30Time (sec)
NS
−0.35
0.00
0.35
0 5 10 15 20 25 30Time (sec)
UD
−0.35
0.00
0.35
0 5 10 15 20 25 30Time (sec)
AQU station, f = 0.05 - 0.5 Hz
syn. Case 1
syn. Case3
syn. Case 1
syn. Case 3
Case1
Case2
(a)
(c)
Case 2.
Case 3.
syn. Case 2
syn. Case 2
GSA
(d)
−0.07
0.00
0.07
0 5 10 15 20 25 30Time (sec)
−0.07
0.00
0.07
0 5 10 15 20 25 30Time (sec)
−0.07
0.00
0.07
0 5 10 15 20 25 30Time (sec)
EW NS UDGSA station, f = 0.05 - 0.5 Hz
Acc. (cm/s/s)
GSA station, f = 0.2 - 10 Hz
0 5 10 15Time (sec)
Vel. (cm/s)
0 5 10 15Time (sec)
Displ. (cm)
0 5 10 15Time (sec)
Acc. (cm/s/s)
0 5 10 15Time (sec)
Vel. (cm/s)
0 5 10 15Time (sec)
Displ. (cm)
0 5 10 15Time (sec)
−150
0
150
−10
0
10
−2.5
0.0
2.5
−150
0
150
GS
A
−10
0
10
−2.5
0.0
2.5
syn. Case 2
syn. Case 1
syn. Case 2
syn. Case 1
EW NS
EW NS
Case 1.
Case 2.
Case 3.
(b)
AQU
AQU / GSA
Onna
N. Poiata et al. Pure Appl. Geophys.
required for generation of the rupture directivity
effect (e.g., Somerville et al. 1997). This is appro-
priate for the purpose of the study. The station-
specific 1-D velocity models used in calculations of
low-frequency synthetics were extracted from Poiata
et al. (2012). All the finite fault parameters are
summarized in Table 1 and the 1-D velocity models
are presented in Fig. 4.
3. Results
3.1. Effect of Ground Motion Pulses on the Response
Spectra
We investigated the response spectra of the fault-
normal component and the extracted long-period
ground motion pulse recorded during the 1992
Landers (Mw 7.3) earthquake at Lucerne station
located in the direction of the rupture propagation. By
comparing the velocity response spectra of the
original record and the extracted pulse (Fig. 3), we
observed that the peaks at longer periods ([4 s)
originate from the pulse, whereas the response
spectrum of the residual waveform is roughly com-
parable to that of the station situated in the backward
direction of the rupture propagation (Fig. 3c). Similar
results were provided by the analysis of the strike
normal (SN) component of the velocity record from
the AQK hanging-wall station (Fig. 3d) recorded
during the 2009 L’Aquila, Italy (Mw 6.3) normal-
faulting event (Fig. 3e). In this case, we observed
significantly larger values of the *1.8 s-period
response spectra for the 2009 L’Aquila earthquake.
This interval also corresponds to the period range of
the pulse recorded at the AQK station (Fig. 3b).
These examples illustrate that near-fault ground
motion pulses can generate significantly larger
demands on structures than ordinary records. This
also underlines that the design criteria established
without considering such long-period near-fault
pulses will inevitably result in underestimation of
the seismic demands.
3.2. Conditions for the Generation of Near-Fault
Ground Motion Pulses: The 2009 L’Aquila
Earthquake
The results of the forward simulations for differ-
ent rupture scenarios indicated the following: (1) both
up-dip and along-strike rupture propagations con-
tribute to the generation of near-fault ground motion
pulses, although the up-dip rupture propagation
contribution is more significant, and (2) in a more
general case (Fig. 4d), for all stations (hanging wall
or footwall), the generation of pulses is conditioned
by rupture propagation toward the site. More partic-
ularly, the results of forward simulations for the pure
along-strike and pure up-dip rupture propagation
scenarios (Fig. 4b, c) show that for the near-fault
stations (both hanging wall and footwall) not located
above the ruptured fault plane (e.g., GSA station
Fig. 5d), both along-strike and up-dip rupture prop-
agations contribute to the generation of pulses.
Furthermore, scenarios testing the different rupture
initiation points for the SMGA (Fig. 4d) indicate that
the main condition for pulse generation at these
stations is that the direction of rupture propagation
should be toward the site. The analysis also points out
that in case of a more complex rupture configuration,
it is not possible to identify a predominant direction
of rupture propagation that will be more efficient in
generating the ground motion pulses at near-fault
stations not located above the ruptured fault plane.
Figure 5d illustrates this conclusion for the example
of the GSA footwall station.
Next, we discuss in more detail the results of the
forward simulations for the rupture scenarios, assum-
ing different rupture starting points (Fig. 5b, c) and
different locations of the SMGA for near-fault
stations positioned on the hanging wall above the
bFigure 5
a Map showing the relative location of the AQU and GSA stations
(blue triangles) and the rupture scenario corresponding to the
predominant along-strike rupture propagation (Cases 1 and 2;
rupture starting points are shown by the gray and green stars,
respectively) and the focusing effect (Case 3; rupture starting point
is shown by the brown star). b Vertical view of the rupture
scenarios. The stars correspond to the location of the rupture
starting points. The blue triangle shows the station location. c,
d Synthetic broadband and low-frequency waveforms for the
rupture scenarios of Case 1 (gray traces), Case 2 (green traces),
and Case 3 (brown traces) of the AQU and GSA stations,
respectively
Ground Motion Pulses for Dip-Slip Faulting
ruptured fault plane (AQ* stations from the Aterno
Valley).
3.2.1 Near-Fault Stations Located on the Hanging
Wall above the Ruptured Fault
Figure 5 shows examples of rupture scenarios assum-
ing the rupture of the entire SMGA and different
rupture starting points (Fig. 4d). These scenarios
result in different directions of rupture propagation
relative to the AQ* stations. We select two extreme
cases of these scenarios, noting them Case 1 and Case
2 (Fig. 5a, b). In Case 1, rupture propagates toward
the AQ* stations, and the fraction of the ruptured area
of the fault between the hypocenter and the stations is
maximum; while in Case 2, rupture propagates away
from the stations, and the fraction of the ruptured area
of the fault between the hypocenter and the stations is
minimum. The results of the forward simulations for
the broadband and low-frequency ranges (Fig. 5c)
show that near-fault ground motion pulses are
efficiently generated for the scenarios implying the
propagation of rupture toward the stations. We also
confirm that the amplitude of the pulses, for both the
broadband and the low-frequency components at
station AQU, is maximum when the rupture starting
point is located in the southeast end of the fault (Case
1, Fig. 5c). This scenario includes the along-strike
(predominant) and the up-dip propagations of rupture
in the direction of the station. We will refer to this
rupture scenario as ‘‘rupture propagation toward the
site’’, in order to distinguish it from the rupture
directivity effect.
The rupture scenarios assuming different loca-
tions of the SMGA relative to the AQ* stations are
examined to determine the worst-case scenario of
pulse generation for the city of L’Aquila and compare
it to the SMGA model (Fig. 4e) based on the EGF
simulation of Poiata et al. (2012), as well as observed
ground motions. Here, we focus on the synthetic
waveforms at the AQU station. We identify two
possible locations of the rupture area and the rupture
starting point relative to the AQU station that result in
pulse generation for both the broadband and low-
frequency synthetic waveforms. These rupture sce-
narios are summarized in Fig. 6, showing their
location relative to the AQU station. The first
scenario (Case 1) corresponds to the rupture propa-
gation toward the station (Fig. 5a–c, Case 1), whereas
the second scenario (Figs. 5a–c, 6, Case 3) represents
the case when the station is located above the
ruptured fault plane and the distance to the rupture
starting point is minimum (Fig. 5b, right). This case
corresponds to the ‘‘focusing effect’’ discussed by
Kagawa (2009). The effect is caused by the simul-
taneous arrival of the rupture front from different
points of the fault, which are at the same distance (on
the same isochrones) from the rupture starting point
(Fig. 7b). The synthetic broadband and low-fre-
quency waveforms for Cases 1 and 3 are presented
in Figs. 5c and 6a. Rupture propagation toward the
site (Case 1) and the focusing effect (Case 3) generate
ground motion pulses significantly exceeding the
observations as well as the simulation results for the
SMGA based on the EGF model of Poiata et al.
(2012) (Fig. 6a). It can be noticed from Fig. 6a that
the SMGA model of Poiata et al. (2012) reproduces
well the ground motion pulses recorded at the AQU
station during the 2009 L’Aquila earthquake. The
comparison of the velocity response spectra for the
observed and synthetic waveforms (Fig. 6c, d) indi-
cates that in the period 2.0–3.5 s, in both cases, the
spectral ordinates of these observed ground motions
are significantly below the level provided by the two
scenarios of Cases 1 and 3. Moreover, in Case 3, the
spectral values of synthetics are significantly higher
than the observations for the period range of
1.0–1.3 s (Fig. 6c), corresponding to the resonance
period of typical residential buildings in the area.
These results point out the important role that the
direction of rupture propagation and the location of
the SMGA relative to the hanging-wall sites play in
generation of the near-fault ground motion pulses in
dip-slip faulting events.
4. Discussion and Conclusions
In this study, we demonstrated that long-period
near-fault ground motions pulses observed at near-
source regions can strongly affect the duration and
spectral characteristics of the observed ground
motions. Thus, the design criteria established without
considering such ground motion pulses will
N. Poiata et al. Pure Appl. Geophys.
EW componentAcc. (cm/s/s)
−1000
0
1000
0 5 10 15Time (sec)
Vel. (cm/s)
−80
0
80
0 5 10 15Time (sec)
Displ. (cm)
−20
0
20
0 5 10 15Time (sec)
NS componentAcc. (cm/s/s)
−1000
0
1000
0 5 10 15Time (sec)
Vel. (cm/s)
−80
0
80
0 5 10 15Time (sec)
Displ. (cm)
−20
0
20
0 5 10 15Time (sec)
obs.
syn. EGF
syn. Case 1
syn. Case 3
(a)
(c)
(d)
238.0 20.5 3.8
280.2 17.5 3.7
327.1 31.3 12.1
623.0 46.6 14.7
289.4 28.9 7.7
269.0 23.2 6.0
340.5 37.8 12.0
1143.0 78.8 13.9
EW obs.NS obs.EW syn.3NS syn.3
Case 3
0
100
200
300
400
SV
(cm
/s),
(da
mpi
ng=
5%)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Period (sec)
EW syn.EGFNS syn.EGF
Case1
0
100
200
300
400
SV
(cm
/s),
(da
mpi
ng=
5%)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Period (sec)
EW obs.NS obs.EW syn.1NS syn.1EW syn.EGFNS syn.EGF
13˚24'E 13˚36'E
42˚1
2'N
42˚1
8'N
42˚2
4'N
5 km
Onna
Paganica
Castelnuovo
0.0
0.2
0.4
0.6
0.8
1.0Slip(m)
AQUL’Aquila
Case1
Case 3
EGF
(b)
Figure 6a Comparison of the observed (black traces) acceleration, velocity, and displacement waveforms for the horizontal components of the
broadband record of the 2009 L’Aquila mainshock from the AQU station, and the synthetics calculated for the SMGA by the EGF simulation
of Poiata et al. (2012; red traces), the rupture scenario of the rupture propagation towards the AQU station (Case 1; gray traces), and the
rupture scenario corresponding to the case of the focusing effect (Case 3; brown traces). The numbers above the waveforms correspond to the
maximum amplitudes. b Map of the near-fault area of the 2009 L’Aquila earthquake showing the relative location of the AQU station (blue
triangle), the city of L’Aquila, and the rupture areas for the assumed scenario and EGF model (gray rectangles). The stars correspond to the
location of the rupture starting points. c 5% damped velocity response spectra comparing the observed horizontal components of the AQU
record (in black), and d the synthesis calculated for the two cases of the rupture scenarios and the EGF model (red traces)
Ground Motion Pulses for Dip-Slip Faulting
inevitably result in underestimation of seismic
demands in building structures.
The forward rupture directivity effect is generally
considered the main cause of the near-fault ground
motion pulses observed in cases of strike-slip and
dip-slip faulting. The conditions for the rupture
directivity effects summarized by Somerville et al.
(1997) provided a good explanation of the fault-
normal components observed during the past strike-
slip faulting events (e.g., 1992 Landers). In the case
of dip-slip faulting, however, both the near-fault
ground motion pulses and the hanging-wall effect
(Abrahamson and Somerville 1996) are observed at
some stations. The inspection of the 3-D S-wave
radiation patterns for dip-slip faulting, as well as the
near-fault strong ground motion records from the
stations on the hanging-wall sites above the causative
fault of the 2007 Chuetsu-oki (reverse faulting), the
2009 L’Aquila (normal faulting) and the 2015 Gor-
kha (low-angle dip-slip faulting) earthquakes point
toward a more complex mechanism of pulse gener-
ation for such type of events.
We also inspected the main aspects of the rupture
configuration that contribute to the generation of the
near-fault ground motion pulses for a dip-slip faulting
event by performing the forward simulations of the
different rupture scenarios for the 2009 L’Aquila,
Italy, (Mw 6.3) earthquake. We examined rupture
scenarios based on the broadband source model of the
2009 L’Aquila earthquake determined by Poiata et al.
(2012) and performed forward simulations of the
waveforms for the near-fault strong motion stations in
both broadband and low-frequency ranges. The
results of our analysis indicate that the generation
mechanism of the near-fault ground motion pulses in
the case of dip-slip faulting depend on both the rup-
ture configuration and the location of the site relative
to the fault plane. We confirmed that the rupture
directivity effect is predominant in stations located on
the footwall of the causative fault. In all hanging-wall
stations, pulse generation occurs as the result of the
radial rupture propagation toward the site (Fig. 7a).
However, in the stations located on the hanging wall
above the ruptured fault plane, where the hanging-
wall effect is observed (Abrahamson and Somerville
1996), the focusing effect (Kagawa 2009) could be an
alternative mechanism of pulse generation. Thus, our
results indicated that that the actual 2009 L’Aquila
earthquake did not constitute a worst-case scenario
for the city of L’Aquila regardless of its proximity to
the epicenter. We confirmed that, in the case of dip-
slip faulting earthquakes, along-strike propagation
toward the site and the focusing effect (Fig. 7) are the
main candidates for the worst-case rupture scenarios
of pulse generation for the hanging-wall sites located
above the area of large slip (strong motion generation
area; Fig. 6).
Acknowledgements
The authors would like to thank Luis Dalguer and
two anonymous reviewers for their help in improving
the manuscript. We would also like to express our
gratitude to the Italian Department of Civil Protection
and the Istituto Nazionale di Geofisica e
AQU
AQU
δ
hanging wallfoot wall
(a) (b)
Figure 7Schematics of a the radial rupture propagation toward the site and b the focusing effect
N. Poiata et al. Pure Appl. Geophys.
Vulcanologia of Italy for providing free access to
seismological data and the strong ground motion
records of the 2009 L’Aquila earthquake and its
aftershocks. The original near-field acceleration
waveforms used in the study were downloaded from
the website of the database of the Italian strong
motion records (http://itaca.mi.ingv.it) maintained by
the Working Group ITACA. Some figures were made
using the Generic Mapping Tools software (Wessel
and Smith 1995).
REFERENCES
Abrahamson, N. A., & Somerville, P. G. (1996). Effects of the
hanging wall and footwall on ground motions recorded during
the Northridge earthquake. Bulletin of the Seismological Society
of America, 86, S93–S99.
Aki, K. (1968). Seismic displacement near a fault. Journal of
Geophysical Research, 73, 5359–5376.
Aki, K., & Richards, P. G. (1980). Quantitative seismology (p.
932). San Francisco: W. H. Freeman.
Atzori, S., Hunstad, I., Chini, M., Salvi, S., Tolomei, C., Bignami,
C., et al. (2009). Finite fault inversion of DInSAR coseismic
displacement of the 2009 L’Aquila earthquake (central Italy).
Geophysical Research Letters, 36, L15305. doi:10.1029/
2009GL039293.
Baker, J. W. (2007). Quantitative classification of near-fault ground
motions using wavelet analysis. Bulletin of the Seismological
Society of America, 97, 1486–1501.
Baker, J.W. & Shahi, S. (2007). Quantitative classification of near-
fault ground motions using Wavelet analysis (supporting soft-
ware), http://www.stanford.edu/*bakerjw/pulse-classification.
html. Accessed Apr 2017.
Boncio, P., Pizzi, A., Brozzetti, F., Pomposo, G., Lavecchia, G., Di
Naccio, D., et al. (2010). Coseismic ground deformation of the 6
April 2009 L’Aquila earthquake (central Italy, Mw6.3). Geo-
physical Research Letters, 37, L06308. doi:10.1029/
2010GL042807.
Boore, D. M., & Joyner, W. B. (1978). The influence of rupture
incoherence on seismic directivity. Bulletin of the Seismological
Society of America, 68, 283–330.
Boore, D. M., & Joyner, W. B. (1989). The effect of directivity on
the stress parameter determined from ground motion observa-
tions. Bulletin of the Seismological Society of America, 79,
1984–1988.
Cirella, A., Piatanesi, A., Cocco, M., Tinti, E., Scognamiglio, L.,
Michelini, A., et al. (2009). Rupture history of the 2009 L’Aquila
(Italy) earthquake from non-linear joint inversion of strong
motion and GPS data. Geophysical Research Letters, 36,
L19304. doi:10.1029/2009GL039795.
Heaton, T. H. (1990). Evidence for and implications of self-healing
pulses of slip in earthquake rupture. Physics of the Earth and
Planetary Interiors, 64, 1–20.
Heaton, T. H., Hall, D. F., Wald, D. J., & Halling, M. W. (1995).
Response of high-rise and base-isolated buildings to a hypo-
thetical M 7.0 blind thrust earthquake. Science, 267, 206–211.
Irikura, K. (1986). Prediction of strong acceleration motions using
empirical Green’s function. In Proceedings of 7th Japan Earth-
quake Engineering symposium (pp. 151–156).
Istituto Nazionale di Geofisica e Vulcanologia (INGV) (2009), The
L’Aquila seismic sequence—April 2009, Italian Accelerometric
Archive (ITACA). Available at: http://itaca.mi.ingv.it/ItacaNet/.
Accessed Feb 2015.
Joyner, W. B. (1991). Directivity for nonuniform ruptures. Bulletin
of the Seismological Society of America, 81, 1391–1495.
Kagawa, T. (2009). Conditions of fault rupture and site location
that generate damaging pulse waves. In Proceedings of the 6th
International Conference on Urban Earthquake Engineering (pp.
53–58).
Kasahara, K. (1981). Earthquake mechanics, (p. 248), Cambridge:
Cambridge University Press.
Kohketsu, K. (1985). The extended reflectivity method for syn-
thetic near-field seismograms. Journal of Physics of the Earth,
33, 121–131.
Koketsu, K., & Miyake, H. (2008). A seismological overview of
long-period ground motion. Journal of Seismology, 12,
133–143.
Koketsu, K., Miyake, H., Guo, Y., Kobayashi, H., Masuda, T.,
Davuluri, S., et al. (2016). Widespread ground motion distribu-
tion caused by rupture directivity during the 2015 Gorkha, Nepal
earthquake. Scientific Reports, 6, 28536. doi:10.1038/srep28536.
Lay, T., & Wallace, T. C. (1995). Modern global seismology (p.
521). San Diego: Academic Press.
Liu, H., & Helmberger, D. V. (1985). The 23:19 aftershock of the
15 October 1979 Imperial valley earthquake: more evidence for
an asperity. Bulletin of the Seismological Society of America, 75,
689–708.
Madariaga, R. (1976). Dynamics of an expanding circular fault.
Bulletin of the Seismological Society of America, 66,
163–182.
Miyake, H., Iwata, T., & Irikura, K. (1999). Strong ground motion
simulation and source modeling of the Kagoshima-ken Hoku-
seibu earthquakes of March 26 (MJMA 6.5) and May 13 (MJMA
6.3), 1997, using empirical Green’s function method. Zisin, 51,
431–442. (in Japanese with English abstract).
Miyake, H., Koketsu, K., Hikima, K., Shinohara, M., & Kanazawa,
T. (2010). Source fault of the 2007 Chuetsu-oki, Japan, earth-
quake. Bulletin of the Seismological Society of America, 100,
384–391.
Miyatake, T. (1998). Generation mechanism of strong ground
motion pulse near the earthquake fault. Zisin, 2(51), 161–170. (in
Japanese with English abstract).
Poiata, N., Koketsu, K., Vuan, A., & Miyake, H. (2012). Low-
frequency and broadband source models for the 2009 L’Aquila,
Italy, earthquake. Geophysical Journal International, 191,
224–242.
Somerville, P. G. (2003). Magnitude scaling of the near fault
rupture directivity pulse. Physics of the Earth and Planetary
Interiors, 137, 201–212.
Somerville, P. G., Smith, N. F., Graves, R. W., & Abrahamson, N.
A. (1997). Modification of empirical strong ground motion
attenuation relations to include the amplitude and duration
effects of rupture directivity. Seismological Research Letters, 68,
199–222.
Strasser, F. O., & Bommer, J. J. (2009). Review: strong ground
motions—have we seen the worst? Bulletin of the Seismological
Society of America, 99, 2613–2637.
Ground Motion Pulses for Dip-Slip Faulting
Takemura, S., Furumura, T., & Saito, T. (2009). Distortion of the
apparent S-wave radiation pattern in the high-frequency wave-
field: Tottori-Ken Seibu, Japan, earthquake of 2000. Geophysical
Journal International, 178(2), 950–961.
Takenaka, H., Mamada, Y., & Futamure, H. (2003). Near-source
effect on radiation pattern of high-frequency S waves: strong
SH–SV mixing observed from aftershocks of the 1997 North-
western Kagoshima, Japan, earthquakes. Physics of the Earth and
Planetary Interiors, 137, 31–43.
Walters, R. J., Elliott, J. R., D’Agostino, N., England, P. C.,
Hunstad, I., Jackson, J. A., et al. (2009). The 2009 L’Aquila
earthquake (central Italy): a source mechanism and implications
for seismic hazard. Geophysical Research Letters, 36, L17312.
doi:10.1029/2009GL039337.
Wessel, P., & Smith, W. H. F. (1995). New version of the generic
mapping tools released, EOS. Transactions American Geophys-
ical Union, 76, 329.
(Received July 30, 2016, revised March 23, 2017, accepted March 24, 2017)
N. Poiata et al. Pure Appl. Geophys.