retrieving impulse response function amplitudes from the · 2017-01-20 · 1 retrieving impulse...
TRANSCRIPT
submitted to Geophys. J. Int.
Retrieving impulse response function amplitudes from the1
ambient seismic field2
Loıc Viens1, Marine Denolle1, Hiroe Miyake2,3, Shin’ichi Sakai2 and Shigeki Nakagawa2
1 Earth and Planetary Sciences, Harvard University, Cambridge, MA, USA
E-mail: [email protected]
2 Earthquake Research Institute, University of Tokyo, Tokyo, Japan.
3 Center for Integrated Disaster Information Research, Interfaculty Initiative in Information Studies,
University of Tokyo, Tokyo, Japan
3
Received ...; in original form ...4
Key words: Seismic interferometry; Seismic noise; Surface waves and free oscillations; Wave5
propagation; Body waves; Earthquake ground motions6
2 L. Viens et al.
SUMMARY7
8
Seismic interferometry is now widely used in passive seismology to retrieve the impulse re-9
sponse of the Earth between two distant seismometers. The phase information has been the fo-10
cus of most studies, as conventional seismic tomography uses travel-time measurements. The11
amplitude information, however, is harder to interpret because it has been argued to strongly12
depend on the distribution of ambient seismic field sources and on the multitude of processing13
methods. Our study focuses on the latter by comparing the amplitudes of the impulse responses14
calculated between seismic stations in the Kanto sedimentary basin, Japan, using several pro-15
cessing techniques. This region provides a unique natural laboratory to test the reliability of16
the amplitudes with dense instrumentation, notably the Metropolitan Seismic Observation net-17
work (MeSO-net), and complex wave propagation with strong attenuation and basin amplifi-18
cation. We compute the impulse response functions using the cross correlation, coherency and19
deconvolution techniques of the raw ambient seismic field and the cross correlation of 1-bit20
normalized data. To validate the amplitude of the impulse response functions, we use a shal-21
low Mw 5.8 earthquake which occurred on the eastern edge of Kanto Basin close to a station22
that is used as a virtual source. Both S and surface waves are retrieved in the causal part of the23
impulse responses computed with the different techniques, but their amplitudes agree better24
with those of the earthquake with the deconvolution method. The amplification related to the25
structure of the Kanto Basin, which overcomes the attenuation, is clearly retrieved from the26
ambient seismic field. To test whether or not the anticausal part of the impulse response from27
deconvolution also contains reliable amplitude information, we use a virtual source on the28
western edge of the basin and show that their surface-wave amplitudes agree well with those29
of a nearby shallow Mw 4.7 event. This study demonstrates that the deconvolution technique30
is the best strategy to retrieve reliable relative amplitudes from the ambient seismic field in the31
Kanto Basin and can be used to simulate earthquake ground motions.32
Retrieval of impulse response function amplitudes 3
1 INTRODUCTION33
Over the past decade, new opportunities have emerged in seismology through ambient seismic field34
seismic interferometry. Shapiro & Campillo (2004) empirically showed that the impulse response35
function (IRF) of the Earth can be retrieved by cross correlating ambient seismic field time series36
recorded by two distant seismometers. Under certain distribution of ambient seismic field sources,37
the IRF of the medium retrieved by cross correlation yields the true Green’s function (Snieder38
2004; Wapenaar 2004). The ambient seismic field on Earth is, however, mainly excited by the39
interaction of oceanic waves with the seafloor and the shores (Longuet-Higgins 1950), which leads40
to an uneven distribution of the ambient seismic field sources that potentially biases both phase41
and amplitude information of the IRFs.42
The travel-time information of the IRF has been, up to now, the primary focus of passive43
seismology. To reduce the biases caused by the non-uniform distribution of ambient seismic field44
sources, pre-processing of the raw ambient seismic field (1-bit normalisation, pre-whitening) is45
routinely performed before computing the cross correlations (Bensen et al. 2007). The travel-time46
measurements of the pre-processed IRFs have been used to image the Earth structure at local (Lin47
et al. 2013; Mordret et al. 2014; Roux et al. 2016), regional (Shapiro et al. 2005; Lin et al. 2008;48
Nishida et al. 2008), and global (Boue et al. 2013) scales, to estimate seismic anisotropy (Mordret49
et al. 2013; Takeo et al. 2013), monitor velocity changes in volcanoes (Brenguier et al. 2008b),50
and study the response of the crust after a large earthquake (Brenguier et al. 2008a, 2014).51
The amplitude information, in contrast, has attracted less attention because it appears to be52
more affected by the intensity and directionality of the ambient seismic field. Numerical (Cupil-53
lard & Capdeville 2010; Lawrence et al. 2013) and theoretical (Weaver 2011; Tsai 2011) studies54
predict a sensitivity of the amplitudes to the distribution of ambient seismic field sources as well55
as to the data processing. Despite this, several empirical studies demonstrated that the amplitude56
information seems to be preserved and can be used to simulate the long-period ground motions57
of moderate Mw 4–5 (Prieto & Beroza 2008; Denolle et al. 2013; Viens et al. 2015) and large58
Mw > 6.5 (Denolle et al. 2014a; Viens et al. 2016b) earthquakes, map site amplification (Denolle59
4 L. Viens et al.
et al. 2014b; Bowden et al. 2015) and infer seismic attenuation (Prieto et al. 2009; Lawrence &60
Prieto 2011).61
The amplitude of seismic waves can be affected by several effects: the elastic effects, which62
are the geometrical spreading, multipathing (focusing and defocusing) and scattering effects, and63
the anelastic effects, which are also called intrinsic attenuation. These processes are particularly64
efficient in complex shallow crustal structures, such as sedimentary basins. The low seismic-wave65
velocity of the soft sediments that compose basin and the shape of the basement both contribute66
to the trapping of seismic waves and thus their amplification. The Kanto Basin, Japan, is a sed-67
imentary basin that is locally deeper than 4 km and is located beneath the Tokyo Metropolitan68
area. The Japan Integrated Velocity Structure Model (JIVSM) (Koketsu et al. 2008, 2012) shows69
that the basin is constituted of sediments that have S-wave velocities ranging between 0.5 and 1.570
km/s and that lie over a stiff bedrock of S-wave velocity equal to 3.2 km/s. The contours of the71
basement are shown in Figure 1. This complex and deep structure has a resonance period of 6–1072
s (Kudo 1978, 1980; Furumura & Hayakawa 2007; Denolle et al. 2014b) and the seismic waves73
amplified by the basin are a potential threat to the large scale urban structures, such as high-rise74
buildings or long-span bridges, of the Tokyo Metropolitan area.75
The Tokyo Metropolitan area is extremely well instrumented. The dense Metropolitan Seis-76
mic Observation network (MeSO-net), which is composed of 296 accelerometer stations, was77
deployed in shallow 20-m deep boreholes (Kasahara et al. 2009; Sakai & Hirata 2009) to assess78
the seismic hazard in the region. There are also dozens of instruments of the Hi-net of the National79
Research Institute for Earth Science and Disaster Resilience (NIED) (Okada et al. 2004; Obara80
et al. 2005), University of Tokyo, and Japan Meteorological Agency (JMA) networks (Figure 1).81
All stations record continuous signals that contain the waveforms of several moderate Mw 4–582
earthquakes that occurred close to seismic stations (epicentral distances < 10 km). The ambient83
seismic field recorded by the stations has been used in several seismic interferometry studies to84
infer the complexity of the wave propagation and the seismic amplification in the basin (Denolle85
et al. 2014b; Viens et al. 2016a; Boue et al. 2016). Therefore, the combination of the large number86
of stations, the complexity of wave propagation, and the numerous shallow earthquakes makes the87
Retrieval of impulse response function amplitudes 5
area the ideal natural laboratory to investigate the recovery of reliable amplitudes from the ambient88
seismic field.89
In this study, we aim to determine the best processing strategy to retrieve reliable IRF ampli-90
tudes from the ambient seismic field. After introducing the different processing techniques, we91
compare the waveforms of the vertical-to-vertical IRFs computed using one virtual source station92
located on the eastern edge of the basin. To validate the surface-wave amplitude of the IRFs, we93
compare their long-period peak ground velocities (PGVs) with the ones of a shallow earthquake94
that occurred in the vicinity of the virtual source. Finally, we discuss the effect of seasonal varia-95
tions, the retrieval of realistic S-wave amplitudes, and the possibility of using the anticausal part96
of the IRFs using one virtual source located on the western edge of the basin.97
2 METHODS: CROSS CORRELATION, COHERENCY, AND DECONVOLUTION98
We use the ambient seismic field data recorded in October 2014 by the seismic stations located99
close to the surface of the Kanto Basin. After instrumental response correction, the acceleration100
records of the MeSO-net stations are integrated once in time to retrieve the corresponding velocity101
waveforms and the data-set is divided into 1-hour time windows. Waveforms are band-pass filtered102
between 0.05 and 2 Hz (0.5–20 s) and down-sampled to 4 Hz to speed up the computation process.103
To reduce the bias that might be caused by earthquakes (Bensen et al. 2007), we remove the104
windows that have spikes larger than 10 times the standard deviation of each time window. We105
finally compute the Fourier transform of each 1-hour non-overlapping time series after using a106
zero-padding of 10 times the length of the window to increase the resolution when performing the107
computation in the frequency domain.108
For each pair of virtual source-receiver stations, we compute the IRFs using three different109
techniques. The first method is the cross correlation and can be written as110
CZZ(xr, xs, t) =⟨F−1 (vZ(xr, ω)v
∗Z(xs, ω))
⟩, (1)
where vZ(xr, ω) and v∗Z(xs, ω) are the Fourier transformed time series recorded by the vertical111
6 L. Viens et al.
component Z at the receiver station xr and the virtual source xs. The ∗ symbol is the complex112
conjugate and ω is the frequency. The inverse Fourier transform, denoted by F−1, of the cross113
correlation is computed for each 1-hour time window to retrieve the waveforms in the time domain114
t. We finally stack the cross correlations over 1 month, represented by the 〈〉 brackets, to increase115
the signal to noise ratio (i.e., equivalent to spatially averaging the effect of the ambient seismic116
field sources). For the cross correlation, we use both raw and 1-bit normalized ambient field data.117
The 1-bit normalization is a widely used technique (Campillo & Paul 2003; Shapiro & Campillo118
2004; Bensen et al. 2007) that only retains the sign of the data, setting the positive values to 1 and119
the negative ones to −1.120
The second technique is the coherency. It is similar to pre-whitening in the frequency do-121
main (Bensen et al. 2007), and consists of using the amplitude spectrum of the raw data from122
both stations as a denominator term when performing the cross correlation. This process can be123
summarized as124
CoZZ(xr, xs, t) =
⟨F−1
(vZ(xr, ω)v
∗Z(xs, ω)
{|vZ(xs, ω)|}{|vZ(xr, ω)|}
)⟩, (2)
where | | represents the absolute value and {} denotes a smoothing of the spectra using a moving125
average over 20 points to stabilize the denominator terms. All the other notations are the same as126
in Equation 1.127
The third method is the deconvolution, where only the amplitude spectrum of the data recorded128
by the virtual source is used in the denominator, and can be written as129
DZZ(xr, xs, t) =
⟨F−1
(vZ(xr, ω)v
∗Z(xs, ω)
{|vZ(xs, ω)|}2
)⟩, (3)
where all the notations are the same as in Equations 1 and 2.130
The travel-time (or phase) information is preserved through all three techniques as the denom-131
inator term in Equations 2 and 3 only affects the amplitude information. The regularization of the132
amplitude with the denominator term, however, prevents us from directly comparing the ampli-133
Retrieval of impulse response function amplitudes 7
tudes from these three techniques without any scaling. In the following, we either normalize the134
amplitudes or calibrate them with earthquake records before comparing the waveforms.135
3 EFFECTS OF THE PROCESSING ON THE IMPULSE RESPONSE FUNCTIONS136
We use the techniques described in Section 2 to retrieve IRFs between the TENNOD station,137
which is the virtual source, and receivers in the basin. The virtual source station has been chosen138
because of its location close to the epicentre of a shallowMw 5.8 earthquake, which occurred in 14139
March 2012 at a depth of approximately 11 km (F-net/NIED catalog), that is used to validate the140
amplitude of the IRFs in Section 4. In the Kanto Basin, the primary source of the ambient seismic141
field is the Pacific Ocean. Such distribution of ambient seismic field sources and the location of142
the virtual source and receiver stations both lead to a strong asymmetry between the causal and143
anticausal parts of the IRFs. As the signal of the anticausal part is very weak, we only focus on the144
causal part of the IRFs for the TENNOD virtual source.145
Figure 2 shows the four waveforms, which are normalized to their respective peak absolute146
value, computed between the TENNOD station and the E.SRTM and E.YMMM receivers (lo-147
cations shown in Figure 1). In the 1–10 s period range, several groups of waves, which can be148
explained by the seismic wave propagation in the layered Kanto Basin (Boue et al. 2016; Viens149
et al. 2016a), can be observed. The earliest group of waves arriving at 15 and 25 s for the E.SRTM150
and E.YMMM stations, respectively, has been identified as S waves by Viens et al. (2016a) and151
is seen in the four waveforms. However, differences in the amplitudes are observed among the152
waveforms, with smaller amplitudes retrieved with the deconvolution method. Nevertheless, the153
recovery of clear body waves from seismic interferometry without any additional computation re-154
mains unusual and has only been observed in a few studies (Roux et al. 2005; Poli et al. 2012;155
Boue et al. 2013).156
The second and third wave packets in Figure 2 are likely surface waves. The first group is well157
retrieved by all the methods, but its duration is shorter for the IRFs retrieved with the raw cross158
correlation. The second group is very weak for the waveforms obtained from raw and 1-bit cross159
correlations compared to the two other techniques. To explain this weak signal, we compute the160
8 L. Viens et al.
Fourier spectra of the IRFs retrieved with the cross correlation and deconvolution techniques of161
raw data for 28 stations in Figure 3. These receivers are located within the azimuth range of 274±162
4◦, between 15 and 100 km from the virtual source. The IRFs obtained by cross correlation have163
a narrower frequency content (0.18 to 0.3 Hz), which corresponds to the frequency range of the164
secondary microseism peak, compared to that of the deconvolution (0.1 to 1 Hz). Vasconcelos &165
Snieder (2008) demonstrated that this discrepancy between the two techniques is expected because166
the dependence on the ambient seismic field sources is stronger for the cross correlation than167
for the deconvolution. Therefore, the last group of surface-wave arrivals, which has a dominant168
period content of 1 to 3 s (Viens et al. 2016a), can only be retrieved by both the coherency and169
deconvolution methods that generate a broader frequency content.170
To study the effect of the processing on the reliability of the amplitudes, we assume that the171
dominant amplitude in the waveforms in sedimentary basins comes from the surface wave. We172
band-pass filter the IRFs at the same stations as in Figure 3 between 3 and 10 s. The two main173
reasons of the narrower band-pass filter are the following: first, the cross correlation can only174
retrieve waveforms at periods longer than 3 s; and second, Viens et al. (2016a) demonstrated that175
both phases and amplitudes of the Z–Z (vertical-to-vertical) IRFs between 3 and 10 s are highly176
similar to those of theMw 5.8 earthquake. Then, we correct their relative amplitude for the surface-177
wave geometrical spreading (1/√xs, where xs is the virtual source–receiver distance) and compute178
the envelope of the signals using the Hilbert transform. Finally, the waves travelling faster than 5179
km/s are probably non-physical and are tapered out. Figure 4 shows the envelopes, which are180
normalized by a single calibration factor taken as the peak amplitude at the station the closest to181
the virtual source, against the distance from the virtual source TENNOD. Despite the geometrical182
spreading correction, the amplitudes of the first surface-wave group generally decays with the183
increasing distance, a likely signature of the attenuation in Kanto Basin. However, remarkable184
local amplifications can be observed at some stations, as for example, at the two stations located185
between 60 and 70 km from the virtual source. We can also note that the amplitude of S waves186
retrieved with the cross correlation of raw and 1-bit data is higher than that of the two other187
techniques for most of the stations. Despite the surface-wave geometrical spreading correction,188
Retrieval of impulse response function amplitudes 9
the amplitude of S waves decreases faster than that of the surface waves, which validates that the189
long-period surface waves are dominant in the Kanto Basin.190
4 VALIDATION AGAINST EARTHQUAKE RECORDS: ATTENUATION AND191
AMPLIFICATION192
We showed in Section 3 that the dominant signal is contained by the first group of surface-wave193
arrivals. To validate the amplitude of this group of waves, we compare their long-period PGVs194
with those of the Mw 5.8 earthquake. However, only the relative amplitude is preserved by seismic195
interferometry techniques and the IRFs must be calibrated against the earthquake velocity records.196
To scale up the amplitude of the IRFs to that of the earthquake, we calculate the ratio of earthquake197
to IRFs long-period PGV at each selected station, compute the mean of the ratio, and correct the198
IRFs by this factor. This correction is the same for the selected stations but is different for the cross199
correlation, coherency and deconvolution techniques. We also account for the difference between200
the epicentre and virtual source locations, which leads to a difference in surface-wave geomet-201
rical spreading. We effectively shift the virtual source to the earthquake epicentre by correcting202
the amplitudes with the ratio of the geometrical spreading terms√xs/√x, where xs is the vir-203
tual source–receiver distance and x is the epicentre–receiver distance. This correction is relatively204
minor (mean value of 0.96) given the proximity of the virtual source to the real source epicentre205
(8.3 km), the period range of interest (3–10 s), and the fact that the virtual source and the earth-206
quake are located outside of the Kanto Basin (higher seismic wave velocity and therefore longer207
wavelengths).208
To reduce the azimuthal effects related to the non-uniform distribution of the ambient seismic209
field sources and to the surface-wave radiation pattern at the earthquake source (Denolle et al.210
2014a; Viens et al. 2016b), we only calibrate the IRFs within narrow azimuth ranges of 8◦. Finally,211
we select the stations located at distances greater than 30 km from the epicentre to ensure a similar212
path effect for the earthquake and IRF waveforms. The two corrections applied to the amplitudes213
and the station selection allow us to compare the simulated IRF PGVs with the observed, ground214
truth, earthquake PGVs.215
10 L. Viens et al.
For the stations located within the 264±4◦ azimuth angle from the epicentre, we plot the216
natural logarithm of the long-period PGVs as a function of the distance from the epicentre in Figure217
5. We also show the 1/√x and 1/
√x× e−αx theoretical curves, where x is the epicentre–receiver218
distance and α is an effective attenuation coefficient. We choose a constant value of α = 14×10−3219
km−1, which is close to the values found by Prieto et al. (2011) in the Los Angeles sedimentary220
basin (e.g., α = 2–5 × 10−3 km−1 in the 5 to 10 s period range). In Figure 5, we can observe221
that the amplitude decay of the IRF and earthquake PGVs follows the 1/√x × e−αx theoretical222
curve relatively well for distances smaller than 80 km and larger than 140 km from the epicentre.223
However, between 80 and 140 km, there is a large amplification for both predicted and observed224
values. This amplification correlates well with the deepest part of basin (Koketsu et al. 2008, 2012),225
which is also shown in Figure 5. Such amplification at large distance demonstrates that elastic 3-D226
effects of wave focusing overcome the attenuation in Kanto Basin.227
To further quantify the match between the observed and predicted amplitudes, we compute the228
root mean square error (RMS error) between the natural logarithm (ln) of the PGVs, which can be229
written as230
RMS error =
√∑Nn=1(ln(PGV
neq)− ln(PGV n
IRF ))2
N, (4)
231
where PGV nGF and PGV n
eq are the IRF and earthquake PGVs, respectively, and n is the nth sta-232
tion of a total number of stations N located within an azimuth range. For the 67 stations located233
within the 264±4◦ azimuth range, the RMS errors are equal to 0.239, 0.457, and 0.251 for the de-234
convolution, coherency, and cross correlation, respectively. The minimum RMS error is obtained235
for the IRFs retrieved with the deconvolution method. The RMS error computed from the PGVs236
retrieved with the cross correlation technique is comparable to that from the deconvolution and237
much smaller than the one from the coherency technique.238
To study the azimuthal dependence of the RMS error, we compute the RMS error for stations239
located within 8◦ azimuth ranges, every 2◦ between 254◦ and 290◦ from the epicentre. Results are240
shown for the raw and 1-bit cross correlation in Figure 6a and for the coherency and deconvolution241
Retrieval of impulse response function amplitudes 11
in Figure 6b. The largest RMS error is obtained for the coherency technique at almost all azimuth242
angles and the smallest RMS errors are found with the raw cross correlation and deconvolution243
methods. We also find that the RMS error is higher for the azimuth angles larger than 284◦. To244
evaluate the consistency of the PGVs in the RMS estimates, we bootstrap Equation 4 by randomly245
resampling 1000 times the original PGV values and evaluating the 95% confidence interval of the246
RMS error of the new set of PGV values (Figure 6). As expected, fewer stations in the azimuth247
angles larger than 284◦ lead to a greater uncertainty in the RMS value, but the overall mean RMS248
value may arise from biases due to non-uniform source distribution.249
5 DISCUSSION250
5.1 Seasonal variations251
The IRFs studied in the previous sections are computed using 1 month of data recorded in Octo-252
ber 2014. To investigate the effect of possible seasonal variations, we compute the deconvolutions253
using Equation 3 for the months of April, June, August, and December 2014. The waveforms254
between the TENNOD virtual source and the E.SRTM and E.YMMM receivers, which are nor-255
malized with the maximum value of month of April, are shown in Figures 7a and 7b, respectively.256
The shape of the waveform slightly changes over the year, but the different wave arrivals are ob-257
served for each month. Moreover, the maximum amplitude is almost constant over the year and no258
clear seasonal pattern can be deduced for these two stations.259
We compute the maximum values of the waveforms normalized with the month of April for260
all the stations located in the 250–294◦ azimuth range from the epicentre, and summarize their261
mean and standard deviation values in Table 1. The amplitudes are the lowest in April, but weakly262
vary with the seasons (less than 10%) and the standard deviation is around 0.20 for each month.263
This shows that the peak value of the waveforms is stable throughout the year in the Kanto Basin.264
This contrasts with the study of Stehly et al. (2006) that noticed dramatic seasonal variations in265
the amplitudes with the seismic network in southern California.266
To verify whether or not the small amplitude variations observed over the year impact the267
retrieval of earthquake PGVs, we calibrate the IRF amplitudes for each month following the same268
12 L. Viens et al.
procedure as in Section 4. Figure 8 shows the RMS errors computed using Equation 4 for each269
month against the azimuth angle from the epicentre. Small variations can be observed over the270
different months, but the general trend remains the same throughout the year. One interesting271
feature is that the RMS error for the month of August is the highest for azimuth angles smaller than272
262◦, but the smallest for azimuth angles greater than 280◦. This can be interpreted that August273
experienced a more localized, or nearer noise source to influence the amplitude with the somewhat274
narrow azimuth coverage. December experiences the opposite trend, likely for a similar reason.275
The month of October allows to retrieve minimum RMS errors in the central part of the basin276
where strong and complex 3-D effects are observed. One possible explanation for the minimum277
RMS error for the data recorded in October 2014 is that two typhoons passed over the Kanto Basin278
during this month and might have excited the ambient seismic field more efficiently. However,279
further work is needed to understand the potential effect of strong meteorological events on our280
results.281
5.2 S-wave amplitudes282
Along the 274±4◦ azimuth angle, clear S waves can be retrieved. To investigate the reliability of283
their amplitudes, we perform the same kind of analysis as done for the surface waves. We first284
band-pass filter the data between 4 and 10 s and we select the phases traveling between 1.7 and285
5 km/s. We calibrate the amplitude of the S waves following the same procedure as in Section 4,286
but with the ratio of the virtual source–receiver and hypocentre–receiver distances to account for287
the geometrical spreading of body waves (xn/xh, where xn is the virtual source–receiver distance288
and xh is the hypocentre–receiver distance). The S-wave long-period PGVs of the earthquake and289
IRF are shown in Figure 9. The amplitude of the S waves are reliable with the coherency and290
deconvolution methods for distances larger than 60 km from the hypocentre. Similarly to surface291
waves, S waves are clearly amplified in the deepest part of the basin. The overestimation of the IRF292
PGVs for distances smaller than 60 km from the hypocentre might be explained by the fact that293
body waves travel directly from the hypocentre to the stations through the basement of the Kanto294
Basin. Therefore, body waves from the surface-to-surface instruments likely undertake a different295
Retrieval of impulse response function amplitudes 13
ray path than from the deep source to the surface instrument. Nevertheless, reliable amplitudes296
of body waves can be extracted from the ambient seismic field at sufficient distance from the297
hypocentre using the deconvolution technique.298
5.3 Can the anticausal part of the deconvolution be used?299
So far, we have only studied the amplitude information of the causal part of the IRFs for seismic300
waves propagating westward, away from the coast. For earthquakes located on the west of the301
basin, however, seismic waves will propagate toward the East and most of the energy of IRFs will302
be contained in the anticausal part. Such configuration is, for example, similar to Southern Cali-303
fornia (Stehly et al. 2006) where earthquakes are expected along the San Andreas Fault and their304
seismic waves will radiate towards the coast where the Los Angeles Metropolitan area is located.305
Therefore, evaluating the reliability of the amplitude information contained in the anticausal part306
of IRFs is critical to predicting shaking from those earthquakes using the ambient seismic field.307
We retrieved the IRFs with the deconvolution method for the data recorded in October 2014308
by regarding the E.ZKUM station as the virtual source. This station is located close to a Mw 4.7309
earthquake (Figure 1), which occurred in 28 January 2012 at a depth of 20 km (F-net/NIED cata-310
log). We also use the same calibration procedure as in Section 4 for the surface waves propagating311
at velocities between 0.5 and 3 km/s, and show the results for two azimuth angles in Figure 10.312
We choose the 70±4◦ angle because of the large number of stations and the 90±4◦ to sample the313
deepest part of the Kanto Basin. For these two azimuths, the RMS errors (e.g., 0.412 and 0.476)314
are comparable to those of the causal part. Observed and simulated PGVs for the stations located315
above the deepest part of the basin, which reaches a depth of 4.1 km, are clearly amplified com-316
pared to the stations located above a basin depth of approximately 3 km. This demonstrates that317
the amplitude of the anticausal part of IRFs can also be trusted.318
6 CONCLUSIONS319
We retrieved IRFs from the ambient seismic field recorded in the Kanto Basin using three seismic320
interferometry techniques. We first showed that the waveforms retrieved with the cross correla-321
14 L. Viens et al.
tion technique have a narrower frequency content than the IRFs extracted with the coherency322
and deconvolution methods. To validate the surface-wave amplitudes of the amplitude calibrated323
IRFs, we compared their long-period PGVs to those of a shallow Mw 5.8 earthquake that occurred324
nearby the virtual source station. The long-period PGVs of the waveforms retrieved with the cross325
correlation and deconvolution techniques agree well with the ones of the earthquake and complex326
3-D effects, likely caused by the basin structure, are well captured by the ambient seismic field.327
We also investigated the sensitivity of the amplitudes to seasonal variations and showed that328
despite small variations, they remain relatively constant throughout the year for the deconvolution329
method. We also showed that the amplitudes of S waves are preserved for distances greater than330
60 km from the hypocentre with the deconvolution and coherency techniques. Finally, we con-331
ducted the same analysis on the surface-wave amplitude of the anticausal part of the IRFs from332
the deconvolution using an earthquake that occurred on the western part of the basin. The large333
amplification caused by the deepest part of the Kanto Basin is also well retrieved.334
This analysis demonstrates that the deconvolution technique produces broadband IRFs that335
contain both surface- and S-wave amplitudes that reproduce those of earthquakes in the complex336
Kanto Basin. Finally, IRFs obtained from deconvolution can be used to predict accurate seismic337
amplitudes of future earthquakes.338
ACKNOWLEDGMENTS339
We acknowledge Hi-net/NIED and JMA for providing the continuous data and F-net/NIED for the340
information about the earthquakes used in this study. We thank Naoshi Hirata and all the MeSO-341
net project members for the MeSO-net data. The MeSO-net project is supported by the Special342
Project for Reducing Vulnerability for Urban Mega Earthquake Disasters from the Ministry of343
Education, Culture, Sports, Science, and Technology (MEXT) of Japan.344
REFERENCES345
Bensen, G. D., Ritzwoller, M. H., Barmin, M. P., Levshin, A. L., Lin, F., Moschetti, M. P., Shapiro, N. M.346
& Yang, Y., 2007. Processing seismic ambient noise data to obtain reliable broad-band surface wave347
Retrieval of impulse response function amplitudes 15
dispersion measurements, Geophys. J. Int., 169, 1239–1260.348
Boue, P., Poli, P., Campillo, M., Pedersen, H., Briand, X. & Roux, P., 2013. Teleseismic correlations of349
ambient seismic noise for deep global imaging of the Earth, Geophys. J. Int., 194, 844–848.350
Boue, P., Denolle, M., Hirata, N., Nakagawa, S. & Beroza, G. C., 2016. Beyond basin resonance: char-351
acterizing wave propagation using a dense array and the ambient seismic field, Geophys. J. Int., 206,352
1261–1272.353
Bowden, D. C., Tsai, V. C. & Lin, F. C., 2015. Site amplification, attenuation, and scattering from noise354
correlation amplitudes across a dense array in Long Beach, CA, Geophys. Res. Lett., 42, 1360–1367.355
Brenguier, F., Campillo, M., Hadziioannou, C., Shapiro, N. M., Nadeau, R. M., & Larose, E., 2008a. Post-356
seismic relaxation along the San Andreas fault at Parkfield from continuous seismological observations,357
Science, 321, 1478–1481.358
Brenguier, F., Shapiro, N. M., Campillo, M., Ferrazzini, V., Duputel, Z., Coutant, O. & Nercessian, A.,359
2008b. Towards forecasting volcanic eruptions using seismic noise, Nature Geoscience, 1, 126–130.360
Brenguier, F., Campillo, M., Takeda, T., Aoki, Y., Shapiro, N. M., Briand, X., Emoto, K. & Miyake, H.,361
2014. Mapping pressurized volcanic fluids from induced crustal seismic velocity drops, Science, 345,362
80–82.363
Campillo, M. & Paul, A., 2003. Long-range correlations in the diffuse seismic coda, Science, 299, 547–364
549.365
Cupillard, P. & Capdeville, Y., 2010. On the amplitude of surface waves obtained by noise correlation and366
the capability to recover the attenuation: a numerical approach, Geophys. J. Int., 181, 1687–1700.367
Denolle, M. A., Dunham, E. M., Prieto, G. A. & Beroza, G. C., 2013. Ground motion prediction of realistic368
earthquake sources using the ambient seismic field, J. Geophys. Res. Solid Earth, 118, 2102–2118.369
Denolle, M. A., Dunham, E. M., Prieto, G. A. & Beroza, G. C., 2014a. Strong ground motion prediction370
using virtual earthquakes, Science, 343, 399–403.371
Denolle, M. A., Miyake, H., Nakagawa, S., Hirata, N. & Beroza, G. C., 2014b. Long-period seismic372
amplification in the Kanto Basin from the ambient seismic field, Geophys. Res. Lett., 41, 2319–2325.373
Furumura, T. & Hayakawa, T., 2007. Anomalouspropagation of long-period ground motions recorded in374
Tokyo during the 23 October 2004 Mw 6.6 Niigata-ken Chuetsu, Japan, earthquake, Bull. Seismol. Soc.375
Am., 97, 863–880.376
Kasahara, K., Sakai, S., Morita, Y., Hirata, N., Tsuruoka, H., Nakagawa, S., Nanjo, K. Z. & Obara, K.,377
2009. Development of the Metropolitan Seismic Observation network (MeSO-net) for Detection of378
Mega-thrust beneath Tokyo Metropolitan Area, Bull. Earthq. Res. Inst., 84, 71–88.379
Koketsu, K., Miyake, H., Fujiwara, H. & Hashimoto, T., 2008. Progress towards a Japan integrated velocity380
structure model and long-period ground motion hazard map, Proceedings of the 14th World Conference381
on Earthquake Engineering, Paper No S10–038.382
16 L. Viens et al.
Koketsu, K., Miyake, H. & Suzuki, H., 2012. Progress towards a Japan integrated velocity structure model383
and long-period ground motion hazard map, Proceedings of the 15th World Conference on Earthquake384
Engineering, Paper No 1773.385
Kudo, K., 1978. The contribution of Love waves to strong ground motions, Proceedings of the 2nd Inter-386
national Conference on Microzonation, 2, 765–776.387
Kudo, K., 1980. A study on the contribution of surface wave to strong ground motions, Proceedings of the388
7th World Conference on Earthquake Engineering, pp. 499–506.389
Lawrence, J. F. & Prieto, G. A., 2011. Attenuation tomography of the western united states from ambient390
seismic noise, Journal of Geophysical Research: Solid Earth, 116, B06302.391
Lawrence, J. F., Denolle, M., Seats, K. J. & Prieto, G. A., 2013. A numeric evaluation of attenuation from392
ambient noise correlation functions, J. Geophys. Res. Solid Earth, 118, 6134–6145.393
Lin, F.-C., Moschetti, M. P. & Ritzwoller, M. H., 2008. Surface wave tomography of the western United394
States from ambient seismic noise: Rayleigh and Love wave phase velocity maps, Geophys. J. Int., 173,395
281–298.396
Lin, F.-C., Li, D., Clayton, R. W. & Hollis, D., 2013. High-resolution 3d shallow crustal structure in long397
beach, california: Application of ambient noise tomography on a dense seismic array, Geophysics, 78,398
Q45–Q56.399
Longuet-Higgins, M. S., 1950. A Theory of the Origin of Microseisms, Philosophical Transactions of the400
Royal Society of London A: Mathematical, Physical and Engineering Sciences, 243, 1–35.401
Mordret, A., Shapiro, N. M., Singh, S., Roux, P., Montagner, J.-P. & Barkved, O. I., 2013. Azimuthal402
anisotropy at valhall: The helmholtz equation approach, Geophys. Res. Lett., 40, 2636–2641.403
Mordret, A., Landes, M., Shapiro, N. M., Singh, S. C. & Roux, P., 2014. Ambient noise surface wave404
tomography to determine the shallow shear velocity structure at Valhall: depth inversion with a Neigh-405
bourhood Algorithm, Geophys. J. Int., 198, 1514–1525.406
Nishida, K., Kawakatsu, H. & Obara, K., 2008. Three-dimensional crustal S wave velocity structure in407
Japan using microseismic data recorded by Hi-net tiltmeters, J. Geophys. Res. Solid Earth, 113, B10302.408
Obara, K., Kasahara, K., Hori, S. & Okada, Y., 2005. A densely distributed high-sensitivity seismograph409
network in Japan: Hi-net by National Research Institute for Earth Science and Disaster Prevention, Rev.410
Sci. Instrum., 76, 021301.411
Okada, Y., Kasahara, K., Hori, S., Obara, K., Sekiguchi, S., Fujiwara, H. & Yamamoto, A., 2004. Recent412
progress of seismic observation networks in Japan - Hi-net, F-net, K-NET and KiK-net, Earth Planets413
Space, 56, xv–xxviii.414
Poli, P., Pedersen, H. A., Campillo, M. & the POLENET/LAPNET Working Group, 2012. Emergence of415
body waves from cross-correlation of short period seismic noise, Geophys. J. Int., 188, 549–558.416
Prieto, G. A. & Beroza, G. C., 2008. Earthquake ground motion prediction using the ambient seismic field,417
Retrieval of impulse response function amplitudes 17
Geophys. Res. Lett., 35, L14304.418
Prieto, G. A., Lawrence, J. F. & Beroza, G. C., 2009. Anelastic Earth structure from the coherency of the419
ambient seismic field, J. Geophys. Res. Solid Earth, 114, B07303.420
Prieto, G. A., Denolle, M., Lawrence, J. F. & Beroza, G. C., 2011. On amplitude information carried by421
the ambient seismic field, Comput. Geosci., 343, 600–614.422
Roux, P., Sabra, K. G., Kuperman, W. A. & Roux, A., 2005. Ambient noise cross correlation in free space:423
Theoretical approach, The Journal of the Acoustical Society of America, 117, 79–84.424
Roux, P., Moreau, L., Lecointre, A., Hillers, G., Campillo, M., Ben-Zion, Y., Zigone, D. & Vernon, F.,425
2016. A methodological approach towards high-resolution surface wave imaging of the san jacinto fault426
zone using ambient-noise recordings at a spatially dense array, Geophys. J. Int., 206, 980–992.427
Sakai, S. & Hirata, N., 2009. Distribution of the Metropolitan Seismic Observation network, Bull. Earthq.428
Res. Inst., 84, 57–69.429
Shapiro, N. M. & Campillo, M., 2004. Emergence of broadband Rayleigh waves from correlations of the430
ambient seismic noise, Geophys. Res. Lett., 31, L07614.431
Shapiro, N. M., Campillo, M., Stehly, L. & Ritzwoller, M. H., 2005. High-resolution surface-wave tomog-432
raphy from ambient seismic noise, Science, 307, 1615–1618.433
Snieder, R., 2004. Extracting the Green’s function from the correlation of coda waves: a derivation based434
on stationary phase, Phys. Rev. E, 69, 046610.435
Stehly, L., Campillo, M. & Shapiro, N. M., 2006. A study of the seismic noise from its long-range436
correlation properties, J. Geophys. Res. Solid Earth, 111, B10306.437
Takeo, A., Nishida, K., Isse, T., Kawakatsu, H., Shiobara, H., Sugioka, H. & Kanazawa, T., 2013. Radially438
anisotropic structure beneath the Shikoku Basin from broadband surface wave analysis of ocean bottom439
seismometer records, J. Geophys. Res. Solid Earth, 118, 2878–2892.440
Tsai, V. C., 2011. Understanding the amplitudes of noise correlation measurements, J. Geophys. Res. Solid441
Earth, 116, B09311.442
Vasconcelos, I. & Snieder, R., 2008. Interferometry by deconvolution: Part 1–Theory for acoustic waves443
and numerical examples, Geophysics, 73, S115.444
Viens, L., Miyake, H. & Koketsu, K., 2015. Long-period ground motion simulation of a subduction445
earthquake using the offshore-onshore ambient seismic field, Geophys. Res. Lett., 42, 5282–5289.446
Viens, L., Koketsu, K., Miyake, H., Sakai, S. & Nakagawa, S., 2016a. Basin-scale Green’s functions from447
the ambient seismic field recorded by MeSO-net stations, J. Geophys. Res. Solid Earth, 121, 2507–2520.448
Viens, L., Miyake, H. & Koketsu, K., 2016b. Simulations of long-period ground motions from a large449
earthquake using finite rupture modeling and the ambient seismic field, J. Geophys. Res. Solid Earth,450
121, 2507–2520.451
Wapenaar, K., 2004. Retrieving the Elastodynamic Green’s Function of an Arbitrary Inhomogeneous452
18 L. Viens et al.
Medium by Cross Correlation, Phys. Rev. Lett., 93, 254301.453
Weaver, R. L., 2011. On the amplitudes of correlations and the inference of attenuations, specific intensities454
and site factors from ambient noise, Comptes Rendus Geoscience, 343, 615 – 622.455
Retrieval of impulse response function amplitudes 19
Table 1. Mean and standard deviation of the peak values of the waveform for the months of April, June,
August, October, December 2014, normalized to peak value of the month of April 2014. For each month,
we use the deconvolution method between the TENNOD virtual source and 239 receiver stations located
within the 250–294◦ angle from the epicentre.
Normalized peak value April June August October December
Mean 1.00 1.06 1.06 1.04 1.11
Standard deviation 0.00 0.19 0.22 0.20 0.20
20 L. Viens et al.
290 °
254°
TENNOD
E.ZKUM
E.SRTM
E.YMMM
25 km
Mw 5.8
Mw 4.7
139.0° E 139.5
° E 140.0
° E 140.5
° E 141.0
° E
35.0° N
35.5° N
36.0° N
36.5° N
MeSO−net Hi−net JMA Univ. Tokyo Earthquakes
Basin depth (km)
4
2
0
PhilippineSea Plate
Pacific Plate
Figure 1. Map of the Kanto Basin, Japan, including stations of the MeSO-net (purple triangles), Hi-net
(black triangles), JMA (blue triangles), and University of Tokyo (green triangles) networks. Coastlines are
represented by the black lines and five hundred meter spaced basin depth contours derived from the JIVSM
(Koketsu et al. 2008, 2012) are shown by the coloured lines. The names of the TENNOD and E.ZKUM
virtual sources and the two receivers for which the waveforms are shown in Figures 2 and 7 are also showed.
The epicentre of theMw 5.8 and 4.7 earthquakes are represented by the red stars and their focal mechanisms
are plotted. The two azimuth angles of 254◦ and 290◦ that are used in the following are represented by the
two black lines departing from the Mw 5.8 earthquake epicentre. The upper right insert shows the Japanese
Islands (black lines), plate boundaries (grey lines), and the zoomed region (red square).
Retrieval of impulse response function amplitudes 21
0 50 100 150 200 250
Time (s)
-1
0
1
2
3
4
5
6
7
8
No
rma
lize
d a
mp
litu
de
TENNOD - E.SRTM (44 km), BP: 1-10 s
Vertical component
Deconvolution
Coherency
1-bit
cross correlation
Raw
cross correlation
(a)
S wave
Surface waves
1st arrivals 2nd arrivals
0 50 100 150 200 250
Time (s)
-1
0
1
2
3
4
5
6
7
8
No
rma
lize
d a
mp
litu
de
TENNOD - E.YMMM (67 km), BP: 1-10 s
Vertical component
Deconvolution
Coherency
1-bit
cross correlation
Raw
cross correlation
(b)
S wave
Surface waves
1st arrivals 2nd arrivals
Figure 2. Impulse response functions retrieved from the raw cross correlation (red), 1-bit cross correlation
(orange), coherency (green), and deconvolution (blue) techniques between the virtual source (TENNOD)
and the (a) E.SRTM and (b) E.YMMM stations, which are located at 44 and 67 km from the virtual source,
respectively. The location of these two stations is shown in Figure 1. All the waveforms are shown for the
Z–Z component and are bandpass filtered between 1 and 10 s. The name of each group of arrivals is also
indicated.
22 L. Viens et al.
0.1 0.2 0.3 0.4 0.5 0.6 0.8 1.0
Frequency (Hz)
10-4
10-2
100
102
Norm
aliz
ed a
mplit
ude
Fourier spectra
Vertical component
Deconvolution
Raw cross correlation
Figure 3. Fourier spectra of the impulse response functions extracted using the deconvolution (black) and
the cross correlation (red) techniques between the TENNOD virtual source and 28 receiver stations. The
thick lines represent the geometric mean of the 28 Fourier spectra. The waveforms are normalized with the
peak value of the closest station to the virtual source (18 km) before being Fourier transformed.
Retrieval of impulse response function amplitudes 23
0 50 100 150 200 250 300 350
Time (s)
20
30
40
50
60
70
80
90
100
110
Dis
tan
ce
fro
m t
he
TE
NN
OD
sta
tio
n
Band-pass filter 3 - 10 s
Vertical component
1st arrival of
surface waveS wave
Raw cross correlation
1-bit cross correlation
Coherency
Deconvolution
Figure 4. Envelope of the Z–Z impulse response functions retrieved from the raw cross correlation (red),
1-bit cross correlation (orange), coherency (green), and deconvolution (black) techniques as a function of
the distance from the virtual source. All the impulse response functions are normalized with the maximum
value of the waveform at the closest station from the virtual source, corrected for the geometrical spreading
of surface waves (1/√xs, where xs is the virtual source–receiver distance), and band-pass filtered in the
period range of 3 to 10 s.
24 L. Viens et al.
TENNOD
25 km
264 ± 4°
Mw 5.8
139° E 140
° E 141
° E
35° N
36° N
Basin depth (km)
4
2
0
-4
-3
-2
-1
Ve
rtic
al co
mp
on
en
t
ln o
f lo
ng
-pe
rio
d P
GV
(ln
(cm
/s))
Deconvolution
RMS error: 0.239
50 100 150 200
Distance from epicentre (km)
3
2
1
0
-1
Ba
sin
de
pth
(km
)
Vs = 3.2 km/s
1.5
0.9
0.5
Coherency
RMS error: 0.457
Earthquake
IRF
50 100 150 200
Distance from epicentre (km)
3
2
1
0
-1
Vs = 3.2 km/s
1.5
0.9
0.5
-4
-3
-2
-1
Raw cross correlation
RMS error: 0.251
50 100 150 200
Distance from epicentre (km)
3
2
1
0
-1
Vs = 3.2 km/s
1.5
0.9
0.5
Figure 5. (Upper panel) Map of the Kanto Basin including the basin depth contours from the JIVSM and
all the seismic stations. The receiver stations included in the 264±4◦ azimuth range, which is represented
by the two black lines, are used in this figure. (Middle panels) Natural logarithm of the long-period PGVs
of the earthquake (red) and impulse response functions (IRFs) after amplitude calibration (blue) waveforms
for the three different techniques. The black curves are the theoretical curves 1/√x and 1/
√xe−αx, where
x is the epicentre–receiver distance and α is an attenuation coefficient taken as 14 × 10−3 km−1. For each
panel, the RMS error computed using Equation 4 is also shown. (Bottom panels) Velocity profile extracted
from the JIVSM (Koketsu et al. 2008, 2012) for an azimuth angle of 264◦ from the epicentre. The S-wave
velocity of each layer is indicated and the thick line represents the basin depth.
Retrieval of impulse response function amplitudes 25
252 256 260 264 268 272 276 280 284 288 292
Azimuth angle (° )
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
0.7
RM
S e
rror
(a)
Raw cross correlation
1-bit cross correlation
252 256 260 264 268 272 276 280 284 288 292
Azimuth angle (° )
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
0.7
RM
S e
rror
(b)
Coherency
Deconvolution
Figure 6. (a) Mean of the RMS error computed from the long-period PGVs of the earthquake and the
impulse response functions retrieved with the raw cross correlation (red squares) and 1-bit cross correlation
(orange squares) techniques as a function of the azimuth angle from the epicentre. For each azimuth angle,
the 95% confidence interval of the bootstrap is indicated by the error bars. (b) Same as Figure 6a for the
RMS errors computed from the long-period PGVs of the earthquake and the impulse response functions
computed with the coherency (green) and deconvolution (black) methods.
26 L. Viens et al.
0 50 100 150 200−1
0
1
2
3
4
5
6
7
8
9
Norm
aliz
ed a
mplit
ude
TENNOD − E.SRTM (44 km), BP: 3−10 sVertical component
April (1)
June (1.18)
August (0.99)
October (1.09)
December (1.19)
Time (s)
(a)
0 50 100 150 200−1
0
1
2
3
4
5
6
7
8
9
Norm
aliz
ed a
mplit
ude
TENNOD − E.YMMM (67 km), BP: 3−10 sVertical component
April (1)
June (0.87)
August (0.88)
October (0.93)
December (1.03)
Time (s)
(b)
Figure 7. Impulse response functions retrieved with the deconvolution method between the virtual source
(TENNOD) and the (a) E.SRTM and (b) E.YMMM stations from the raw data recorded in April (black),
June (blue), August (orange), October (red) and December (green) 2014. The waveforms are normalized
with the peak amplitude of the impulse response retrieved for the month of April and the maximum value
of each waveform is indicated between parenthesis. All the waveforms are shown for the Z–Z component
and are bandpass filtered between 3 and 10 s.
Retrieval of impulse response function amplitudes 27
252 256 260 264 268 272 276 280 284 288 292
Azimuth angle (° )
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
RM
S e
rror
April
June
August
October
December
Figure 8. Mean of the RMS error computed from the long-period PGVs of the earthquake and impulse
response functions extracted with the deconvolution technique using the data recorded in April (black),
June (blue), August (orange), October (red), and December (green) 2014 as a function of the azimuth angle
from the epicentre.
28 L. Viens et al.
TENNOD
25 km
274 ± 4°
Mw 5.8
139° E 140
° E 141
° E
35° N
36° N
Basin depth (km)
4
2
0
-4.5
-4
-3.5
-3
-2.5
-2
-1.5
-1
Ve
rtic
al co
mp
on
en
t
ln o
f L
on
g-p
erio
d P
GV
(ln
(cm
/s))
Deconvolution
RMS error: 0.392
50 100 150 200
Distance from hypocentre (km)
3
2
1
0
-1
-2
Ba
sin
de
pth
(km
)
Vs = 3.2 km/s
1.5
0.9
0.5
Coherency
RMS error: 0.401
Earthquake
IRF
50 100 150 200
Distance from hypocentre (km)
3
2
1
0
-1
-2
Vs = 3.2 km/s
1.5
0.9
0.5
-4.5
-4
-3.5
-3
-2.5
-2
-1.5
-1
Raw cross correlation
RMS error: 0.464
50 100 150 200
Distance from hypocentre (km)
3
2
1
0
-1
-2
Vs = 3.2 km/s
1.5
0.9
0.5
Figure 9. (Upper panel) Map of the Kanto Basin including the basin depth contours from the JIVSM and all
the seismic stations. The receiver stations included in the 274±4◦ azimuth range, which is represented by the
two black lines, are used in this figure. (Middle panels) Natural logarithm of the S-wave long-period PGVs
of the earthquake (red) and impulse response functions (IRFs) after amplitude calibration (blue) waveforms
for the three different techniques as a function of the distance from the hypocentre. For each panel, the RMS
error computed using Equation 4 is also shown. (Bottom panels) Velocity profile extracted from the JIVSM
(Koketsu et al. 2008, 2012) for an azimuth angle of 274◦ from the hypocentre. The S-wave velocity of each
layer is indicated and the thick line represents the basin depth.
Retrieval of impulse response function amplitudes 29
E.ZKUM
25 km
Mw 4.7
70 ± 4°
90 ± 4°
139° E 140
° E 141
° E
35° N
36° N
Basin depth (km)
4
2
0
-6.5
-6
-5.5
-5
-4.5
-4
-3.5
-3
-2.5
Ve
rtic
al co
mp
on
en
t
ln o
f L
on
g-p
erio
d P
GV
(ln
(cm
/s))
Azimuth: 70 4°
RMS error: 0.412
Earthquake
IRF
30 50 100 150
Distance from epicentre (km)
4
3
2
1
0
Ba
sin
de
pth
(km
)
Vs = 3.2 km/s
1.5
0.9
0.5
-6.5
-6
-5.5
-5
-4.5
-4
-3.5
-3
-2.5V
ert
ica
l co
mp
on
en
t
ln o
f L
on
g-p
erio
d P
GV
(ln
(cm
/s))
Azimuth: 90 4°
RMS error: 0.476
30 50 100 150
Distance from epicentre (km)
4
3
2
1
0
Ba
sin
de
pth
(km
)
Vs = 3.2 km/s
1.5
0.9
0.5
Figure 10. (Upper panel) Map of the Kanto Basin including the basin depth contours from the JIVSM and
all the seismic stations. The receiver stations included in the 70 ±4◦ and 90 ±4◦ azimuth ranges, which
are represented by the black lines, are used in this figure. (Middle panels) Natural logarithm of the long-
period PGVs of the surface waves of the Mw 4.7 earthquake (red) and that of the anticausal IRF from
deconvolution after amplitude calibration (blue) waveforms as a function of the distance from the epicentre.
For each panel, the RMS error computed using Equation 4 is also shown. (Bottom panels) Velocity profile
extracted from the JIVSM (Koketsu et al. 2008, 2012) for the 70 and 90◦ azimuth angles from the epicentre.
The S-wave velocity of each layer is indicated and the thick line represents the basin depth.