You are on page 1of 19

This article was downloaded by: [UNAM Ciudad Universitaria]

On: 20 May 2014, At: 14:40


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Earthquake Engineering


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/ueqe20

Use of Ambient Noise: From Spectral


Amplitude Variability to H/V Stability
a a
Bertrand Guillier , Jean-Luc Chatelain , Sylvette Bonnefoy-
a b
Claudet & Ebrahim Haghshenas
a
Laboratoire de Géophysique Interne et de Tectonophysique ,
Université Joseph Fourier , Grenoble, France
b
International Institute of Earthquake Engineering and Seismology ,
Tehran, Iran
Published online: 29 Nov 2007.

To cite this article: Bertrand Guillier , Jean-Luc Chatelain , Sylvette Bonnefoy-Claudet & Ebrahim
Haghshenas (2007) Use of Ambient Noise: From Spectral Amplitude Variability to H/V Stability, Journal
of Earthquake Engineering, 11:6, 925-942, DOI: 10.1080/13632460701457249

To link to this article: http://dx.doi.org/10.1080/13632460701457249

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Journal of Earthquake Engineering, 11:925–942, 2007
Copyright © A.S. Elnashai & N.N. Ambraseys
ISSN: 1363-2469 print / 1559-808X online
DOI: 10.1080/13632460701457249

Use of Ambient Noise: From Spectral Amplitude


1559-808X
1363-2469
UEQE
Journal of Earthquake Engineering
Engineering, Vol. 11, No. 6, June 2007: pp. 1–27

Variability to H/V Stability

BERTRAND GUILLIER, JEAN-LUC CHATELAIN, and


Use of Ambient
Guillier et al. Noise

SYLVETTE BONNEFOY-CLAUDET
Laboratoire de Géophysique Interne et de Tectonophysique, Université Joseph Fourier,
Grenoble, France
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

EBRAHIM HAGHSHENAS
International Institute of Earthquake Engineering and Seismology, Tehran, Iran

The study of the variation over time of both spectral amplitudes and H/V curves, has been performed
on three different sites, two close to cities and one in the countryside, during periods varying from
week to over a month. It demonstrates the robustness of the H/V technique to give consistent peak
frequency values. In particular, H/V peak frequencies, either fundamental (f0) or natural (fx, x•1),
are not affected by weather nor the level of human activity. However, while fundamental H/V peak
amplitudes are stable, they proved rather unstable for natural (secondary) peak. Spectral amplitude
curves are very variable but follow human activity cycles from week-week end and day-night varia-
tions down to a very small scale, such as lunch breaks. Finally, the frequency limit between
anthropic noise and natural noise, commonly taken at 1 Hz, is not straightforward and is varying
from site to site from 0.7–0.8 Hz up to 2–3 Hz.

Keywords Ambient noise; Spectral amplitude; H/V spectral ratio; Site effect

1. Introduction
One of the main issues in site effect studies using H/V from ambient vibration recordings
[Nakamura, 1989, 1996] is the stability over time of the H/V frequency and amplitude.
The question is: “on a defined site, are the fundamental frequency of the soil (f0) and its
amplitude stable over time?,” i.e., is the H/V result dependant upon the instant of the
recording (day-night, week-weekend, meteorological activity . . . ) or not?
No noticeable H/V variations over time have been evidenced at various time
scales: week [Volant et al., 1998; Cara et al., 2003], month [Mucciarelli and
Monachesi, 1998], and year [Bour et al., 1998; Mucciarelli et al., 2003; Di Giacomo
et al., 2005]. However, only the week scale studies are based on a continuous data set,
when the others were performed with a time lap of one recording per day over a month
or a year.
In this article we present a study of the behavior over time of both spectral amplitudes
and H/V curves from ambient noise recordings at three sites, characterized by different
geotechnical characteristics, i.e., various H/V peak frequencies (f0) conditions, and human

Received 19 September 2005; accepted 9 May 2007.


Address correspondence to Bertrand Guillier, Laboratoire de Géophysique Interne et de Tectonophysique,
Université Joseph Fourier, BP 53, 38041 Grenoble cedex 9, France; E-mail: Bertrand.Guillier@bondy.ird.fr

925
926 Guillier et al.

activity types: Grenoble, Tehran, and La Gibonnière. Ambient noise recordings time
series from a week to over a month were used. The studied frequency domain {0.2–20
Hz} has been divided into three sub-domains, as proposed in SESAME [2003]: the low-
frequency domain (LF) corresponding to frequencies below 1 Hz, the medium-frequency
domain (MF) corresponding to frequencies between 1 and 4 Hz, and the high-frequency
domain (HF) for frequencies over 4 Hz.
These sites have been chosen for the following reasons:
• the three sites present a peak in the LF domain;
• the presence of secondary peaks: one at Grenoble and La Gibonnière, two at
Tehran;
• human activity varies from heavy at Grenoble and Tehran to light at La Gibonnière.
In this study, local time is used in order to control the human factor. For the three sites, we
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

fix as night the timetable between 7 PM and 7 AM the next day, and as day the timetable
from 7AM to 7 PM the same day. Following the local way of living, in France (Grenoble
and La Gibonnière), the weekend is set from 12-midnight on Friday to 12-midnight on
Sunday, while in Tehran it is set as the whole Friday.
Let’s make it clear that the aim of this article is to test the stability over time of both
spectral and H/V amplitudes and peaks from “classic” ambient noise recording, i.e., not,
for example, from triggered noise. It is by no means at all to study the underlying structure
at the chosen sites. Except for La Gibonnière, the sites have been studied by the above-
mentioned authors, whose results are used only to confirm that we get the same fundamen-
tal frequency peaks. Anyway, any serious basin structure study cannot be based on the
results of a single recording site only. Also, there is no attempt in this study to compare H/
V peak amplitudes to earthquake amplification.

2. Data Acquisition, Processing and Analysis

2.1. Data Acquisition


In Grenoble and La Gibonnière, we used 5-s Lennartz LE-3D seismometers, when in
Tehran we used a 30-s Guralp CMG-40T. In the SESAME instrument tests [SESAME,
2002], the Lennartz 5-s is described as acceptable and even “best performing sensor,” and
the Guralp CMG-40T as acceptable for H/V studies. The recording devices were CitySark
[Chatelain et al., 2000] at Grenoble and La Gibonnière, and a Reftek 72A07 in Tehran,
respectively, ranked 2nd and 4th over 13 tested stations in SESAME tests [Guillier et al.,
2002; SESAME, 2002] for H/V studies. Therefore, the equipment used in this study
should not raise any question or problem concerning the results.
At Grenoble, ambient noise has been recorded over a 1.5 month (02/07/2002-18/08/
2002), during 15 min every h, at a 100-Hz sample rate, giving a total set of 891 recordings.
Tehran recordings were taken from a study carried out over the whole city, from
which we selected a station (DAR) with continuous 1-week recording (168 1-h record-
ings), at a 100-Hz sample rate, from Monday 15/04/2002 (0 AM) up to Sunday 21/04/
2002 (12-midnight).
At La Gibonnière, during a period of a week, from Monday 29/03/2004 (0 AM) up to
Sunday 04/04/2004 (12-midnight), 168 15-minutes recordings were obtained with a one
hour time lag, at a 100-Hz sample rate.
Finally, in Grenoble the sensor was installed in an open space without shelter, while
in the two other cases the seismometer was installed inside a building, i.e., protected
against adverse weather conditions.
Use of Ambient Noise 927

2.2. Data Processing


Data from each site have been processed using a LGIT software [SESAME, 2003], follow-
ing the same procedure and using the same processing parameters as in SESAME [2003]:
• offset removal: the mean of the entire recorded signal is deducted from each sample
value. Neither sensor nor station corrections have been applied to the signal;
• determination of stable windows, in the 30–40 s range, using an anti-trigger, with a
1-s STA, a 30-s LTA, and STA/LTA ratio thresholds of 0.3 and 2.0;
• for each window, a cosine taper with a length of 5% is applied on both side of the win-
dow signal of the vertical (V), North-South (NS), and East-West (EW) components;
• for each window, a FFT is applied to the signal of the three components to obtain
the three spectral amplitudes, to which a Konno and Ohmachi smoothing [1998] is
applied, with a bandwidth of 40 and an arithmetical average;
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

• spectral ratios (NS/V, EW/V, and average (NS, EW)/ V) are computed.
One could argue that the influence of selecting stable windows through the anti-trigger algo-
rithm in conditions so different as big cities and a small hamlet should be important, as in the
countryside its is much less probable to have transients due to human activity. However, (1)
the H/V experiment has been conducted following the “classic” way, i.e., suppressing tran-
sients during processing, and (2) nearby transients (few tens of meters) are not useful nor
meaningful to study the behavior in time of ambient vibration spectral amplitudes.
For the sake of simplicity, only the velocity spectral amplitudes of the vertical com-
ponent and the average (NS, EW)/ V ratio are presented in the figures.

2.3. H/V Curves Reliability


Data has been processed in blind conditions, i.e., neither the waveform nor the window-
ing, automatically defined in the program, have been checked. Once the spectral ampli-
tudes and spectral ratios are obtained, the general conditions for H/V curve reliability are
tested according to the SESAME user guidelines [SESAME, 2004]. These tests allow
evaluating if a curve is suitable for use in the study or not: if one of the SESAME criteria
is not fulfilled, the corresponding H/V and spectral amplitude curves are rejected. The
SESAME H/V reliability tests are:
• for a given window length (lw), each H/V peak frequency (f0, f1 . . . ) should be over
10/ lw, i.e., the H/V curve below10/ lw cannot be used;
• for each H/V peak frequency, the total of significant cycles over all selected win-
dows should be equal to or greater than 200;
• over each H/V curve, the standard deviation should be less than 2.0.

2.4. H/V Peaks Reliability


H/V peaks were selected according to the SESAME recommendations [Koller et al.,
2004; SESAME, 2004], in which a “clear H/V peak” is defined when at least 5 of the 6
following criteria are fulfilled:
• H/V peak amplitudes (f0, f1 . . . ) should be equal to or greater than 2;
• between f0 and f0/4, there must be a frequency which amplitude is less than half the
amplitude of the frequency peak;
• between f0 and 4*f0, there must be a frequency which amplitude is less than half the
amplitude of the frequency peak;
928 Guillier et al.

• the frequency standard deviation (σf) for each peak (f0, f1 . . . ) should be less than a
frequency dependant threshold, detailed in SESAME [2004];
• the amplitude standard deviation (σA( f0)) of a H/V peak (A0, A1 . . . ) should be less
than a threshold (frequency dependant) fixed in SESAME [2004];
• the peak should appear at the same frequency (within ± 5%) on the H/V curves cor-
responding to the mean plus and minus one standard deviation.
Finally, as to determine if a given H/V peak has a natural or anthropic (machinery, indus-
try, etc.) origin, each H/V peak is tested using the Randomdec method [Huang and Yeh,
1999], as suggested in SESAME [2004] as being a very effective way to check the nature
of the peak, and which has been successfully used in buildings [e.g., Dunand et al., 2002]
as well as on soil [Haghshenas, 2005]. The Randomdec technique applied to the ambient
vibration recordings allows to derive the “impulse response” around the frequency of
interest: if the corresponding damping is very low (e.g., below 5%), an anthropic origin
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

may be assumed almost certainly, and the frequency is not considered in the interpretation
[SESAME 2004].
If there is a single peak on the H/V curve, it is called the “fundamental frequency
peak,” while if there is more than one peak the one with the lowest frequency is called
“fundamental frequency peak” and the others “natural peaks.”
It is only after these tests that the fx peaks were considered as real and suitable. The
peak frequency fx and its amplitude Ax are automatically averaged from the [fx-Ax] pairs
obtained from the H/V curve of each single window [SESAME, 2003].

2.5. H/V Peaks Coherency


The stability over time of both the fx frequency value and its corresponding Ax amplitude
is investigated through a similarity Student t-test, that has been applied to analyze the
degree of similarity between H/V peaks [e.g., Giampiccolo et al., 2001; SESAME, 2003].
It has been decided to use the Student t-test because on one hand this statistical method
deals with small sample number (less than 30, which is the case of H/V window number)
and on the other hand we are not comparing samples randomly selected, as H/V peaks
come from an identical processing and are changing only by the time of recording.

3. Grenoble Results
Recordings on the Grenoble basin have been performed at Le Versoud (10 km, ENE from
Grenoble, 450,000 inhabitants), located on top of a narrow and deep sedimentary basin
(10 × 8 km, 500–600 m depth), made of late quaternary post-glacial deposits overlaying
Jurassic marls and a marly limestone bedrock [Dietrich et al., 2001; Cornou et al., 2003].
Previous H/V studies demonstrate a strong amplification in the 0.2–3 Hz frequency band
[Lebrun, 1997; Lebrun et al., 2001], which has been confirmed by numerical simulations
of ground motion (up to 1.6 Hz) demonstrating basin-edge-induced waves [Cotton et al.,
1998; Bard et al., 1999; Cornou et al., 2003]. Again, the goal of this work is not to study
the basin structure, and this short description is given only as to show that our results fit
with these more detailed structural studies, especially in terms of H/V frequency.

3.1. Spectral Amplitude Behavior


At Le Versoud, spectral amplitudes clearly exhibit three distinct peaks, a very marked one
at 2.8 Hz and two secondary peaks at 0.32 and 0.60 Hz (Fig. 2A, left).
Use of Ambient Noise 929

1
A 6
f0 Grenoble, f0 f
Grenoble H/V 0(V) = 0.33 Hz D0(V) = 5.71%
5
H/V AMPLITUDE

Normalized amplitudes
0 f0(NS) = 0.35 Hz D0(NS) = 5.34%
4 f0(EW) = 0.34 Hz D0(EW) = 5.08%
3 f1 –1
0 10 20 30
2 1
Grenoble, f1 f
1(V) = 5.05 Hz D1(V) = 0.08%
1
0 f1(NS) = 5.05 Hz D1(NS) = 0.02%
0
1 10 f1(EW) = 5.05 Hz D1(EW) = 0.02%
FREQUENCY (Hz) –1
0 0.5 1 1.5 2
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

Time (second)

B 1
Normalized amplitudes

Tehran, f0 f0(V) = 0.87 Hz D0(V) = 6.21%

0 f0(NS) = 0.82 Hz D0(NS) = 5.99%


3
Tehran H/V f0(EW) = 0.85 Hz D0(EW) = 5.55%
H/V AMPLITUDE

f0 –1
2
f2 0 5 10
f1 1
Tehran, f1 f
1(V) = 1.32 Hz D1(V) = 0.68%

1 0 f1(NS) = 1.35 Hz D1(NS) = 0.42%


f1(EW) = 1.35 Hz D1(EW) = 0.85%
–1
0 0 2 4 6
1 10
1
FREQUENCY (Hz) Tehran, f2
f2(V) = 4.76 Hz D2(V) = 8.78%
0 f2(NS) = 4.67 Hz D2(NS) = 6.12%
f2(EW) = 4.67 Hz D2(EW) = 5.49%
–1
0 1 2
Time (second)

C 7 1
“La Gibonnière” H/V La Gibonnière, f0 f
6 0(V) = 0.62 Hz D0(V) = 7.18%
H/V AMPLITUDE

f1
Normalized amplitudes

5 0 f0(NS) = 0.60 Hz D0(NS) = 5.42%


4 f0(EW) = 0.62 Hz D0(EW) = 5.45%
f0 –1
3
0 5 10 15
2 1
La Gibonnière, f1
1 f1(V) = 3.02 Hz D1(V) = 5.59%
0 0 f1(NS) = 2.91 Hz D1(NS) = 5.24%
1 10 f1(EW) = 2.86 Hz D1(EW) = 5.21%
FREQUENCY (Hz)
–1
0 1 2 3 4
Time (second)

FIGURE 1 H/V curves at the three studied sites with definition of the H/V peak (left) and
the tests for the origin of the peak using a random decrement method [right; Dunand et al.,
2002] for the three sites: Grenoble (A), Tehran (B), and La Gibonnière (C). To be consid-
ered from natural origin, the tested frequency peak fx must exhibit a damping factor over
5% on each channel [Dx(component)]. For the damping graphics (right): thick lines represent
the vertical components, thin lines the N-S components, and dashed lines the E-W compo-
nents. For Grenoble (A), f0 has a natural origin while the f1 peak is human generated. For
Tehran (B), f0 and f2 are natural peaks when the f1 peak has a human origin. For La Gibon-
nière (C), both f0 and f1 are natural peaks.
930 Guillier et al.

VELOCITY SPECTRAL AMPLITUDE Low frequency site Daily spectral A


6 Night (Grenoble - France) amplitude
10 variations (Z) VELOCITY
Day
24/07/2003 SPECTRAL
5 AMPLITUDE

FREQUENCY (Hz)
6.7
4

1
2

0 0.0
1 10 0h 3h 6h 9h 12h 15h 18h 21h
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

FREQUENCY (Hz)
7
B
VELOCITY SPECTRAL AMPLITUDE

Night Low frequency site Weekly spectral


(Grenoble - France) amplitude
6 Day 10 VELOCITY
variations (Z)
Week-end 22–28/07/2002 SPECTRAL
AMPLITUDE
FREQUENCY (Hz)

5
7.2
4

3
1
2

0 0.0
1 10 MON TUE WED THU FRI SAT SUN
FREQUENCY (Hz)
Monthly spectral
amplitude
C
Low frequency site
10 (Grenoble - France) variations (Z) VELOCITY
19–07/2002 – 17/08/2002 SPECTRAL
W1 W2 W3 W4 W5 AMPLITUDE
FREQUENCY (Hz)

7.5

0.0
F S S M T W T F S S M T W T F S S M T W T F S S M T W T F S

FIGURE 2 Absolute spectra at Grenoble on the vertical component. A: daily set of abso-
lute spectra curves (left) and daily historical absolute spectra (right). B: weekly set of
absolute spectra curves (left) and weekly historical absolute spectra (right). C: monthly
historical absolute spectra. The black lines on the historical graphs indicate the main
energy release frequency.

3.1.1. Daily Variations. Over a day, the 2.8 Hz peak spectral amplitudes exhibit a strong
heterogeneity, the amplitude varying from 0.5–6.8 (i.e., a factor of 14), while below 1 Hz
the curves are relatively homogeneous (Fig. 2A, left; factor of 3.6). On the entire
Use of Ambient Noise 931

frequency band, spectral amplitudes are systematically higher during the day than during
the night. The spectral amplitude variation can be divided into five periods (Fig. 2A,
right):
• 0 AM – 7 AM: during this time period, spectral amplitudes are homogeneous,
although an MF increase is observed at the end of the period;
• 7 AM – 12 AM: a very strong MF increase is observed, with a strong jump between
7 and 8 AM, and a maximum between 10 and 11 AM. In the LF domain, amplitudes
are steady and weak, while in the HF domain, amplitudes increase strongly
(+500%) but remain weak;
• 12 AM – 1 PM: MF amplitudes crumble, while they are steady in the LF and HF
domains;
• 1 PM – 6 PM: the main behavior is a brutal increase of the amplitudes in the MF
domain between 1 and 2 PM, with a maximum between 2 and 3 PM;
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

• 6 PM – 12-midnight: the amplitude behavior is progressively going back to what it


was at the beginning of the day (0–7AM).
The maximum spectral amplitude at 2.8 Hz is the main peak only during daytime
(maximum human activity) and becomes a secondary peak at night (minimum human
activity) during which the main energy release is found at 0.32 Hz (Fig. 2A, right, black
curve).

3.1.2. Weekly Variations. Along a week (Fig. 2B, left), the 2.8 Hz peak spectral ampli-
tudes show a very strong variability 0.35–7.2 (factor of 20). Below 1 Hz and above 8 Hz,
curves are relatively homogeneous, even though a slight difference can be observed
between workdays and weekend, by a factor of about 4. Furthermore, on the whole fre-
quency band, spectral amplitudes during the day are systematically higher than at night,
and workdays spectral amplitudes are systematically higher than during the weekend at
the same hour.
Based on the 2.8 Hz peak, the weekly spectral amplitudes variation can be divided
into two different stages (Fig. 2B, right):
• from Monday to Friday (workdays) it is observed 5 times the daily variations (see
above);
• from Saturday to Sunday (weekend), the variations are weaker, and the 2.8 Hz peak
amplitudes are lower than during workdays.
As for daily variations, the 2.8 Hz peak is the main one only during the day period and
becomes secondary at night (Fig. 2B, right). Additionally, during the weekend the domi-
nance of the 2.8 Hz peak is strongly contested by the 0.32 Hz peak.

3.1.3. Monthly Variations. Monthly spectral amplitudes (Fig. 2C) show the same pat-
terns as the daily and weekly alternations as described in the previous sections. Interest-
ingly, amplitudes on Thursday of week 5 (w5) do not present the “normal” shape of a
“normal” Thursday (such as during w2, w3, and w4), which is explained by the fact that
this Thursday (15/08) is a national holiday and therefore human activity is comparable to
the weekend activity. Moreover, the next day (16/08) amplitudes are lesser than observed
on previous Fridays, but higher than during a weekend or a national holiday. This can be
explained by the peculiar French way of living consisting in not going to work during a
day squeezed between a holiday and a weekend (commonly termed as a “bridge”) for a
significant part of the population, reflecting in a higher human activity than during the
weekend but lower than a normal workday.
932 Guillier et al.

3.2. H/V Behavior


The H/V curve exhibits two clear peaks at f0 = 0.35 Hz and f1 = 5 Hz (Fig. 1A, left). Both
H/V curves and peaks complied with the reliability tests (Table 1). Application of the ran-
dom decrement technique demonstrated that while the f0 peak had a natural origin, the f1
peak was of anthropic origin (Fig. 1A, right), and therefore the latter has been discarded
from further considerations. The whole data set satisfied the required conditions with a
single reliable and usable peak f0, taken as the fundamental frequency peak of the soil.
Some recordings are missing, due to station power problems (week w3, Fig. 2C and 3C).

3.2.1. Daily Variations. Over a day (Fig. 3A, left), H/V curves are strongly stable over
the whole studied frequency range, especially in the LF domain in which the fundamental
frequency peak (f0) of the Grenoble basin is well expressed (Fig. 3A, right). Moreover, the
f0 peak amplitude is steady over time. The only variations are observed around 5 Hz (f1),
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

but as this peak has been shown to be of anthropic origin it is not taken into account for the
H/V curve interpretation. Finally, over a complete day (24 h), there are no observable H/V
day-night variations, neither at the fundamental frequency of the site (f0 = 0.35 Hz) nor on
the rest of the curve.

3.2.2. Weekly Variations. Over a week (Fig. 3B, left), although the amplitude of the f0
peak varies from 3.9–7.3, Student-t tests confirm its relative stability, when the variation
of the f0 frequency value is not meaningful. Additionally, in the LF and HF domains, the
observed variations are meaningless. In general (Fig. 3B, right), H/V curves are similar
and do not exhibit any difference neither between day and night nor between workdays
and weekends.

3.2.3. Monthly Variations. The f0 fundamental frequency peak does not present any
significant variation over a month period, neither in frequency value nor in amplitude
(Fig. 3C), whether diurnal-nocturnal, weekly or monthly. The only clear variation is
observed around 5 Hz (f1) reflecting the influence of a waste incinerator located about
4 km NW from the site. This factory, which worked preferentially at night, was very
active on the first 10 nights only. Afterwards, while still active it had less impact on the
H/V curves.

TABLE 1 Values for the reliability of the H/V spectral ratios using both Koller et al.
[2004] and SESAME [2004] guidelines. lw mini.: minimal useful windows length (in sec-
onds) from guideline; lw real: used window length; nw mini.: minimal useful number of
windows from guidelines; nw real: used number of windows; cw mini.: minimal useful
number of cycles from guidelines; cw real: used number of cycles; fx: frequency peak
Frequency lw mini. (s.), nw cw lw real nw cw real
(Hz) lw = 10/fx mini. mini. (s.) real cw = nw*lw*fx
Grenoble, f0 0.35 28.57 10 200 38 16 212
Grenoble, f1 5.05 2.00 10 200 38 16 3040
Tehran, f0 0.90 11.11 10 200 30 70 1890
Tehran, f1 1.35 7.40 10 200 30 70 2835
Tehran, f2 4.50 2.22 10 200 30 70 9450
La Gibonnière, f0 0.60 16.67 10 200 30 17 306
La Gibonnière, f1 2.80 3.57 10 200 30 17 1428
Use of Ambient Noise 933

6
f0
Low frequency site Daily H/V A
Night (Grenoble - France) variations
5 Day 10 24/07/2003
H/V
AMPLITUDE
4 6.0
H/V AMPLITUDE

FREQUENCY (Hz)
f1

2 1

1
f0

0.0
0
1 10
0h 3h 6h 9h 12h 15h 18h 21h
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

FREQUENCY (Hz)
7 f0 Low frequency site Weekly H/V B
Night (Grenoble - France) variations
Day 10 22–28/07/2002
6
Week-end H/V
AMPLITUDE
5 7.3
FREQUENCY (Hz)
H/V AMPLITUDE

4 f1

3
1

1 f0

0 0.0
1 10 MON TUE WED THU FRI SAT SUN
FREQUENCY (Hz)

Low frequency site Monthly H/V C


(Grenoble - France) variations
10 19–07/2002 – 17/08/2002
W1 W2 W3 W4 W5
H/V
AMPLITUDE
FREQUENCY (Hz)

7.5

0.0
F S S M T W T F S S M T W T F S S M T W T F S S M T W T F S

FIGURE 3 Spectral ratio at Grenoble. A: daily set of H/V curves (left) and daily histori-
cal H/V variation (right). B: weekly set of H/V curves (left) and weekly historical H/V
variation (right). C: monthly historical H/V variation. The black lines on the historical
graphs indicate the main H/V frequency peak.

3.3. Grenoble Conclusions


Low-frequency spectral amplitudes (< 1Hz) are not markedly affected neither by human
activity nor weather conditions, although strong climatic variations (wind, rain,
934 Guillier et al.

rainstorms, etc.) occurred, during the 1.5–month, when the experiment took place and the
sensor was not sheltered. Above 1 Hz, spectral amplitudes exhibit important diurnal-noc-
turnal fluctuations, as well as workweek-weekend variations, which easily permit to char-
acterize in detail changes of human activity over time, down to levels such as work-
timetable, periods of rest, lunch breaks, etc.
To the contrary, H/V curves do not exhibit notable variation of the soil fundamental
frequency value (f0). Even though weak amplitude variations of the fundamental peak (f0)
are observed, they are not meaningful enough to define a real difference, and in any case
are not correlated to climatic phenomena, such as strong wind or rain. The level of energy
released over the studied frequency range (climatic phenomena, human activity), has no
impact on the H/V curves, especially for the soil fundamental peak, which remains stable
in both frequency value (f0) and amplitude (A0).
Hence, while spectral amplitudes exhibit a strong variability over time, it is not the
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

case for the f0 H/V peak, which, moreover, is not affected by climatic conditions even
though the sensor was installed on the ground without any protection. However, in our
case “strong wind” is only a rough description, as no measurement of the wind speed was
performed. The wind case should be handled with care, especially for low-frequency sites.
It is better to shelter the sensor and even to avoid recording under strong wind conditions,
especially with nearby structures (building, trees, and even tall grass).

4. Tehran Results
Recordings at Tehran (9,000,000 inhabitants) have been performed in the northern center
of the city, on a Plio-Quaternary basin (about 50 km NS × 70 km EW) without good geom-
etry definition and characterized by stiff soil (coarse grained alluvium), overlying a bed-
rock at an unknown depth [SESAME, 2004; Haghshenas, 2005]. The aim of this study is
not to precise the characteristics of this basin but to check out H/V peak stability over time.

4.1. Spectral Amplitude Behavior


Spectral amplitude curves exhibit a strong peak close to 10 Hz and a smaller one, not
always present, around 2.5 Hz (Fig. 4A, left). In the HF domain, amplitudes around 10 Hz
are systematically higher during daytime than during nighttime. Amplitude fluctuations
are stronger at night (amplitude from 7–30) than during daytime (25–50). During the
weekend (night and day), the amplitude around 10 Hz is of the same magnitude as at
workday nights with the same variability. Below 4 Hz, only weak day-night and work-
week-weekend changes are observed.
The maximum energy release is close to 10 Hz (Fig. 4A, right), frequency that is very
consistent over time even though one hour per night the main energy release switches bru-
tally to 2.5 Hz. Based on the daily spectral amplitude fluctuations (Fig. 4A, right), days
can be divided into three time periods:
• 0 AM–5 AM, characterized by weak human activity in the HF domain;
• 6 AM–3 PM, with a strong human activity over 4 Hz, with a maximum around
7 AM and a second one between 11 AM and 1 PM;
• 4 PM–12-midnight, HF activity decreases progressively down to the level of the
0 AM – 5 AM period.
• Weekly variations clearly evidence two periods of human activity (Fig. 4A, right):
• Saturday to Thursday, with a strong day-night alternation and a maximum activity
in the morning;
Use of Ambient Noise 935

50 Weekly spectral
Night
Low to high frequency site
(Tehran - Iran)
amplitude A
variations (Z)
Day 10
15–21/04/2002
VELOCITY SPECTRAL AMPLITUDE

40
Week-end SPECTRAL
AMPLITUDE
50

FREQUENCY (Hz)
30

20
1

10

0.0
0
MON TUE WED THU FRI SAT SUN
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

1 10
FREQUENCY (Hz)
3

Night
f2 Low to high frequency site Weekly H/V B
(Tehran - Iran) variations
Day f1 10 15–21/04/2002
Week-end f0 H/V
AMPLITUDE
2 f2
FREQUENCY (Hz)

3.0
H/V AMPLITUDE

1
1 f0

0 0.0
1 10 MON TUE WED THU FRI SAT SUN
FREQUENCY (Hz)

FIGURE 4 Absolute spectra and spectral ratio (H/V) at Tehran. A: weekly set of absolute
spectra curves (left) and weekly historical absolute spectra (right). B: weekly set of H/V
curves (left) and weekly historical H/V variation (right). The black lines on the historical
graphs indicate either the main energy release frequency (spectral amplitude graph) or the
main H/V peak frequency value (H/V graph).

• Friday, the HF activity is way weaker than during the other days of the week
because in Iran Friday is the nonworked day of the week.

4.2. H/V Behavior


At this site, three peaks are observed on the H/V curves at f0 = 0.9 Hz, f1 = 1.3 Hz, and f2 =
4.7 Hz (Fig. 1B, left). All the H/V curves passed the reliability tests (Table 1). However,
the random decrement technique shows that the f1 clearly exhibits an anthropic behavior
(Fig. 1B, right). Otherwise, the complete data set complied with the requirements defined
in the previous section, and the H/V curve exhibits two clear frequency peaks of natural
origin: f0 (fundamental peak) and f2 (natural peak).
The coherency of the f0 peak (0.9 Hz) has been tested on whole H/V curves (Fig. 4B,
left). Student-t tests performed on f0 demonstrate the stability over time of this frequency
peak (Fig. 4B, right, black line), as well as that of f0 amplitudes, even though they vary
from 1.3–2.25. The same tests performed on the f2 peak (Fig. 4B, right, black line),
936 Guillier et al.

demonstrate that its frequency value is very robust while its amplitude is unstable, even
for a same period of a day, night or weekend, and hence in a week. In the MF domain, it is
noticeable that there are slight alternations with low H/V values from 10 PM–5 AM.
However, according to the SESAME criteria [Koller et al., 2004; SESAME, 2004],
the amplitude of a H/V peak has to be over 2.0 to be considered as usable. Therefore, at
the Tehran site, the f0 peak (0.9 Hz) would be taken as a peak only 20% of the time, as
during 80% of the time its amplitude is less than 2.0. This “instability” in f0 amplitude
cannot be linked to human activity, as 22% of the f0 peaks are less than 2.0 during the day,
19% at night and 21% during the weekend. Meanwhile, the instability of the f2 peak
amplitude (4.7 Hz) may be correlated to human activity. This is confirming the results of
numerical modeling [Bonnefoy-Claudet et al., 2006], which demonstrate that the ampli-
tude of secondary H/V peaks is sensitive to the position and intensities of the sources.
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

4.3. Tehran Conclusion


Below 1 Hz, spectral amplitudes fluctuations cannot be correlated to human activity, but
are probably linked to weather changes. Above 1 Hz, as in the Grenoble case, there is a
very clear day-night alternation as well as a workday-weekend change both clearly linked
to human activity.
H/V frequency values of both f0 and f2 peaks are robust, while, however, only the f0
amplitude is stable in spite of the climatic fluctuations highlighted by the spectral ampli-
tude variations.
As in the Grenoble case, the f0 from H/V curves is stable over time, not influenced
neither by meteorological conditions nor by human activities, to the contrary of the spec-
tral amplitude, which is strongly linked to the latter. Our observations also confirm that
natural peak amplitudes are sensitive to the position and intensity of ambient vibration
sources.
The results at the Tehran site raise the question of peak determination. Following the
SESAME recommendations [Koller et al., 2004; SESAME, 2004], it is needed to fulfill
5 of 6 criteria, so as long as there is five criteria fulfilled, there is no problem. If, however,
the amplitude criterion [A0 > 2] is one of the five fulfilled, f0 might either be considered as
the fundamental peak or discarded, depending upon when the recording has been per-
formed. In the latter case, if the peak is not observed at neighboring sites it is recom-
mended to perform other recordings at different times of the day.

5. La Gibonnière Results
The La Gibonnière site is located in the Perche region, in the center-west of France. This
hamlet (10 inhabitants) is located in the countryside (Dorceau), about 5 km away from
Rémalard city (1,700 inhabitants) and 80 km East from Le Mans (250,000 inhabitants).
Recordings were performed about 10 m away from a house and 5 m from a 20-m tall tree,
200 m from a forest. The soil is composed of about 10 m of green to red compact clays
(alteration), overlaying 70–90 m of cretaceous sands themselves overlaying cretaceous
limestones.

5.1. Spectral Amplitude Behavior


The spectral amplitude curves exhibit a very strong peak close to 0.35 Hz and two smaller
ones close to 0.7 and 10 Hz (Fig. 5A, left). Both smaller peaks are not always apparent
(Fig. 5A). For both low-frequency peaks (0.35 and 0.7 Hz), amplitudes are similar during
Use of Ambient Noise 937

10

Night
Medium frequency site
(La Gibonnière - France)
Weekly spectral A
9 amplitude
Day 10 variations (V)
VELOCITY SPECTRAL AMPLITUDE

8 29/03/2004 – 04/04/2004
Week-end SPECTRAL
AMPLITUDE
7
10

FREQUENCY (Hz)
6

4
1
3

0 0.0
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

1 10 MON TUE WED THU FRI SAT SUN


FREQUENCY (Hz)
7

Night
f1 Medium frequency site Weekly H/V B
(La Gibonnière - France) variations
6 Day 10 29/03/2004 – 04/04/2004
Week-end H/V
5 AMPLITUDE
7.0
FREQUENCY (Hz)
H/V AMPLITUDE

f1
4
f0

3
1
2
f0

0 0.0
1 10 MON TUE WED THU FRI SAT SUN
FREQUENCY (Hz)

FIGURE 5 Absolute spectra and spectral ratio (H/V) at La Gibonnière. A: weekly set of
absolute spectra curves (left) and weekly historical absolute spectra (right). B: weekly set
of H/V curves (left) and weekly historical H/V variation (right). The black lines on the his-
torical graphs indicate either the main energy release frequency (spectral amplitude graph)
or the main H/V peak frequency value (H/V graph).

daytime and during nighttime. The maxima observed over the weekend are much higher
than those of the rest of the week, which is explained by a strong windstorm that occurred
during the weekend of the April 3–4. Over 1 Hz, daytime amplitudes are generally stronger
than nighttime ones, and weekend amplitudes are of the same order of size as during each
day of the week. The 0.35 Hz peak (Fig. 5A, right, black line) does not exhibit any fre-
quency value fluctuation over the entire week, even during the weekend when the wind-
storm occurred. In the HF domain (Fig. 5A, right), it is easy to distinguish the day-night
fluctuation, while there is no week-weekend alternation. This is probably due to the fact
that the site is far away from any important urban center and close to a residence, leading
to the stability in the diurnal-nocturnal fluctuation over the 7 days of the week.

5.2. H/V Behavior


The H/V curves (Fig. 1C, left) exhibit two peaks at f0 = 0.6 Hz and f1 = 2.9 Hz. For the
entire data set an average of 57% of each record has been used, i.e., 510 over 900 s, due to
938 Guillier et al.

the windowing selection. Out of the 168 recordings, 167 satisfied the H/V curve reliability
tests (Table 1), one recording being rejected due to a standard deviation greater than 2.5 at
0.48 Hz. Both f0 and f1 H/V peaks satisfy the H/V peak reliability tests, and even though
the f1 frequency is slightly less than 5*f0, it has been considered as independent from f0.
Finally, the natural origin of both f0 and f1 frequencies has been confirmed with the ran-
dom decrement technique (Fig. 1C, right).
The f0 peak is stable around 0.6 Hz during the week, but shifts slightly towards 0.5 Hz
during the weekend (Fig. 5B). This variation is not meaningful according to Student-t
tests. It is probably the strong wind, coupled to nearby structures (house, trees, etc.), that
induces this shift in frequency. Likewise, Student-t tests confirm that the amplitude of the
f0 peak is stable over time, in spite of the strong spectral amplitude variability below 1 Hz.
For the f1 peak (2.9 Hz) Student-t tests demonstrate a strong robustness in frequency
(Fig. 5B) but a weakness in amplitude. Furthermore, neither diurnal-nocturnal nor week-
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

weekend fluctuations are observed (Fig. 5B, right).

5.3. La Gibonnière Conclusions


At this site, spectral amplitude variations are observed below 1 Hz, which may be linked
to weather conditions. As observed in Grenoble and Tehran, there is a diurnal-nocturnal
spectral amplitude alternation over 1 Hz, but there is no week-weekend difference.
The H/V f0 fundamental peak is consistent in frequency as well as in amplitude,
despite a small f0 variation and an important variation in spectral amplitudes during the
weekend. The f1 peak, as the Tehran f2 peak, is robust with respect to the frequency but
not with the amplitude, showing again that natural peak amplitudes are sensitive to the
position and intensity of ambient vibration sources.

6. Discussion
Stability of H/V over time has been studied, at a weekly scale, with continuous recording
by Volant et al. [1998] during one week and Cara et al. [2003] during 10 days, showing no
f0 variation. These studies are confined to the evaluation of weather conditions on H/V
curves and do not consider the influence of human activity. At a larger time-scale, Bour
et al. [1998], with 2 recordings at 1-year interval on 3 sites, found very stable H/V curves.
Mucciarelli et al. [2003] and Di Giacomo et al. [2005] present results over a 674-day and
244-day period, respectively, both getting “at least one noise record per day” from trig-
gered noise (defined by the authors as “man-made seismic waves strong enough to reach
the threshold set for earthquake detection”), i.e., much more energetic noise than that
recorded during “classic” H/V ambient noise studies. The main goal of these two studies is
a comparison between noise and earthquake results, although Mucciarelli et al. [2003]
evidence “a weak, negative correlation between the rainfall and fundamental frequency
and between the fundamental peak’s amplification and magnitude.”
Cara et al. [2003] argue for spectral amplitude variations due to the wind, which,
however, does not affect the f0 peak, but significantly alters H/V below 0.2 Hz. While
other studies [Mucciarelli, 1998; Mucciarelli et al., 2005] claim no visible influence of the
wind, provided that the sensor are well protected against it, our results show no direct
influence of the wind on H/V curves obtained with a sensor installed without shelter over
1.5 month, as also shown by SESAME [2004] and Chatelain et al. [2007], who in return,
both warn against indirect influence transmitted to the ground by features next to the
sensor, as also observed during the windstorm in La Gibonnière with a tree close to the
sensor.
Use of Ambient Noise 939

While they converge towards advocating that noise H/V is stable over time and show
possible influence of weather conditions (mainly rain and wind), none of the aforemen-
tioned studies deals with a comparison between H/V and hourly spectral amplitude behav-
ior linked to human activity as analyzed in this study.

6.1. Spectral Amplitude


Spectral amplitudes are clearly influenced by human activities, and to a lesser extent, by
weather conditions, in two different frequency bands.
Below 1 Hz, ambient vibration spectral amplitude variations may be correlated to
local climatic conditions such as during the weekend at La Gibonnière where the wind
influences spectral amplitudes up to 2–3Hz. Over 1 Hz, Whiters et al. [1996] found that
wind influences ambient level noise (spectral amplitudes) beyond 3–4 m/s.
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

Meanwhile, at Grenoble, if it is possible to observe changes in the spectral amplitudes


below 1 Hz (week w2 and w4; Fig. 2C), these changes are not propagated over 1 Hz.
Over 1 Hz, the three sites exhibit clear spectral amplitude variations that may be
directly related to human activity, as they reflect very precisely human life cycles (day-
night, week-weekend) at a city scale (Grenoble, Tehran) down to a very local scale, such
as a house (La Gibonnière). These variations are even precise enough as to determine
human activity down to a very fine level (lunch break, holidays, etc.).
However, the frequency band most impacted by human activity is site dependent,
since in Grenoble the maximum energy release is around 3 Hz while in Tehran and La
Gibonnière it is around 10 Hz. Surprisingly, the two sites located next to cities do not
behave the same way, and one of them behave in the same way as the countryside site,
which might be an indication on the type of city-related activities.
Finally, there is no clear downwards frequency limit of human activity, which influ-
ences spectral amplitude down to 0.7–0.8 Hz in Grenoble, (Fig. 2A; Fig. 2B), 1.5–2.0 Hz
in Tehran (Fig. 4A), and 4 Hz at La Gibonnière (Fig. 5A).

6.2. H/V Curves


Below 1 Hz, H/V peaks are very stable both in frequency values and amplitude, even
though spectral amplitude are unstable and linked to climatic conditions. For example, at
La Gibonnière the fundamental frequency peak (0.6 Hz) is stable in amplitude even
though the energy released at 0.6 Hz has changed by an order of 10 during a windstorm.
Over 1 Hz, H/V peaks are very robust in frequency values but unsteady in amplitudes,
although no clear relationship can be established between the amplitude of the peak and
the anthropic noise energy release at this frequency. This amplitude instability of natural
H/V peaks is certainly due to the sensitivity of the secondary peak to position and intensi-
ties of noise sources [Bonnefoy-Claudet, 2004, Bonnefoy-Claudet et al., 2006].

7. Conclusions
The main result of this study is that H/V fundamental (f0) and natural peak (fx, x•1) fre-
quencies from ambient noise recordings are extremely robust over time, independently of
both the frequency of the peak and the energy level released at the frequency of the peak,
i.e., whatever the level of energy of natural (wind, rain, etc.) and anthropic activities. In
particular, the wind does not influence the results, except when it acts on features coupled
to the ground next to the sensor (for example, a tree at La Gibonnière) and not when acting
directly without coupling on the recording equipment (sensor not sheltered in the Grenoble
940 Guillier et al.

case). However, while the amplitude of the H/V fundamental frequency (f0) is rather
steady over time, it is not the case for higher frequency natural peaks (fx, x•1), generally
located over 1 Hz. The amplitude variation of natural H/V peaks cannot be correlated to
human activity, and could be rather explained by the noise source locations and intensities
as proposed by Bonnefoy-Claudet et al. [2006]. Therefore, on the basis of our results,
ambient noise H/V f0 surveys are not biased by the moment in time factor: f0 may be
considered as reliable, as well as its amplitude to be used for relative comparisons. For
natural peaks (fx ; x•1), the frequency is reliable, but amplitude studies should be han-
dled with care.
Three other points are evidenced by this study. The limit between natural noise and
anthropic noise is generally taken at 1 Hz [e.g., Bonnefoy-Claudet et al., 2006]. Our study
evidences that this limit is site dependent, and may go down to 0.7–0.8 Hz or up to 2–3
Hz, slightly different from the commonly admitted 1-Hz threshold.
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

One of the SESAME [2004] peak definition criteria states that in order to define a
clear H/V peak its amplitude should be over 2.0. This criterion should not be applied in a
too conservative way, since, at the same site f0 peaks may have up and down amplitudes
around the fatidic limit. In the case of amplitudes a little bit below 2.0 observed on more
than one site, (1) complementary recordings should be performed, preferably in a different
time window, and (2) interpreted by comparison with neighboring sites.
Finally, spectral amplitude variations over time allow describing human life cycles
down to a very detailed scale. We plan, for further test, the use of this characteristic as an
objective complement to interview surveys conducted, for example, for establishing seis-
mic scenarios or risk mitigation studies.

Acknowledgments
We thank LGIT (Grenoble) and IIEES (Tehran) people that ably helped for field data col-
lection. We thank the two anonymous reviewers who accepted to spend time for construc-
tive criticisms. This study was supported by IRD, UJF (Grenoble), INSU, and IIEES, four
public research institutes. Support public research.

References
Bard, P.-Y., Mozco, P. and Kristek, J. [1999] “Finite-difference modeling of site effects in the
Grenoble basin,” XXII General Assembly of the International Union of Geodesy and Geophysics
(IUGG99), Birmingham, Session on strong motion: earthquake hazard and risk, 28 July 1999,
Abstract ST3/W/20–B3.
Bonnefoy-Claudet, S. [2004] “Nature du bruit de fond sismique: implication pour les effets de site,”
PhD Thesis, Joseph Fourier University, Grenoble (France), p. 241.
Bonnefoy-Claudet, S., Cornou, C., Bard, P.-Y., Cotton, F., Moczo, P., Kristek, J. and Fäh, D. [2006]
“H/V ratio: a tool for site effects evaluation. Results from 1-D noise simulations,” Geophysics
Journal International, 167, 827–837, doi: 10.1111/j.1365-246X.2006.03154.x
Bour, M., Fouissac, D., Dominique, P., and Martin, C. [1998] “On the use of microtremor record-
ings in seismic Microzonation,” Soil Dynamics and Earthquake Engineering, 17, 465–474.
Cara, F., Di Giulio, G., and Rovelli, A. [2003] “A study on seismic noise variations at Colfiorito,
Central Italy: implications for the use of H/V spectral ratios,” Geophysics Research Letters,
30(18), 1972, doi:10.1029/2003GL017807.
Chatelain, J. L., Guéguen, P., Guillier, B., Fréchet, J., Bondoux, F., Sarrault, J., Sulpice, P., and
Neuville, J. M. [2000] “Cityshark : A user-friendly instrument dedicaced to ambient noise
(microtremor) recording for site and building response studies,” Seismology Research Letters,
71, 698–703.
Use of Ambient Noise 941

Chatelain, J.-L., Guillier, B., Cara, F., Duval, A.-M., Atakan, K., Bard, P.-Y. and the WP02 SES-
AME team [2007] “Evaluation of the influence of experimental conditions on H/V results from
ambient noise recordings,” Accepted. Bulletin of Earthquake Engineering. accepted.
Cornou, C., Bard, P.-Y., and Dietrich, M. [2003] “Contribution of dense array analysis to basin-
edge-induced waves identification and quantification. Application to Grenoble basin, French
Alps (II),” Bulletin of the Seismological Society of America, 93, 2624–2648.
Cotton, F., Berge, C., Lemeille, F., Pitarka, A., Lebrun, B., and Vallon, M. [1998] “Three-
dimensional simulation of earthquakes in the Grenoble’s basin,” Second International Symposium
on the Effects on Surface Geology on Seismic Motion, 2, Balkema, Yokohama, 873–878.
Di Giacomo, D., Gallipoli, M. R., Mucciarelli, M., Parolai, S. and Richwalski, S. M. [2005]
“Analysis and Modeling of HVSR in the Presence of a Velocity Inversion: The Case of
Venosa, Italy,” Bulletin of Seismological Society of America, 95, 6, 2364–2372, doi: 10.1785/
0120040242.
Dietrich, M., Cornou, C., Bordes, C., Bard, P.-Y., and Lemeille, F. [2001] “Geophysical exploration
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

for site-effects assessment: borehole measurements and Vibroseis profiling in the Isère valley
near Grenoble, France,” March 25–30 Nice, France, 26th General Assembly of the European
Geophysical Society.
Dunand, F., Bard, P.-Y., Chatelain, J.-.L., Guéguen, PH., Vassail, T. and Farsi, M. N. [2002]
“Damping and frequency from Randomdec method applied to in situ measurements of ambient
vibrations: evidence for effective soil structure interaction,” Proceedings of the 12th European
Conference on Earthquake Engineering, London, September 9–13.
Giampiccolo, E., Gresta, S., Mucciarelli, M., De Guidi, G., Gallipoli, M. [2001] “Information on
subsoil geological structure in the city of Catania (eastern Sicily) from microtremor measure-
ments,” Annali di Geophysica 44(1), 1–11.
Guillier, B., Atakan, K., Duval, A.-M., Ohrnberger, M., Azzara, R., Cara, F., Havskov, J., Alguacil, G.,
Teves-Costa, P., Theodulidis, N., and the SESAME Project WP02-Team, [2002] “Influence of
instrumentation on H/V spectra of ambient noise,” EGS, 22–26 April, Nice, France.
Haghshenas, E. [2005] “Condition géotechnique et aléa sismique local à Téhéran,” PhD Thesis,
Joseph Fourier University, Grenoble (France), p. 288.
Huang, C. S. and Yeh, C. H. [1999] “Some properties of randomdec signatures,” Mechanical Sys-
tems and Signal Processing, 13(3), 491–507.
Koller, M. G., Chatelain, J.-L., Guillier, B., Duval, A. M., Atakan, K., Lacave, C., Bard, P.-Y., and
the SESAME participants [2004] “Practical user guideline and software for the implementation
of the H/V ratio technique on ambient vibrations: measuring conditions, processing method and
results interpretation,” 13th World Conference on Earthquake Engineering, Vancouver, B.C.,
Canada, August 1–6, Paper No 3132.
Konno, K. and Ohmachi, T. [1998] “Ground-motion characteristics estimated from spectral ratio
between horizontal and vertical components of microtremor,” Bulletin of the Seismological Soci-
ety of America, 88, 228–241.
Lebrun, B. [1997] “Les effets de site : étude expérimentale et simulation de trois configurations,”
PhD thesis, Université Joseph Fourier, Grenoble (France), p. 208.
Lebrun, B., Hatzfeld, D., and Bard, P.-Y. [2001] “A site effect study in urban area: experimental
results in Grenoble (France),” Pure and Applied Geophysics 158, 2543–2557.
Mucciarelli, M. [1998] “Reliability and applicability of Nakamura’s technique using microtremors:
an experimental approach,” Journal of Earthquake Engineering 4, 1–14.
Mucciarelli, M. and Monachesi, G. [1998] “A quick survey of local amplifications and their correla-
tion with damage observed during the Umbro–Marchesan earthquake of September 26, 1997,”
Journal of Earthquake Engineering 2, 325–337.
Mucciarelli, M., Gallipoli, M. R., and Arcieri, M. [2003] “The stability of the horizontal-to-vertical
spectral ratio of triggered noise and earthquake recordings,” Bulletin of the Seismological Society
of America, 93, (3), 1407–1412.
Mucciarelli, M., Gallipoli, M. R., Di Giacomo, D., Di Nota, F. and Nino, E. [2005], “The influence
of wind on measurements of seismic noise,” Geophysics Journal International, 161, 303–308,
doi:10.1111/j.1365-246X.2004.02561.x.
942 Guillier et al.

Nakamura, Y. [1989] “A method for dynamic characteristics estimation of subsurface using


microtremor on the ground surface,” Quaterly Report of the Railway Technology Research Insti-
tute, 30, 25–30.
Nakamura, Y. [1996] “Real-time information systems for seismic hazards mitigation,” Proceedings
of the 11th World Conference on Earthquake Engineering, Acapulco, Mexico.
SESAME [2002] “Instrument workshop 22–26 October 2001,” European Commission – Research
General Directorate Project No. EVG1-CT-2000–00026 SESAME, report D01.02, http://SES-
AME-FP5.obs.ujf-grenoble.fr.
SESAME [2003] “H/V technique: experimental conditions. Final report on Measurement Guideline
on WP02,” European Commission – Research General Directorate Project No. EVG1-CT-2000-
00026 SESAME, report D08.02, http://SESAME-FP5.obs.ujf-grenoble.fr.
SESAME [2004] “Guidelines for the implementation of the H/V spectral ratio technique on ambient
vibrations. Measurements, processing and interpretation,” European Commission – Research
General Directorate Project No. EVG1-CT-2000–00026 SESAME, report D23.12, http://SES-
Downloaded by [UNAM Ciudad Universitaria] at 14:40 20 May 2014

AME-FP5.obs.ujf-grenoble.fr.
Volant, P., Cotton, F., and Gabriel, J. C. [1998] “Estimation of site response using the H/V method:
applicability and limits of this technique on Garner Valley Downhole Array dataset (California),”
Proceedings of the 11th European Conference on Earthquake Engineering, Rotterdam.
Withers, M. M., Aster, R. C., Young, C. J. and Chael, E. P. [1996] “High frequency analysis of
seismic background noise as a function of wind speed and shallow depth,” Bulletin of the Seis-
mological Society of America, 86, 1507–1515.

You might also like