Are transients carrying useful information for
estimating H/V spectral ratios?
S. Parolai1, M. Picozzi1, A. Strollo1,2, M. Pilz1, D. Di Giacomo1,2, B. Liss1, and
D. Bindi3
1
GeoForschungsZentrum Potsdam, Telegrafenberg, 14473 Potsdam, Germany
2
Univertität Potsdam, KarlLiebknechtStrasse, 14476 Potsdam, Germany
3
Istituto Nazionale di Geofisica e Vulcanologia, via Bassini 15, 20133 Milano, Italy.
Abstract
Defining a procedure to calculate the H/V spectral ratio of seismic noise has
recently been the focus of several studies. Although some issues have been
addressed and a general procedure to follow in choosing the optimal seismological
equipment, carrying out the measurements and analyzing the data has been agreed
upon by the majority of specialists in this field, some points still remain open. One
open question is whether transients should be avoided when H/V spectral ratios
are calculated. Starting from contradictory results of previous papers, an adhoc
experiment at the Nauen test site (Germany) was conducted to verify the influence
of transients on the H/V spectral ratio of seismic noise. We show, consistent with
some previous studies, that transients generated by high frequency content sources
can amplify spectral ratio peaks at low frequencies. After careful polarization
analysis of the data and estimation of the spectral amplitude decay with distance,
we provide an explanation in terms of the composition of the wavefield over
certain distance ranges. Our results highlight the importance of a better
understanding of the wavefield composition for better exploiting the potential of
H/V spectral ratio method.
1 Introduction
The Nakamura technique (Nakamura 1989), first proposed by Nogoshi and
Igarashi (1970, 1971), is today one of the most commonly applied methods for
microzonation studies of large urban areas (e.g. Parolai et al. 2001). A large
number of studies using this rapid, economical, and therefore attractive technique
have been published (e.g. Field and Jacob 1993; Lermo and Chávez-García 1994;
Mucciarelli 1998; Bard 1999; Fäh et al. 2001). More recent studies (Fäh et al.
2
2003; Scherbaum et al. 2003; Arai and Tokimatsu 2004; Parolai et al. 2005;
Picozzi et al. 2005; Parolai et al. 2006) proposed and tested the possibility of
inverting the HorizontaltoVertical (H/V) spectral ratio of noise (alone or in a
joint inversion scheme) for investigating the Swave subsoil structure. All these
studies require the apriori assumption that the noise wavefield is generally
dominated by surface waves. Moreover, attempts to provide standards for the
analysis of seismic noise have only recently been carried out (Bard 1999;
SESAME 2004; Picozzi et al. 2005). To this regard, a point frequently debated by
the seismological community is whether only the stationary part of the recorded
signal should be considered in the analysis, or could transients (e.g. due to human
activities) also be considered (excluding obviously very strong and clipped
signals). Most authors disregarded the nonstationary noise from the analysis (e.g.
Horike et al. 2001), while others (e.g. Mucciarelli et al. 2003) showed that the H/V
ratio of nonstationary noise might be more similar, especially in amplitude, to the
H/V spectral ratio determined for smallsize earthquakes. By applying an optimal
filtering procedure to empirical data and using numerical simulations, Parolai and
GalianaMerino (2006) showed that transients might cause large variability in the
H/V shape, depending on source type and distance from the receiver relative to the
thickness of the sedimentary cover. However, they showed that when dealing with
a sufficient number of real data, the H/V ratio is not biased by including
transients. Recently, Chatelain et al. (2007) suggested to avoid the use of
transients in the H/V analysis. Nevertheless, they showed that H/V does not
significantly change if the sensor is located just a few meters from the major
source of transients (in their case, cars on a highway). Furthermore, at only 2 m
from the highway, the peak in the H/V spectral ratio is still visible, while it
disappears when the sensor is placed directly on the highway. These results
(except for those from the stations located directly on the highway) seem to be in
agreement with the pioneering study of Taniguchi and Sawada (1979) who made
systematic noise measurements close to a highway under constructions and
observed that (1) the frequency peak in the transient spectra was stable
independent of the mass of the source (a truck), hinting to the fact that the
frequency of trafficinduced vibration is mainly determined by ground soil
conditions, and (2) trafficgenerated vibrations were dominated by Rayleigh
waves. A recent investigation of the seismic noise wavefield induced by several
source types can be found in Kim and Lee (2007).
Additional evidence of the role played by transients was reported by
Mucciarelli (1998) who showed that when generating transients with a sledge
hammer near the recording sensor, the H/V peak becomes more clear. This result,
3
probably too often overlooked, stimulates many intriguing questions. Since the
energy released by the blow of a sledge hammer peaks at frequencies much higher
(5070 Hz) than the fundamental resonance frequency of the site (1 Hz), how can
the transients modify the H/V? Similarly, human generated seismic noise mainly
affects frequencies higher than 12 Hz, with cars generating signals with
frequencies mainly between 1020 Hz (e.g. Taniguchi and Sawada 1979;
McNamara and Buland 2004). Therefore, the question arises if there is any
important information carried by the transients that one could try to extract.
We endeavour to answer these questions, by analysing recordings of seismic
noise that include transients generated by a sledge hammer. Vertical and
horizontal sources have been simulated. We first calculate the H/V spectral ratio
using only the stationary part of noise. Then, we calculate the H/V spectral ratio
for the transients only. A comparison of the results, considering spectra
inspection, polarization, and amplitude decay analysis is performed. Finally, a
possible explanation of the observed phenomena in terms of wave propagation is
proposed.
2 Data
The experiment was carried out at the Nauen test site (Germany)
(http://www.geophysik.tuberlin.de/menue/testfeld_nauen/), during a day with
favourable weather conditions and absence of wind.
Seismic noise was recorded using a 24bit digitizer (EarthDataLogger)
connected to a Mark L4C3D 1 Hz sensor. The sampling rate was fixed to 200
samples/sec.
Transients were generated by a surface source at distances of 5 m, 10 m, 20 m,
30 m, 50 m, 75 m and 100 m to the recorder. The source was a sledge hammer (5
kg) striking vertically a steel plate (vertical source) and horizontally
(perpendicular to the radial line between the source and the geophone) a railway
sleeper. Series of 5 vertical blows (separated from each other by a few seconds)
were alternated with series of five horizontal blows. For each sourcesensor
distance, this energization scheme was repeated five times (i.e. five series of
vertical and horizontal blows were recorded). Figures 1 and 2 show examples of
recordings for the vertical, radial and transverse component of ground motion for
the source distances 5 m and 100 m, respectively.
4
3 Data Processing
Using a visual inspection procedure, three different kinds of signal windows
were selected from the continuous data stream. In particular, windows of signal
containing only one series of vertical shots, only one series of horizontal shots and
a couple of consecutive vertical and horizontal shots were extracted (Figure 1 and
Figure 2). The length of the signal windows varies between 20.48 sec (for those
including only a series) and 40.96 sec (for those including a couple of vertical and
horizontal blow series). The time series were corrected for trend in the data and
tapered with a 5% cosine function at both ends. The Fast Fourier Transform (FFT)
was calculated for each component, and the spectra were smoothed using a Konno
and Ohmachi (1998) logarithmic window, with the coefficient b, which
determines the bandwidth, fixed to 40. The instrumental response correction was
performed by considering the pole and zero configuration of the sensor. The same
procedure was also applied to twenty 30sec length stationary noise windows.
Then, the HorizontaltoVertical spectral ratios (H/V) were calculated considering
the rootmeansquare (rms) spectra of the two horizontal components. The
procedure was repeated for each sourcesensor distance and source type
(horizontal and vertical). Finally, the logarithmic average of the H/V spectral
ratios was computed.
4 Amplitude Fourier spectra and H/V ratios
Figure 3 (left and middle frames) shows the average vertical and horizontal
spectra of stationary noise and of the windows including transients, for vertical
sources at different distances. The spectra are the logarithmic average of the five
series. Figure 3 (right) shows the H/V results obtained for stationary noise and for
transients generated at different source distances.
The former shows two peaks at around 0.2 Hz and 1 Hz that are consistent with
the H/V peaks obtained by Picozzi (personnel communication) after analyzing
data obtained from monitoring the site for one week. Since borehole
measurements and geophysical surveys (geoelectric and georadar) carried out at
the Nauen test site provide information about the structure of the upper 20 m only,
the two peaks cannot be linked to any apriori known impedance contrast below
the site, that is strong enough to generate resonance peaks. Note that both peaks
occur within the frequency band exploitable for the considered short period sensor
(Strollo et al. 2008a; Strollo et al. 2008b), and therefore can be considered as
reliable. However, since the lowest frequency peak, as shown in the following
5
analysis, occurs at frequencies too low to be affected by the high frequency
transients generated by the hammer blow, we will focus our attention mainly on
the peak at around 1 Hz. Figure 3 suggests that the peak at 1 Hz is mainly related
to a trough in the vertical component spectra, suggesting a link with the ellipticity
of Rayleigh waves (Konno and Ohmachi 1998).
Considering the transient windows, the spectral energy for both vertical and
horizontal components of ground motion is concentrated at frequencies higher
than 10 Hz, with a broader peak for the vertical component. The amplitude of the
spectral peaks, occurring at about 30 Hz, decreases with distance due to the
seismic wave attenuation, but at 100 m it is still higher than the spectral amplitude
of the stationary noise.
It is remarkable that only when the source is located at a distance smaller than
20 m, do the spectra also show an increase in amplitude between 0.3 and 2 Hz. In
particular, the largest amplitudes are of the radial component. The effect on the
H/V (note that here only that obtained by averaging the radial and transverse
component is shown) is also worth noting. In fact, while for short sourcesensor
distances (5 m and 10 m) the peak around 1 Hz is strongly magnified (the large
scattering in the results may be due to the fact that only 5 H/V spectral ratios have
been averaged), for larger source distances it approaches the stationary noise H/V.
Taking into account the amplitudes of the average spectra, this effect can mainly
be related to the radial component behaviour. Some transientrelated effects on the
H/V shape are visible only at frequencies larger than 10 Hz.
Figure 4 shows the results obtained when only the horizontal sources are
considered. The spectra of both horizontal and vertical components show large
amplitudes at frequencies higher than 10 Hz. These amplitudes are, however,
much smaller than those obtained using a vertical source. For example, the
spectral amplitudes of the signals generated by a horizontal source at a distance of
10 meters are comparable with the amplitudes of a vertical source at 20 meters.
This is probably due to the less efficient energization in the horizontal direction
and weak coupling of the sleeper with the soil. The spectral amplitude, as
expected, decays with increasing sourcetostation distance, and there is no large
significant effect of the transients at frequencies around 1 Hz, neither when the
short nor long sourcetostation distances are considered. All the H/Vs show
shapes similar to the stationarynoise H/V up to 10 Hz. Only at higher frequencies,
and especially for the distance range 1030 m, do transients appear to affect the
H/V spectral ratio shape.
Figure 5 shows the results when vertical and horizontal sources are considered
simultaneously. The results are similar to those obtained when considering only
6
the vertical sources since the spectral amplitude of the waves they generated are
higher.
Results and Discussions
The evidence that the H/V spectral ratio is enhanced at frequencies around the
resonance frequencies of the site when active sources are considered is in
agreement with the results of Mucciarelli (1998). Indeed, the sourcetodistance
range used in the Mucciarelli (1998) test was larger (50 m) than the range of
distances (510 m) we observed that provide the largest effects. Nevertheless, this
can be explained by considering the numerical simulation results of Parolai and
GalianaMerino (2006), who showed that the influence of transients varies with
the sourcetoreceiver distance relative to the thickness of the sediments.
In order to explain the mechanism generating the magnification of the H/V
spectral ratio, we carried out further analyses. After having verified by time
frequency analysis (not shown here) that the increase of the spectral amplitude
between 0.3 and 3 Hz (with welldefined peaks at 0.4 and 0.9 Hz), only results
from vertical energization (i. e, neither in the stationary noise, nor when transverse
sources are used), we focused the analysis only on the time series generated by the
vertical impact of the sledge hammer. First, we filtered all the time series windows
using a Gaussian filter (Dziewonski et al. 1969; Li et al. 1996; Shapiro and Singh
1999)
H (, 0 ) e 0 / 0
2
(1),
where 0 is the central frequency of the filter, fixed to 0.2 is the relative
bandwidth and the frequency. Filtering is performed by first computing the
Fourier transform of the trace, then multiplying the obtained complex spectrum by
the Gaussian filter and then calculating the inverse Fourier transform of the
filtered complex spectrum.
Second, the complex seismic traces (Taner et al. 1979; René et al. 1986; Parolai
2008) are calculated for each component by
7
c(t) = r(t) + iq(t) (2),
where r(t) is the real seismic trace, i is the square root of 1, and q(t) is the
quadrature trace obtained as the Hilbert transform of the real trace. Finally, using
complex seismic traces, we obtained instantaneous polarization attributes. Here
we will focus on reciprocal ellipticity and phase differences, where positive and
negative phase differences correspond to prograde and retrograde ground motion,
respectively; linear polarized waves have 0°Swave) or |180°| (Pwave) phase
differences, and a reciprocal ellipticity of 0. A reciprocal ellipticity of 1
corresponds to circular polarization.
Figure 6 shows the vertical and radialcomponent seismograms filtered using
the above described filter with a central frequency of 0.9 Hz. These time series
include only five consecutive vertical shots. Outside of the time window,
including the first and the last signal generated by the sledgehammer impacts
(e.g. occurring between about 5 and 15 s in Figure 6), small filter effects are
observed. Within this time window, the radial component of ground motion is
much larger than the vertical one, the reciprocal ellipticity is nearly 0 and the
phase difference is close to 0°. Figure 7 shows the results for the radial and
transverse components. The ground motion is predominant in the radial direction,
with small reciprocal ellipticity (0.2) and phase differences of nearly 120°. The
results for the vertical and transverse components (Figure 8) show that the
relatively small ground motion has reciprocal ellipticity that varies greatly (from
0.8 down to 0.3), whereas the phase difference is always around 90°.
Hence, the polarization analysis suggests that the spectral energy between 0.3
and 2 Hz observed at distances less than 20 m is carried out by waves linearly
polarized in the radial direction. Plotting the radial component spectral amplitude
at 0.9 Hz versus distance (scaled to the reference spectral amplitude at 5 m), a
decay slightly larger than R2, where R is the sourcetostation distance is observed
for the first 30 m (Figure 9). At larger distances the spectral amplitudes are
dominated by stationary seismic noise, and, therefore, do not decay any further.
In summary, considering that (1) body waves generated by a surficial source
and recorded at the surface attenuate with R 2 (Richart et al. 1970), (2) substantial
body waves affect the near field (Tokimatsu and Tamura 1995; Tamura 1996) and
(3) we observe an increase in energy also on the vertical component (although
much smaller than on the radial one), we believe that the contribution of surface
waves to the observed spectral amplitudes can be discarded. Moreover,
considering that a vertical source is expected to radiate a considerable amount of
power as shear waves (Miller and Pursey 1955) and the results of the polarization
8
analysis (a linearly polarized wave in the radial direction), we suggest that the
wavefield in the frequency range between 0.3 and 2 Hz for sources at distances
smaller than 20 m is dominated by SV waves.
The amplitude of the waves is relatively small with respect to higher
frequencies, but a resonance mechanism is amplifying them over shorter
distances. Although the resonance mechanism can still act when sources are
further away, the spectral energy of the direct SV arrival (that decays with R 2) is
too small to make the spectral amplitudes larger than those of the stationary noise.
The same mechanism could also explain why the 0.2 Hz resonance peak observed
in the H/V for stationary noise is not affected by the transients.
Therefore, spectral amplitudes can be determined by the superimposition of
spectral energy arriving directly from the source with a resonance mechanism. A
similar effect for P waves is unlikely, because they are expected to have smaller
amplitudes and because resonance should affect principally the vertical
component of ground motion.
The lack of similar evidence when a transverse source is considered (and
therefore generating SH waves more efficiently) could be explained by the lower
spectral amplitude of the horizontal source signal. Note, in fact, that the spectral
amplitudes (between 10 and 50 Hz) for a sourcetostation distance of 10 m
(Figure 4) are of the same order of magnitude than those generated by a vertical
source at 20 m distance. The smaller spectral amplitudes do not allow, even in
presence of resonance, to overstep the spectral amplitudes of stationary noise over
the 0.32 Hz frequency range.
Conclusions
In this study we report on the results of an adhoc experiment performed to
evaluate the influence of transients in H/V spectral ratio calculations. Consistent
with previous studies (e.g. Mucciarelli 1998), we observed that for certain
distance ranges the peak in the H/V spectra ratio can be magnified when energy is
added to the system. We proposed an explanation of the experimental evidence
that agrees with the stateoftheart knowledge in vibration studies and with the
outcomes from previous empirical and numerical studies.
Our results suggest that including transients in the H/V spectral ratio should not
worsen the estimation of the fundamental resonance frequency of a site. Since the
mechanism generating a peak in the H/V spectral ratio may depend on the source
position relevant to the thickness of the sedimentary column (Parolai and Galiana
9
Merino 2006), a full understanding of the composition of the wavefield is
mandatory before inverting the peak to obtain information about the subsoil
structure.
On the other hand, if including active sources can help in better highlighting
the resonance effect of Swaves, the use of transients, consistent with Mucciarelli
et al. (2003), might allow a H/V spectral ratio closer to the H/V spectral ratio of
small earthquakes to be obtained.
We believe that these issues are worthy of the attention of the seismological
community and further adhoc experiments should be planned in order to better
understand the composition of the noise wavefield.
Acknowledgements
K. Fleming kindly improved our English. Figures have been drawn using the GMT software
(Wessel and Smith 1991). R. Milkereit improved the figures.
References
Arai, H. and Tokimatsu, K. (2004). S-wave velocity profiling by inversion of
microtremor H/V spectrum. Bull. Seism. Soc. Am., 94, 1, 53-63.
Bard, P.Y. (1999). Microtremor measurements: a tool for site effect estimation? The
Effects of Surface Geology on Seismic Motion, eds. K. Irikura, K. Kudo, H. Okada and T.
Sasatani (Balkema, Rotterdam), 1251-1279.
Chatelain, J.C., Guillier, B., Cara, F., Duval, A.M., Atakan, K., Bard, P.Y., and The
WP02 SESAME team (2007). Evaluation of the influence of experimental conditions on
H/V results from ambient noise recordings. Bull. Earthquake Eng., doi: 10.1007/s10518-
007-9040-7.
Fäh, D., Kind, F., and Giardini, D. (2001). A theoretical investigation of average H/V
ratios. Geophys. J. Int., 145, 535-549.
10
Fäh, D., Kind, F., and Giardini, D. (2003). Investigation of local S-wave velocity
structures from average H/V ratios, and their use for the estimation of site effects. Journal
of Seismology, 7, 449-467.
Dziewonski, A., Bloch, S., and Landisman, M. (1969). A technique for the analysis of
transient seismic signals. Bull. Seism. Soc. Am., 59, 427-444.
Field, E.H. and Jacob, K. (1993). The theoretical response of sedimentary layers to
ambient seismic noise. Geophys. Res. Lett., 20-24, 2925-2928.
Horike, M., Zhao, B., and Kawase, H. (2001). Comparison of site response
characteristics inferred from microtremors and earthquakes shear waves. Bull. Seism. Soc.
Am., 81, 1526-1536.
Kim. D.S., and Lee, J.S. (2007). Propagation and attenuation characteristics of various
ground vibrations. Soil Dynamics and Earthquake Engineering, 19, 115-126.
Konno, K., and Ohmachi, T. (1998). Ground-motion characteristic estimated from
spectral ratio between horizontal and vertical components of microtremor. Bull. Seism. Soc.
Am., 88, 228-241.
Lermo, J., and Chávez-García, F.J. (1994). Are microtremors useful in site response
evaluation? Bull. Seism. Soc. Am., 84, 1350-1364.
Li, X.P., Zühlsdorf, L., and Liebhardt, G. (1996). Eliminating the effects of the surface
weathered layer by using an inverse transform function. Scientific Drilling, 5, 233-242.
McNamara, D.E., and Buland, R.P. (2004). Ambient noise levels in the continental
United States. Bull. Seism. Soc. Am., 94, 1517-1527.
Miller, G.F., and Pursey, H. (1955). On the partition of energy between elastic waves in
a semi-infinite solid. Proc., Royal Society, London, Series A, 55-69.
Mucciarelli, M. (1998). Reliability and applicability range of the Nakamura’s technique.
Journ. Earthq. Eng., 2, 4, 625-638.
Mucciarelli, M., Gallipoli, M.R., and Arcieri, M. (2003). The stability of the horizontal-
to-vertical spectral ratio of triggered noise and earthquake recordings. Bull. Seism. Soc.
Am., 93, 1407-1412.
Nakamura (1989). A method for dynamic characteristics estimations of subsurface using
microtremors on the ground surface. Q. Rept. Railway Technical Research Institute Japan,
30, 25-33.
Nogoshi, M and Igarashi, T. (1970). On the propagation characteristics estimations of
subsurface using microtremors on the ground surface. J. Seism. Soc. Japan, 23, 264-280.
11
Nogoshi, M and Igarashi, T. (1971). On the amplitude characteristics of microtremor
(part 2). J. Seism. Soc. Japan, 24, 26-40.
Parolai, S., Bormann, P. and Milkereit, C. (2001). Assessment of the natural frequency
of the sedimentary cover in the Cologne area (Germany) using noise measurements. Journal
of Earthquake Engineering, 5, 4, 541-56.
Parolai, S., Picozzi, M., Richwalski, S.M., and Milkereit, C. (2005). Joint inversion of
phase velocity dispersion and H/V ratio curves from seismic noise recordings using a
genetic algorithm, considering higher modes. Geoph. Res. Lett., 32, doi:
10.1029/2004GL021115.
Parolai, S., and Galiana-Merino, J.J. (2006). Effect of transient seismic noise on
estimates of H/V spectral ratios. Bull. Seism. Soc. Am., 96, 228-236.
Parolai, S., Richwalski, S.M., Milkereit, C., and Fäh, D. (2006). S-wave velocity
profiles for earthquake engineering purposes for the Cologne Area (Germany). Bull.
Earthquake Eng., 4, 65-94.
Parolai, S. (2008). Determination of dispersive phase velocities by Complex seismic
trace Analysis of Surface Wave (CASW). Soil Dynamics and Earthquake Engineering in
revision.
Picozzi, M., Parolai, S., and Richwalski, S.M. (2005) Joint inversion of H/V ratios and
dispersion curves from seismic noise: Estimating the S-wave velocity of bedrock. Geoph.
Res. Lett., 32, doi: 10.1029/2005GL022878.
René, R.M., Fitter, J.L., Forsyth, P.M., Kim, K.Y., Murray, D.J., Walters, J.K., and
Westerman, J.D. (1986). Multicomponent seismic studies using complex trace analysis.
Geophysics, 51, 6, 1235-1251.
Richart, F.E., Hall, J.R., and Woods, R.D. (1970). Vibration of soils and foundations.
Englewood Cliffs, NJ: prentice-Hall, 1970.
SESAME European project (2004). Guidelines for the implementation of the H/V
spectral ratio technique on ambient vibrations measurements, processing and interpretation.
Deliverable D23.12. http://sesame-fp5.obs.ujf-grenoble.fr/SES_TechnicalDoc.htm.
Scherbaum, F., Hinzen, K.-G., and Ohrnberger, M. (2003). Determination of shallow
shear-wave velocity profiles in Cologne, Germany area using ambient vibrations. Geophys.
J. Int., 152, 597-612.
Shapiro, N.M., and Singh, S.K. (1999). A systematic error in estimating surface-wave
group velocity dispersion curve and a procedure for its correction. Bull. Seism. Soc. Am.,
89, 1138-1142.
12
Strollo, A., Bindi, D., Parolai, S., and Jäckel, K.-H. (2008a). On the suitability of 1 s
geophone for ambient noise measurements in the 0.1-20 Hz frequency range: experimental
outcomes. Bull of Earthquake Eng (in press).
Strollo A., Parolai, S., Jäckel K.-H., Marzorati, S., and Bindi, D. (2008b). Suitability of
short-period sensors for retrieving reliable H/V peaks for frequencies less than 1 Hz. Bull.
Seism. Soc. Am. (in press).
Tamura, S. (1996). Comparison of body and Rayleigh wave displacements generated by
a vertical point force on a layered elastic medium. Proc., 11 th World Conf. on Earthquake
Engineering, Paper No. 1722.
Taner, M.T., Koehler, F., and Sheriff, R.E. (1979). Complex seismic trace analysis.
Geophysics, 44, 6, 1041-1063.
Taniguchi, E., and Sawada, K. (1979). Attenuation with distance of traffic-induced
vibrations. Soils and Foundations, 19, 2, 16-28.
Tokimatsu, K., and Tamura, S. (1995). Estimation of local site conditions in Kushiro city
based on array observation of microtremors. Proc., 3rd Int. Conf. on Recent Advances in
Geotechnical Earthquake Engineering and Soil Dynamics, 2, 599-602.
Wessel, P., and Smith, W.H.F. (1991). Free software helps map and display data. EOS
Trans. AGU, 72,41,441,445-446.
13
Figure 1: Vertical (top), radial (middle) and transverse component (bottom)
recordings of 5 series of alternated vertical and horizontal energization at 5 m
distance from the sensor. Left panels highlight (gray) signal windows selected for
the vertical energization. Central panels highlight (gray) signal windows selected
for the transverse energization. Right panels show (gray) signal windows
containing signals generated by vertical and horizontal sources. Please note that
each transient in the panels is composed by five blows.
14
Figure 2: Same as Figure 1, but for sourcetosensor distance of 100 m.
15
Figure 3: Vertical and horizontal component Fourier Spectra (left and central
panels, respectively) and H/V spectral ratio (right panels) calculated for increasing
sourcetoreceiver distances (from top to bottom). Only results for vertical sources
are shown here. Average spectra and H/V for stationary noise are shown with
white lines and the 95% confidence interval is shown by light gray. Black lines
indicate average spectra and H/V results for windows including transients. In the
16
central panel, gray indicates the radial component, while dark gray indicates the
transverse component.
Figure 4: As in figure 3 but for horizontal sources.
17
Figure 5: As in Figure 3, but for the signal generated by vertical and horizontal
sources.
18
Figure 6: The vertical (top panel) and the radial (second from the top)
component of ground motion generated by a vertical source at 5 m distance after
applying a Gaussian filter with a central frequency of 0.9 Hz. The reciprocal
ellipticity (third panel from the top) versus time. The phase difference versus time
(bottom panel). The white background indicates the time in which the vertical
sources operated. The histograms are calculated considering the data falling within
the area with white background.
19
Figure 7: Same as Figure 6, but for the radial and transverse components.
20
Figure 8: Same as Figure 6, but for the vertical and transverse components.
21
Figure 9: The spectral amplitude decay for 0.9 Hz versus distance (thin light
gray). The dashed black line indicates a R 2 decay. The thicker gray line indicates
a R1 decay. The dashed gray line indicates a R0.5 decay.
