Geophys. J. Int. (2009) 176, 925–937 doi: 10.1111/j.1365-246X.2008.04020.

Site effect evaluation in the basin of Santiago de Chile using ambient

noise measurements

Sylvette Bonnefoy-Claudet,1 ∗ Stéphane Baize,1 Luis Fabian Bonilla,1

Catherine Berge-Thierry,1 Cesar Pasten,2 Jaime Campos,3 Philippe Volant1
and Ramon Verdugo2
1 Institute
for Nuclear Safety and Radioprotection, Fontenay-aux-Roses, France
2 Department of Civil Engineering, University of Chile, Chile
3 Department of Geophysics, University of Chile, Chile

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

Accepted 2008 October 24. Received 2008 October 24; in original form 2007 November 28

SUMMARY
We performed extensive ambient vibration measurements in the basin of Santiago de Chile
(Chile), and we look for testing the reliability of the horizontal-to-vertical amplitude spectra
ratio method (H/V ) as a tool to provide qualitative and quantitative information of site effects
in complex geological media. The interpretation of the H/V data was carried out conformably
to the SESAME project consensus criteria and outlines three major patterns: (1) clear peaked
H/V curves related to sharp underground velocity contrast; (2) H/V peak of low amplitude
and flat curve related to weak contrast and (3) broad H/V peak indicating the presence
of strong lateral variations of underground structure. H/V measurements, however, reveal
a discrepancy between the computed soil resonance frequencies and the expected building
resonance, therefore not leading to a straight interpretation of the intensity distribution derived

GJI Seismology
from observed damage to one storey houses in Santiago after the 1985 Valparaiso earthquake.
Indeed, the H/V technique mostly maps the first fundamental frequency; however, it fails to
show higher resonance modes. In the case of the city of Santiago, this method works well for
assessing the seismic hazard for high-rise buildings, but is questionable for smaller structures
as is the case of a great percentage of constructions in the city.
Key words: Site effects; South America.

ology. The Michoacan earthquake (Mexico, 1985, M w = 8.1) and

1 I N T RO D U C T I O N
the Great Hanshin-Awaji earthquake (1995 Kobe, M w = 6.9) are
Chile is one of the most seismic areas in the world. Before the outstanding examples of such site effects. In Chile, most of the
great Sumatra–Andaman earthquake (2004, M w = 9.1, Lay et al. seismological studies focus on the properties of the subduction pro-
2005), 46 per cent of the seismic energy spread out during the last cesses; however, a recent study has shown that local site effects may
century was in Chile. The largest events that occurred in Chile happen in Santiago de Chile (Cruz et al. 1993).
(e.g. 1960 Valdivia, M w = 9.5, 1985 Valparaiso, M w = 8, and It was the Valparaiso earthquake in 1985 that showed that the
1995 Antofagasta, M w = 8.1) are mostly due to the subduction of town of Santiago, Chile’s capital city (five million inhabitants), was
the Nazca Plate beneath the South American Plate (Pardo et al. prone to site effects (Monge 1986). The seismic intensities (MSK)
2002; Barrientos et al. 2004; Gardi et al. 2006). In central Chile, derived from observed damages to one storey adobe and one storey
the subduction raises a relative velocity of about 8 cm yr–1 in un-reinforced masonry houses during the 1985 earthquake were
a N78◦ E direction (De Mets et al. 1994) and generates mostly locally much higher (up to IX) than expected (VI) according to re-
interplate earthquakes. The recurrence of the largest events in the gional attenuation relations (Astroza & Monge 1987; Astroza et al.
Metropolitan region—Santiago de Chile area—is about 80 yr. 1993). The highest intensities were mainly reported in areas with
When an earthquake occurs, the most likely factor that con- poor soils conditions, in the fine-grained alluvial deposit and in the
tributes to damages and human injuries is the proximity of the ashes deposits (Astroza & Monge 1991). Local site effects due to
hypocentre to populated areas. However, ground amplification ef- “superficial” geological conditions were first suspected to happen
fect can become predominant at far distances due to the local ge- in which was previously thought to be a good foundation soil. De-
spite this ‘first-order’ correlation between superficial geology and
∗ Now damage, some discrepancies are notable: some parts of the most
at: Résonance Ingénieurs-Conseils SA, 21 rue Jacques
Grosselin, 1227 Carouge (GE), Switzerland. E-mail: sylvette.bonnefoy-
damaged areas are not built on soft sediments, and some soft sedi-
[email protected] ment areas did not show strong damage (Fernandez Mariscal 2003).


C 2008 IRSN 925
Journal compilation 
C 2008 RAS
926 S. Bonnefoy-Claudet et al.

Thus, the geometry of the basin and the geology of deep sedi- agation of seismic waves (body and surface waves) may involve
ments were suspected to be partly responsible for these amplified diffracted surface waves generated along slopes and discontinu-
effects. Such 2-D or 3-D basin structural effects on wave propa- ities. In such situation, the diffracted surface waves, in addition to
gation have been illustrated for long time through observational 1-D resonance effect, may be responsible for site amplification of
(Uetake & Kudo 1998) and numerical studies (Kawase 1996; Olsen seismic ground motion. The contamination of ambient noise wave-
& Archuleta 1996). field by edge-diffracted waves has been already outlined in previous
Past investigations were conducted in Santiago de Chile to eval- studies (Cornou 2002; Di Giulio et al. 2006).
uate site effects. An array of accelerometers (SMACH array) has The scope of this work is to verify the limitation of the H/V tech-
been deployed since 1989 and local site effects have been evalu- nique as a tool to provide qualitative and quantitative information
ated from strong ground motion records (Midorikawa et al. 1991; of site conditions when investigating 3-D structures. The study was
Cruz et al. 1993). Ambient vibrations measurements were also car- carried out in Santiago de Chile basin, which is known to be affected
ried out in the city to correlate the site effects evaluated by strong by 3-D geometrical and geological features.
ground motion (Guéguen 1994; Toshinawa et al. 1996). Ambient
vibration studies such as the H/V technique (Nakamura 1989) are
2 GEOLOGICAL SETTING
powerful tools to study local soil site effects. In addition to mapping
the resonance frequencies of the soil, the H/V technique has been The city of Santiago de Chile is located in the so-called Central

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

used in recent studies as an alternative tool to classical geophysi- Depression, which is a basin surrounded by the Main and Coastal
cal exploration for characterizing sediment infilling—typically, the ranges of the Andes (western 70◦ and southern 33◦ ). This basin
sediment thickness or average shear velocity—by taking advantage spans 80 km long and 30 km wide, and it is mainly elongated in the
of the simple relation (F 0 = β /4h) between the resonance frequency north–south direction. The basin is characterized by the coalescence
(f 0 ), the average shear wave velocity (β) and the average sediment of three main alluvial cones that drain the Andean Cordillera: cones
thickness (h) (Ibs-Von Seht & Wolhenberg 1999; Delgado et al. of Colina, Mapocho and Maipo (Fig. 1).
2000; Parolai et al. 2002). Although reliable in horizontal layering Due to the geomorphological setting, the basin is essentially filled
(Bonnefoy-Claudet et al. 2006a), this simple interpretation—that with alluvial sediments (Valenzuela 1978). These sediments have
assumes 1-D wave propagation—may be misleading in case of 2- extensively been characterized when looking for water and construc-
D or 3-D structures. Such examples have been shown not only tion material resources (Falcon Moreno et al. 1970; Morales-Jerez
experimentally (Uebayashi 2003; Gueguen et al. 2007) but also nu- 2002). They are composed of pebbles, gravels, clays and volcanic
merically (Guillier et al. 2006). This discrepancy is mainly due to ashes (Fig. 2). The pebbles and gravels are mainly located in the
the complexity of surface wave propagation in 2-D/3-D structures. eastern and southern part of the basin. Clayey material is mostly
When the basement geometry is far from 1-D structure, the prop- present in the north; whereas a transition zone is found in the centre

Figure 1. Left-hand panel: location of the Santiago de Chile basin (Chile). Right-hand panel: spatial distribution of the H/V measurement points in the basin
(black dots). Black lines show Santiago districts. Background colour displays the digital elevation model and brown lines indicate the elevation contour lines.
Blue lines exhibit major rivers.


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
 Site effects evaluation in Santiago de Chile 927

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

Figure 2. Left-hand panel: map of geological cross-sections in the North Western part of Santiago de Chile. Right-hand panel: Geological cross-section along
profile 6NS. After Iriarte-Diaz (2003).

Table 1. Mechanical properties (P- and S-waves velocity, density) at various loggers coupled to sensors Lennartz LE-3-D (three components
depths of the three main formations observed in the basin of Santiago velocimeters having a 5 s natural period). 264 measurements of
(gravels, clays and ashes; after Bravo 1992; Guéguen 1994). ambient noise were collected in the city (Fig. 1). The sampling
Depth (m) V P (m s−1 ) V S (m s−1 ) Density (g cm−3 ) rate was 125 Hz and the duration of recording 15 min, except
at some points where the recording duration was extended to 20
Gravels 0–20 950–1250 480–720 2–2.3
200 1300 2.1
minutes because of a strongly disturbed environment (dense and
500 2000 2.1 close car traffic, pedestrians, etc.). Some signals, however, were
still dominated by transients, thus 37 records were rejected. Finally,
Clays 0–20 400–1000 120–350 1.2–1.8
50 550 2.1
227 measurements of ambient noise were considered for the H/V
500 2000 2.1 processing.
The microtremor H/V technique was first proposed by Nogoshi
Ashes 0–20 400–900 180–450 1.15–1.7
& Igarashi (1971) and widespread by Nakamura (1989). It consists
in estimating the ratio between the Fourier amplitude spectra of
of the valley. In the Pudahuel district, a 40-m-thick layer of ashes the horizontal and the vertical components of ambient noise vibra-
(Pudahuel ignimbrite, also known as pumices) is known to seat at the tions. In this study, H/V ratios are calculated using the GEOPSY
top of the sedimentary column. The stiff pumice probably comes software (http://www.geopsy.org). H/V ratios are calculated for the
from a major eruption of the Maipo volcano, located at around frequency range 0.2–10 Hz, using 50 s time windows and removing
120 km to the southeast, at the head of the Maipo valley. time windows contaminated by transients. The objective is to assure
The geometry and depth of the basement geometry are con- the stationary of ambient vibrations, and to avoid the transients of-
strained by gravimetric data (Araneda et al. 2000), geophysical sur- ten associated with specific urban sources (footsteps, close traffic).
veys (Bravo 1992) and borehole data (Iriarte-Diaz 2003; Morales- The procedure to detect transients is based on a classical compar-
Jerez 2002). These data were used to construct a first-order geo- ison between the short term average (‘STA’, the average level of
logical model of the basin (Baize et al. 2006). Although the spatial signal amplitude over a short period of time, 1 s) and the long term
extension of each formation at depth is not well constrained; most average ‘LTA’ (the average level of signal over a much longer pe-
of the studies depict a complexity of the basin geometry, outlining riod of time, 30 s). Only windows with STA/LTA ratio between 0.3
very sharp lateral and vertical variations of units. and 2 are kept for the H/V computation. For every selected time
In addition, the estimation of P- and S-wave velocities of the three window, the Fourier amplitude spectra are smoothed with a Konno
main formations (stiff sediments, unconsolidated sediments and & Ohmachi (1998) filter using a coefficient of 40 for the bandwidth.
pumice) are inferred from seismic refraction (Bravo 1992; Guéguen The quadratic mean of the horizontal amplitude spectra is used here.
1994) and summarized in Table 1. The shear wave velocity in the The final H/V ratio and the associated standard deviation are ob-
rock substratum is about 2000 m s–1 (Bravo 1992). tained by averaging the H/V ratios from all windows. The standard
deviations on H/V ratio curves are estimated considering the arith-
metic average for all individual logarithm of H/V ratio computed
3 H / V A M B I E N T V I B R AT I O N S D ATA for each time window. See Atakan et al. (2004) for further details.
Extensive ambient noise measurements were performed in the basin The field experiment and the parameters of calculation fol-
of Santiago from April 2005 to Spring 2007. The ambient noise lowed the recommendations of the SESAME consortium (Bard &
vibrations were recorded using Cityshark I and Cityshark II data SESAME-Team 2005) to verify the reliability of results:


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
928 S. Bonnefoy-Claudet et al.

(1) The length of time windows (Lw) should be larger to 10 times in Santiago: grassy ground and pavements with different sizes of
the period of interest (i.e. the period corresponding to the H/V ratio paving stones. The sensors were placed at the soil surface over a
peak frequency Fhv): Lw > 10 × Fhv. short distance of about 5 m. The H/V curve recorded on the thin
(2) It is recommended to compute H/V ratio over a sufficient layer of grass is flat and shows low standard deviation (Fig. 4c),
number of windows to assess the average H/V curve. The number whereas the H/V curves recorded on pavements exhibit very large
of windows (Nw) should be set up according to H/V peak frequency standard deviation for frequencies below 1 Hz (Fig. 4b shows a
and windows length: Lw × Nw × Fhv > 200, the value 200 was H/V curve observed at only one kind of pavement) and reproduce
arbitrarily chosen by the SESAME consortium. well the large uncertainties observed on actual H/V curve shown in
(3) An acceptably low level of scattering between all windows Fig. 4(a). These observations strongly suggest that a poor soil-sensor
is needed. The standard deviation of the mean H/V curve should coupling has a large influence on the shape of the H/V curve at low
thus remain low over a frequency range at least equal to (0.5 × Fhv, frequencies. Nevertheless, more systematic measurements should
2 × Fhv). It is recommended that the standard deviation should be be carried out to provide a final explanation. Note that this strong
lower than a factor of two times (for Fhv > 0.5 Hz)—or a factor of disturbance does not influence the H/V curve for frequencies higher
three times (for Fhv < 0.5 Hz)—the mean H/V curve. than 1 Hz.
The average H/V curves were systematically analysed with re-

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

spect to the processing parameters following the SESAME rec- 4 D I S T R I B U T I O N O F T H E H / V C U RV E S
ommendations. At the end, 199 reliable H/V curves fulfilled these
criteria; only these data will be considered in the next sections. 4.1 Topology of the H/V curves
Fig. 3 shows an example of ambient noise record and its associated
H/V curve. Detailed analysis of the H/V data measured in Santiago indicates
Among the 199 reliable H/V curves, 39 data exhibits H/V that the shape of the H/V curves is not uniform: 47 H/V curves
curves showing large standard deviation for frequencies below 1 Hz (24 per cent of reliable data) do not show any peak (Fig. 5a), whereas
(Fig. 4a). The records corresponding to these data are widespread 152 H/V curves (76 per cent of reliable data) show at least one peak.
over the city of Santiago and do not show any spatial correlation. In total, we count 187 H/V peaks which can be separated in two
The very large amplitude of the standard deviation suggests that groups: clear peaks and those with no easy peak identification. The
its origin is not linked to geological or geometrical effects. Such criteria to qualify how neat a peak is are based on the amplitude
large standard deviation associated to a clear trend may be related of the H/V peak, its relative amplitude value with respect to other
to strong wind during the recording (Mucciarelli et al. 2005; Chate- peaks in other frequency bands and the relative value of the standard
lain et al. 2008). However, in the case of Santiago, this hypothesis deviation. We test the clarity of the H/V peaks in Santiago using
has to be discarded since there was no wind at all during the mea- the following recommendations proposed by SESAME guidelines
surements. On the other hand, since the measurements were made (Bard & SESAME-Team 2005).
in urban area and over great distances, a bad soil-sensor coupling (1) The H/V peak amplitude (Ahv) should be higher than 2.
may be responsible for the bad definition of these H/V curves. We (2) There exists one frequency (f − ), lying between Fhv/4 and
tested this hypothesis by simultaneously recording ambient noise at Fhv, such that Ahv/A(f − ) > 2, with A(f − ) the H/V ratio amplitude
four kinds of surfaces, which are representative of the soil coverage at the frequency f − .

Figure 3. (a) Example of ambient seismic noise records: vertical (Z), north–south (N) and east–west (E) components. Greyed areas depict the time windows
selected for the H/V ratio computation. (b) Corresponding H/V curve (black line) and the associated standard deviation (greyed surface).

Figure 4. (a) Example of observed H/V curve showing a large standard deviation at low frequency. H/V ratios computed at two kind of soil: (b) one kind of
pavement, which is typical of Santiago de Chile, and (c) on grass. The distance between the two former measurements was about 5 m, and the ambient noise
was recorded simultaneously by two Lennartz 5 s. See the legend of H/V curve (Fig. 3).


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
 Site effects evaluation in Santiago de Chile 929

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

Figure 5. Topology of the H/V curves observed in Santiago de Chile area: (a) flat curves, (b) curves showing a clear peak, which satisfy the SESAME criteria,
and curves showing unclear peaks which do not fully satisfy the SESAME criteria: peaks of low amplitude (c) and broad peaks (d).

(3) There exists one frequency (f + ), lying between Fhv and 4 × units are characterized by stiff sediments with a rather high S-wave
Fhv, such that Ahv/A(f + ) > 2, with A(f + ) the H/V ratio amplitude velocity at depth (1350 m s–1 ; Table 1). The absence of neat peak
at the frequency f + . on H/V curve in this area is thus very likely due to the local under-
(4) The standard deviation should be lower than a frequency ground structure, which may not exhibit any sharp velocity contrast
dependent threshold (see Bard & SESAME-Team 2005 for further at any depth leading to low-to-moderate amplifications. The corre-
details). lation between the absence of peak on H/V curve and weak velocity
contrast has been already stressed in previous experimental and nu-
When the H/V peak fulfilled at least 3 among these 4 criteria, merical studies (Konno & Ohmachi 1998; Bonnefoy-Claudet et al.
then there is quasi-certitude that the site under study presents a large 2008).
velocity contrast at some depth and is very likely to amplify the The broad H/V peaks (grey squares in Fig. 6) are in great majority
ground motion. The H/V peak frequency Fhv can be considered as located next to topographic slopes and deep depressions. There is
a very reliable estimate of the fundamental frequency. In the case of also a swarm of broad peaks between Renca foothill and Pudahuel
Santiago, 77 H/V peaks satisfy the clarity criteria. Fig. 5(b) displays depression (in the centre of the basin). In addition to the broad
an example of such clear peak. The H/V peaks (110 data) that failed peaks, the other types of H/V data are also present in this area.
to fulfil the clarity criteria could be divided into two groups: peaks In this region, the rock basement rises to the surface at the Renca
with low amplitude (lower than 2) (Fig. 5c) and broad peaks (large hill (350 m above free surface) and reaches 500 m deep below the
amplitude over a broad frequency band; Fig. 5d). surface 5000 m away (in Pudahuel depression; see Figs 2 and 7).
The large variability of H/V data is thus most probably related to a
complexity of the ambient noise wavefield, which may be diffracted
4.2 Spatial distribution of the H/V curves
along sloping bedrock–soft sediment contacts or some sloping in-
The records corresponding to the clear H/V peaks are in great ma- ternal stratification in the sedimentary cover, both cases having a
jority located in the central and northern parts of the basin (black high velocity contrast. Evidence of such effects has already exper-
triangles in Fig. 6). They are distributed at the surface of soft sed- imentally been observed in Colfiorito basin, Italy, (Rovelli et al.
iments: unconsolidated alluvial deposits or stiff pumice deposits. 2001) and in Dinar basin, Turkey, (Yalcinkaya & Alptekin 2005) for
The shear wave velocity at the surface of these units is low (smaller instance.
than 350 m s–1 , Table 1) compared with the velocity in the bedrock To determine the origin of the broad peaks, we performed H/V
(about 2000 m s–1 ). This suggests the existence of strong velocity measurements in the centre of the basin, along a profile from the
contrast at depth between unconsolidated and consolidated under- Renca foothill (point P01) up to the trough of Pudahuel district
lying sediment. Clear H/V peaks in Santiago are thus most probably (450 m deep; point P10). Ten measurements were done over a
correlated with this sharp contrast (Bard 1998). distance of 4500 m (see location of the profile in Fig. 6, red line).
The flat H/V curves and the H/V curves with peaks of low am- Points P01–P09 are located on unconsolidated sediments (silts and
plitude are mainly located in the southern and eastern parts of the clays); point P10 is located on stiff pumice (Fig. 7). The duration of
basin, at the surface of the Santiago and Mapocho gravels (black and recording is 20 min and the H/V ratios are computed as described
grey circles, respectively, in Fig. 6). As shown in Section 2, these in Section 3. Fig. 8 shows spectrogram depicting the relationships


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
930 S. Bonnefoy-Claudet et al.

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

Figure 6. Spatial distribution of the various types of H/V data observed in Santiago: clear peaks (black triangles), peaks of low amplitude (grey circles), flat
curves (black circles) and broad peaks (grey squares). Crosses indicate the location of rejected ambient noise records and no reliable H/V data. See Fig. 5 for
an illustration of the H/V peak types. The black rectangle displays the location of map shown in Fig. 7. Red line indicates the position of the H/V profile.

between the H/V spectral ratio amplitude and location along the have already portrayed the influence of steep slope on the shape of
profile. The H/V peak frequency sharply decreases from point P01 H/V curve. The irregular shape of the basement can lead to focusing
(Fhv = 3.5 Hz) to P03 (Fhv = 0.4 Hz), then it steadily decreases or de-amplification and a loss of spatial coherency in the ambient
up to point P10 (Fhv = 0.35 Hz). In addition, point P03 shows a noise motion at the surface (Woolery & Street 2002; Cornou et al.
second peak having a smaller amplitude at 1 Hz and points P05– 2003). Taking into account the steep slope of the basement, the
P10 show a small bump between 1.5 and 2 Hz. Regarding the shape peculiar shapes of H/V curves (broad-bandwidth peaks) might be
of the H/V curve, point P01 and points P04 to P10 show clear probably related to the more complex wavefield (diffracted body and
peaks (at least for the main peak), whereas H/V peaks associated surface waves) generated near the edge of the basement slope.
to points P02 and P03 are much broader with lower amplitude.
Because all the data are located at the surface of fine-grained soils
(except P10 located on ashes), we assume that the discrepancy in
H/V peak shape might not be related to the geology. On the other 5 I M P L I C AT I O N S F O R S I T E E F F E C T S
hand, gravimetric data (Araneda et al. 2000) shows that points P02 ASSESSMENT
and P03 are located in the most rapidly varying thickness part of the Most examples reported in the literature indicate clear peaked H/V
profile; whereas the points P04–P10 are located in a gentler slope. curve for soft soils and almost flat curves for rock sites (Lermo
Recent experimental studies (Woolery & Street 2002; Uebayashi & Chavez-Garcia 1994; Fäh 1997; Bour et al. 1998; Duval et al.
2003) and numerical studies (Uebayashi 2003; Guillier et al. 2006) 1994). Of course the amplitude and the frequency of the peak vary


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
 Site effects evaluation in Santiago de Chile 931

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

Figure 7. Spatial distribution of the various types of curve H/V in the centre of the basin of Santiago. See Fig. 6 for the spatial location of the map. The colour
background shows the surface geology of the basin (after Valenzuela 1978).

Figure 8. (top) Depth estimation of rock substratum (black circles) given by gravimetric study (Araneda et al. 2000) at points P01 to P10 of the H/V profile
shown Fig. 6. (bottom) H/V spectral ratio amplitude as a function of distance along the profile.

from site to site. In the Santiago case, however, H/V curves exhibit a sites. In addition, under those circumstances, the SESAME rec-
more complex pattern, without any prominent peak, or with several ommendations also claim that the amplification of ground motion
distinct bumps. One question immediately arises after this: is it occurs at the H/V peak frequency. We look for testing this for-
possible to infer the site resonance frequencies from fuzzy H/V mer statement for the Santiago case study by comparing the H/V
data? peak frequencies with fundamental frequency inferred from earth-
quake data. Horizontal-to-vertical spectral ratio method (Lermo &
Chavez-Garcia 1993) was applied to earthquake data recorded at
5.1 Clear peaks three strong ground motion stations operated by the University of
When the H/V peak is clear, then the site under study presents a large Chile (CL2, PCQ and RCDM; Fig. 9). At each frequency, only data
velocity contrast at some depth, and is very likely to amplify the with signal-to-noise ratio higher than 3 was selected. Low to mod-
ground motion (see for instance Konno & Ohmachi 1998; Bard & erate magnitude (3.9–5.3) earthquakes were recorded from 2003 to
SESAME-Team 2005). In the case of Santiago, this fact is supported 2006. Thus, the earthquake Fourier amplitude spectra may lack in-
by Fourier amplitude spectra computed from strong ground motions formation at low frequencies. As a consequence, we consider only
stations by Cruz et al. (1993). These authors report amplification frequencies above 0.6 Hz to compare ambient noise H/V peak fre-
factors of about 2.6 on pumice sites and 4 on unconsolidated-soil quencies and fundamental frequency inferred from earthquake data.


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
932 S. Bonnefoy-Claudet et al.

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

Figure 9. (top) Geological map (see legend Fig. 7) and location of the accelerometers stations (red hexagons) operated by the University of Chile (stations
CL2, PCQ, RCDM) and the Catholic University of Chile (UC1) and nearby H/V data (black dots). (bottom) Earthquake horizontal-to-vertical spectral ratio
computed at the strong ground motion stations (a) CL2, (b) PCQ, (c) RCDM, and (d) amplitude spectra ratio computed at UC1 (after Cruz et al. 1993). Black
thick lines and dashed lines indicate the average spectral ratio and the associated standard deviation, respectively. At each station, H/V ratio (thin black line
and greyed area) computed at the nearby ambient noise data are also shown.

The good agreement between the two former frequencies is depicted non-equally distributed at the surface of the basin. We have then
in Figs 9(a) and (b), outlining the robustness of the H/V method to decided to depict the spatial location of H/V peak frequencies with
provide the fundamental frequency in case of clear H/V peak, at least respect to their frequency: 0.3–0.5 Hz, 0.501–1.2, 1.201–5 and 5.01–
in Santiago. Fig. 9(c), however, illustrates well the limitation of the 10 Hz. The intervals of frequency have been arbitrary chosen so that
H/V method in case of broad band amplification of strong ground there is a roughly similar number of data for the first three intervals.
motion. The seismic amplification starts at the H/V peak frequency H/V peak frequencies are displaying in Fig. 10, and we can make
and remains large even though the H/V amplitude decreases. The the following comments:
H/V method provides thus the lower amplification frequency only,
as already observed by Lebrun et al. (2001). The reason for the • 80 per cent of data with peak frequency between 0.3–0.5 Hz
former limitation is not clear yet. It may be related to the horizontal are located above the thickest parts of the basin, from 550 to 250 m,
polarization of Rayleigh waves, which is involved in the origin of outlying the good correlation between low fundamental frequency
the H/V peak (Konno & Ohmachi 1998; Bonnefoy-Claudet et al. and thick sediment deposits given by gravimetric study, at least to
2006b). The influence of Love wave on the H/V ratio has been also a first order.
pointed out by Arai & Tokimatsu (2000) and Bonnefoy-Claudet • 75 per cent of data with peak frequency between 0.501–1.2 Hz
et al. (2008). are located above thinner sediment deposits, from 250 to 150 m,
In the scientific literature dealing with site effects assessment outlying once again the good correlation between fundamental fre-
and seismic microzonation, the usual practice is to plot the spatial quency and sediment thickness.
distribution of site amplification frequencies as contour map. This • 25 per cent data with peak frequency between 1.201–5 Hz
representation, however, may become inappropriate in the case of are located near basin edges and outcrops where the thickness of
Santiago because we have a low number of data (clear H/V peaks) the sediment deposit ranges between 50 to 100 m, indicating once


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
 Site effects evaluation in Santiago de Chile 933

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

Figure 10. Spatial distribution of the H/V peaks according to their frequencies: (red) 0.3–0.5, (blue) 0.51–1.2, (yellow) 1.21–5 and (green) 5.1–10 Hz. Circles
display frequencies for clear peaked H/V curves, whereas squares depict frequencies for peaks of low amplitude. The surface geology is shown in background,
see Fig. 7 for legend. The location of the accelerometer station PCQ is shown in black hexagon.

more the correlation between fundamental frequency and sediment

thickness. However, a large number of data (40 per cent) are lo- 5.2 Flat curves and H/V peaks of low amplitude
cated above deep depressions. These data cannot be related to the Flat H/V curves and H/V peaks of low amplitude are both located at
resonance of the deep sediment deposits (more than 250–400 m); the surface of dense sediment formations: Mapocho and Santiago
they are more likely linked to the resonance of superficial deposits. gravels. Such low amplitude H/V curves observed on no-rock site
Clear evidence can be found in the Pudahuel district where high fre- are a strong indication of stiff sediment underlain by bedrock and in
quency H/V peaks are distributed at the surface of the stiff pumice such situation one may expect a rather small amplification in case
deposits. of earthquake (Lachet & Bard 1994; Bard 1998; Konno & Ohmachi
• finally, data with peak frequency above 5.01 Hz are scattered 1998; Bonnefoy-Claudet et al. 2008). In the case of Santiago, this
at the surface of the basin, suggesting the presence of a topmost soft assumption is supported by looking at Fourier amplitude spectra
sedimentary layer (few metres thick). computed from strong ground motions by Cruz et al. (1993) in the
dense gravel area. For instance, the low level of site amplification at
Fig. 10 depicts thus the good agreement between the variations
UC1 accelerometer station (the amplitude ratio spectra lay between
of the sediment thickness and the H/V peak frequency. For example,
1.2 and 1.5) is depicted in Fig. 9(d).
the PCQ site is characterized by alternating soft layers of silts and
Following SESAME guidelines (Bard & SESAME-Team 2005),
clays from surface to about 15 m depth; with an S-wave velocity at
in the case of peaks of low amplitude the safest attitude is to refrain
the surface around 150 m s–1 . Below, dense gravels, with an S-wave
from deriving quantitative interpretations from the H/V curve. Al-
velocity estimated at 480 m s–1 (Table 1), reach the bedrock at a
though the SESAME recommendations set the threshold for the H/V
depth of about 100 m (Midorikawa et al. 1991; Toshinawa et al.
peak amplitude at 2, other studies are more relaxed. Bard (1998)
1996). The mean S-wave velocity (by means of average travel time)
suggests that an unclear bump on the H/V curve, followed by a clear
within the sediments is 360 m s–1 ; and the use of the simple relation
trough, may also be used for the estimation of the fundamental fre-
f 0 = β/4h leads to a site resonance frequency of about 0.9 Hz.
quency. In this sense, Rodriguez & Midorikawa (2002) also argue
This value agrees with the expected amplification frequency at this
that proscribe H/V estimations relying solely on the small spectral
site (inferred from microtremor and earthquake H/V data, see Fig.
ratios criterion becomes inadequate. In the case of Santiago, the
9b), suggesting that when the H/V peak frequency is clear, the H/V
spatial distribution of peak frequency of H/V data of low ampli-
method can be safely used as an alternative tool to classical geo-
tude, below 2 (colour filled squares in Fig. 10) agrees well with
physical exploration techniques for characterizing sediment infilling
the location of clear peaked H/V frequency; at least as a first order.
(typically sediment thickness if shear-wave velocity is known, see
This suggests that when peaks of low amplitude are spread among
for instance Ibs-Von Seht & Wolhenberg 1999; Delgado et al. 2000;
clear peaked curves, it may be possible to use the peak frequency
Parolai et al. 2002).


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
934 S. Bonnefoy-Claudet et al.

as a proxy of site resonance frequency. However, one may keep in erable number of houses in Santiago. The intensity data for the
mind that further experiments should be performed before system- Valparaiso earthquake were estimated based on damages report of
atically deriving quantitative information from H/V peaks of low one storey adobe and one storey masonry houses (Astroza et al.
amplitude. 1993). The determination of the intensity degree (in MSK) is based
on the statistical behaviour of damages reported in 288 sectors (of
5.3 Broad peaks one square kilometre) of Santiago city and for (1) only one type of
construction of adobe houses and (2) two types of construction of
Broad H/V peaks are very likely related to a complex wavefield due unreinforced masonry houses (Astroza & Monge 1991). In total, the
to significant 2-D or 3-D variations of the underground structure, number of damaged houses is 16 538 for adobe houses and 21 827
see Section 4. The propagation of microtremor seismic waves may for un-reinforced masonry houses. The vulnerability might thus be
involve diffracted waves (body and surface waves) generated along statistically the same for all adobe houses and for all unreinforced
slope and discontinuities. In such situation, the 1-D wave propaga- masonry houses. Most of the observed damage was concentrated in
tion properties are no longer reliable and the origin of H/V ratio areas with poor soils conditions located in the northwest of the city,
cannot be simply explained by the propagation of waves in a 1-D especially in the fine-grained alluvial deposits (Quilicura, Renca
medium. and Cerro Navia districts), in the stiff pumices (Pudahuel and Lo
For a practical use of H/V ratio, it can be difficult to pick the Prado), and in the transition zone (Quinta Normal; Fig. 11). Local

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

H/V peak frequencies in case of broad peaks (see for instance site effects due to geological conditions were thus first suspected to
Fig. 5d). Picking H/V peak frequency in such cases may lead to bias happen. However, some discrepancies are notable. For instance, a
the actual fundamental site frequency. Recent findings of numerical cluster of damage is also reported in consolidated sediment (grav-
simulations show 1-D resonance frequency underestimated by H/V els), in the southern of the city (Puente Alto district), whereas areas
peak frequencies (up to 80 per cent) at sites with steep underground in unconsolidated sediment report less damage (Conchali and Huer-
slopes (Guillier et al. 2006). Unfortunately, the earthquake data col- churaba districts; Fernandez Mariscal 2003). The surface geology
lection in Santiago is not suitable to quantitatively investigate the cannot thus fully explain the distribution of damage. The hypothesis
shift in resonance frequency inferred from microtremor and strong of lateral propagation, in addition to 1-D resonance effect, responsi-
ground motion data. Although there is no evidence of large devia- ble for the site amplification of seismic ground motion in Santiago
tion on actual data, we strongly suggest not inferring fundamental may be appropriate here. However, clear evidence of such effects
frequency from unclear H/V peak without having earthquake data has to be analysed in more details before drawing any conclusion.
simultaneously. The damage pattern in urban areas during an earthquake depends
on the characteristics of the event and on the interaction between
site response and vulnerability of the exposed structures. During
6 DISCUSSION the last 15 yr several efforts have been made to correlate dam-
The 1985 March 03 Valparaiso earthquake (M s 7.8, 17 Km deep, age with surface geology properties estimated from ambient noise
33.13◦ S–71.87◦ W), was destructive enough to damage a consid- measurements. Several papers present the comparison between the

Figure 11. Comparison between the H/V data and the distribution of the felt intensities (MSK) in Santiago after the Valparaiso 1985 earthquake (Astroza &
Monge 1987). Intensities were evaluated from one-storey—adobe and masonry—houses.


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
 Site effects evaluation in Santiago de Chile 935

fundamental frequency and corresponding H/V amplitude level and deeper strata and upper sedimentary layer, with respect to the uncon-
damage distribution or macroseismic intensities after significant solidated sediments (alluvial or ashes deposits). One should expect
earthquakes (Theodoulidis et al. 2008 and references therein). Some small amplification of strong ground motion during shaking at these
of the studies indicate that the H/V results are correlated with the sites. In addition, we show that, in the case of Santiago, H/V data
spatial distribution of damage, especially when damage variation exhibiting peaks of low amplitude (below 2) may also indicate the
is mostly controlled by near-surface geology. In the case of Santi- soil resonance frequency. These results suggest revising the quality
ago, during the 1985 Valparaiso earthquake, there was considerable criteria, stated by SESAME guidelines (Bard & SESAME-Team
damage to small-sized adobe structures, whereas the many high 2005), about the H/V peak amplitude threshold to assess the relia-
rise structures in the 10–20 storey range performed well with minor bility of the H/V estimate. However, further work will be needed to
damage to finishes and non-structural features (Wyllie et al. 1986). answer this debate.
Surprisingly, in the area of the highest intensities, the compari- (3) Finally, we show, in agreement with previous studies
son between the H/V peak frequencies and damage on small-size (Uebayashi 2003), that broad H/V peaks might be an indication
houses reveals a discrepancy; although the H/V peak frequencies of the presence of a dipping underground interface between softer
do not exceed 4—5 Hz, much damage was observed on small- and harder layers at depth. It is likely that the underground structure
size houses, which have an expected natural frequency higher than of the site under study exhibits significant lateral variations, which
5 Hz. could lead to significant 2-D or 3-D effects. Such broad peak H/V

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

This discrepancy contradicts previous studies showing the corre- curves are indeed observed on many basin or valley edges, and we
lation between the H/V data and the observed damage distribution. strongly suggest not inferring resonance frequency from such peaks
Here, the low frequency H/V peaks are linked to the deep veloc- without having earthquake data simultaneously.
ity contrast at the basement and do not reflect the resonance of
superficial soil layer. See, for instance, the comparison between With reference to site effect assessment in Santiago de Chile, the
microtremor and earthquake H/V curves observed at the PCQ ac- meticulous analysis of H/V data emphasizes three different seis-
celerometer station (Fig. 9b): strong ground motion indicates site mic responses. The first is defined by stiff sediments (Mapocho and
amplification at 5.75 Hz whereas the microtremor H/V ratio remains Santiago gravels) where no large site amplification is expected. The
flat. Conversely, the origin of the discrepancy could be due to other second is characterized by soft sediments (unconsolidated alluvial
processes than site effects that could have contributed to the damage deposits and stiff pumice) where important ground motion ampli-
and they are not captured by the noise measurements (vulnerability fication is very likely to occur. The third relies on the geometry of
of structures, for instance). Beyond the possible explanations, this the stratification, in particular the bedrock inclination, where site
study shows the limits of the H/V technique in providing informa- amplification could be expected. Indeed, in such situation, the gen-
tion that could be correlated to the damage distribution for the 1985 eration of edge-diffracted waves might be expected. These areas are
Valparaiso earthquake. The H/V method could not be used alone mostly located near basin edges and next to underground slopes.
to assess a comprehensive seismic hazard zoning in urbanized area However, further investigations are needed to quantify the seismic
characterized by a large variety of buildings (from small-size houses amplification in such cases.
to high buildings), such as Santiago de Chile. For a practical use of H/V ratio and to ensure the reliability and
the robustness of the H/V technique, we propose to carefully anal-
yse the H/V data before mapping the soil resonance frequencies and
7 C O N C LU S I O N S doing microzonation in urban areas. The peculiar shape of the H/V
curves is a reliable estimate of the underground site features. First,
Extensive ambient vibrations measurements (264 data) were per-
it provides a rough estimation of the strength of the impedance con-
formed in the basin of Santiago de Chile. We seek to verify the
trast: clear H/V peaks depict the presence of a strong underground
possibilities and limitations of the H/V spectral ratios technique
contrast at depth, whereas unclear peaks of low amplitude—or flat
(Nakamura 1989) to provide qualitative and quantitative informa-
curves—depict a low impedance contrast. Second, the broad width
tion of site conditions. The interpretation of the H/V curve was car-
of the H/V peaks might be an indication of the presence of strong
ried out conformably to SESAME project consensus criteria (Bard
underground lateral variations. The results presented in this paper
& SESAME-Team 2005). According to these criteria, we first ver-
suggest using a dense spatial sampling of data and picking H/V
ified the curve reliability and then checked the clarity of the H/V
frequencies only in areas showing a large proportion of clear H/V
peak. The application of the above criteria to the Santiago dataset
peaks.
results in 199 usable data showing 47 sites with flat H/V curves and
The scrutiny of H/V data set recorded in Santiago allows us to
152 sites with peaked H/V curve.
keep only reliable data. This work outlines the need to carefully
The detailed analysis of microtremor data outlines three major
analyse the H/V data before picking the peak frequency and infer-
patterns of H/V curves: (1) clear peaked curves, (2) flat H/V curves
ring quantitative estimation for microzonation purpose. The use of
or curves showing peaks of low amplitude (namely below 2) and
automatic (blind) zoning based on microtremor H/V ratios (see for
(3) curves exhibiting broad peak(s).
instance Bragato et al. (2007)) may lead to severe errors in case of
(1) In the case of clear H/V peaks, we show the reliability of the fuzzy H/V data like those observed in Santiago.
H/V method to provide a confident estimate of the fundamental res- It is worth concluding by emphasizing that the H/V method alone
onance frequency. The H/V method can be safely used, in addition is not enough to assess the seismic hazard in urbanized area espe-
to geophysical and geotechnical data, to obtain either the velocity cially when there is a large variety of buildings type: from small-size
or the sediment thickness at the site of interest. It is thus a valuable residential houses to high rise business buildings. A comprehen-
tool for seismic risk assessment as long as it is used in combination sive seismic-risk evaluation could not be based only on H/V re-
with other methods. sults. However, despite their limitations, the measurements of H/V
(2) Conversely, the H/V curves having flat or low H/V amplitude microtremor spectral ratios are very informative for site effects
are closely related to sites with lower impedance contrast, between estimating and remain a valuable input in urban seismic


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
936 S. Bonnefoy-Claudet et al.

microzonation, in the elaboration of a large earthquake scenario Barrientos, S., Vera, E., Alvarado, P. & Monfret, T., 2004. Crustal seismicity
and in seismic hazard mitigation. in central Chile, J. South Am. Earth Sci., 16, 759–768.
Finally, one should have special attention to the soil-sensor cou- Bonnefoy-Claudet, S., Cornou, C., Bard, P.-Y., Cotton, F., Moczo, P., Kristek,
pling, especially in urban areas where ‘natural’ ground surface is J. & Fäh, D., 2006a. H/V ratio: a tool for site effects evaluation. Results
from 1-D noise simulations, Geophys. J. Int., 167, 827–837.
not always easily available. Records made on non-cemented pave-
Bonnefoy-Claudet, S., Cotton, F. & Bard, P.-Y., 2006b. The nature of noise
ment might produce large amplitude of H/V curve for frequencies
wavefield and its applications for site effects studies. A literature review,
below 1 Hz. In such cases, H/V curves have to be carefully analysed Earth.-Sci. Rev., 79, 205–227.
before inferring site effect estimates. Bonnefoy-Claudet, S., Köhler, A., Cornou, C., Wathelet, M. & Bard, P.-Y.,
2008. Effects of Love waves on microtremor H/V ratio, Bull. seism. Soc.
Am., 98, 288–300.
AC K N OW L E D G M E N T S Bour, M., Fouissac, D., Dominique, P. & Martin, C., 1998. On the use
of microtremor recordings in seismic microzonation, Soil. Dyn. Earthq.
We are very grateful to Chilean students who performed the exten- Eng., 17, 465–474.
sive ambient vibrations measurements (José Lagos). We also thank Bragato, P., Laurenzano, G. & Barnaba, C., 2007. Automatic zonation of
Marc Wathelet for providing the software GEOPSY to compute urban areas based on the similarity of H/V spectral ratios, Bull. seism.
H/V ratio. We also address our gratitude to two anonymous review- Soc. Am., 97, 1404–1412.

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

ers whose comments substantially improve our original manuscript. Bravo P.R., 1992. Estudio geofisico de los suelos de fundacion para une
This research has been performed in the framework of joint collabo- zonificacion sismica del area urbana de Santiago Norte (In Spanish), PhD
ration between the Institute for Radioprotection and Nuclear Safety thesis. University of Chile, Santiago de Chile, Chile.
(Fontenay-aux-Roses, France) and the Department of Geophysics Chatelain, J.-L., Guillier, B., Cara, F., Duval, A.-M., Atakan, K. & Bard,
P.-Y., 2008. Evaluation of the influence of experimental conditions on
(University of Chile, Santiago, Chile). This work was supported
H/V results from ambient noise recordings, Bull. Earthq. Eng., 6, 33–74.
by the Milenio project and ANR research program (France) under
Cornou, C., 2002. Traitement d’antenne et imagerie sismique dans
contract SUBCHILE). l’agglomération grenobloise (Alpes françaises): implications pour les ef-
fets de site (In French), PhD thesis. University Joseph Fourier, Grenoble,
France.
REFERENCES
Cornou, C., Bard, P.-Y. & Dietrich, M., 2003. Contribution of dense array
Arai, H. & Tokimatsu, K., 2000. Effects of Rayleigh and Love waves on analysis to identification and quantification of basin-edge induced waves,
microtremor H/V spectra, in Proceedings of the 12th World Conference on part II: application to the Grenoble basin (French Alps), Bull. seism. Soc.
Earthquake Engineering, Auckland, New Zealand, January 30–February Am., 93, 2624–2648.
4, paper 2232. Cruz, E., Ridell, R. & Midorikawa, S., 1993. A study of site amplification
Araneda, M., Avendaño, M.S. & Merlo C., 2000. Modelo gravimetrico de effects on ground motions in Santiago, Chile, Tectonophysics, 218, 273–
la cuenca de Santiago, Etapa III final (In Spanish), in Proceedings of the 280.
IX Congreso Geologico Chileno, Puerto Varas, Chili, July 31–August 4, DeMets, C., Gordon, R., Argus, D. & Stein, S., 1994. Effect of recent
pp. 404–408, Sociedad Geologica de Chile. revisions to the geomagnetic time scale on estimates of current plate
Astroza, M. & Monge, J., 1987. Zonificacion sismica de la ciudad de San- motions, Geophys. Res. Lett., 21, 2191–2194.
tiago, in Proceedings of the XXIV Jornadas Sudamericanas de Ingeneria Delgado, J., Lopez Casado, C., Giner, J., Estevez, A., Cuenca, A. & Molina,
Estructural,Vol. 5, Porto Alegre, Brasil, June 29–July 3, pp. 221–235, S., 2000. Microtremor as a geophysical exploration tool: Application and
ANAIS, Porto Alegre, Brasil. limitations, Pure appl. Geophys., 157, 1445–1462.
Astroza, M. & Monge, J., 1991. Seismic microzones in the city of Santi- Di Giulio, G., Cornou, C., Ohrnberger, M., Wathelet, M. & Rovelli, A.,
ago. Relation damage-geological unit, in Proceedings of the Fourth In- 2006. Deriving wavefield characteristics and shear-velocity profiles from
ternational Conference on Seismic Zonation,Vol. 3, Stanford, CA, USA, two-dimensional small-aperture arrays analysis of ambient vibrations in
August 25 –29, pp. 595–601, Earthquake Engineering Research Institute, a small-size alluvial basin, Colfiorito, Italy, Bull. seism. Soc. Am., 96,
Stanford, CA, USA. 1915–1933.
Astroza, M., Monge, J. & Valera, J., 1993. Intensidades del sismo del 3 Duval, A.-M., Méneroud, J.-P., Vidal, S. & Singer, A., 1994. Relation be-
de marzo 1985 en la region Metropolitana y el litoral central. Ingeneria tween curves obtained from microtremor and site effects observed after
sismica, el caso del sismo del 3 de marzo 1985 (In Spanish), Instituto Caracas 1967 earthquake, in Proceedings of the 11th European Confer-
Ingenieros de Chile, Ediciones Pedagogicas Chilenas S.A., Ediciones ence on Earthquake Engineering, Paris, France, 6–11 September.
Dolmen, pp. 103–117. Falcon Moreno, E., Castillo Urrutia, O. & Valenzuela Muñoz, M., 1970.
Atakan, K., Duval, A.-M., Theodulidis, N., Guillier, B., Chatelain, J.-L., Hidrogeologia de la Cuenca de Santiago (In Spanish), Publication especial
Bard, P.-Y. & SESAME-Team, 2004. The H/V spectral ratio technique: No3, Instituto de Investigaciones Geologicas, Santiago, Chile.
experimental conditions, data processing and empirical reliability assess- Fäh, D., 1997. Microzonation of the city of Basel, J. Seism., 1, 87–
ment, in Proceedings of the 13th World Conference on Earthquake Engi- 102.
neering, Vancouver, Canada, 1–4 August, paper 2268. Fernandez Marescal, J.C., 2003. Respuesta sismica de la Cuenca de Santi-
Baize, S., Rebolledo, S., Lagos, J. & Rauld, R., 2006. A first-order model of ago, region metropolitana de Santiago, Escala 1:100000, 1 mapa. SER-
the Santiago basin, in Proceedings of the International Conference 1906 NAGEOMIN, carta geologica de Chile, Serie Geologia Ambiental, 1.
Valparaiso Earthquake Centennial, Santiago de Chile, Chile, November ISSN:0717–7305.
6–8. Gardi, A., Lemoine, A., Madariaga, R. & Campos, J., 2006. Modeling of
Bard, P.-Y., 1998. Microtremor measurements: a tool for site effect esti- stress transfer in the Coquimbo region of Central Chile, J. geophys. Res.,
mation? in Proceedings of the Second International Symposium on the 111, B04307.
Effects of Surface Geology on Seismic Motion,Vol. 3, Yokohama, Japan, Guéguen, P., 1994. Microzonage de santiago du Chile (technique de Naka-
December 1–3, pp. 1251–1279, Balkema: Rotterdam, Yokohama, Japan. mura) (In French), Master Thesis. Joseph Fourier University, Grenoble,
Bard, P.-Y. & SESAME-Team, 2005. Guidelines for the implementation of France.
the H/V spectral ratio technique on ambient vibrations–measurements, Gueguen, P., Cornou, C., Garambois, S. & Banton, J., 2007. On the limitation
processing and interpretations, SESAME European research project of the H/V spectral ratio using seismic noise as an exploration tool:
EVG1-CT-2000–00026, deliverable D23.12, available at http://sesame- application to the Grenoble valley (France), a small apex ratio basin, Pure
fp5.obs.ujf-grenoble.fr. appl. Geophys., 164, 115–134.


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS
 Site effects evaluation in Santiago de Chile 937

Guillier, B., Cornou, C., Kristek, J., Moczo, P., Bonnefoy-Claudet, S., Bard, Nogoshi, M. & Igarashi, T., 1971. On the amplitude characteristics of mi-
P.-Y. & Fäh, D., 2006. Simulation of seismic ambient vibrations: does the crotremor, Part 2 (In Japanese with English abstract), J. Seism. Soc. Japan,
H/V provide quantitative information in 2D-3D structures? in Proceedings 24, 26–40.
of the Third International Symposium on the Effects of Surface Geology Olsen, K.B. & Archuleta, R.J., 1996. Three dimensional simulation of earth-
on Seismic Motion, Grenoble, France, August 29–September 1, Paper quakes on the Los Angeles fault system, Bull. seism. Soc. Am., 86, 575–
185. 596.
Ibs-Von Seht, M. & Wohlenberg, J., 1999. Microtremor measurements used Pardo, M., Comte, D. & Monfret, T., 2002. Seismotectonic and stress distri-
to map thickness of soft sediments, Bull. seism. Soc. Am., 89, 250– bution in the central Chile subduction zone, J. South Am. Earth Sci., 15,
259. 11–22.
Iriarte-Diaz, S., 2003. Impact of urban recharge on long-term management Parolai, S., Bormann, P. & Milkreit, C., 2002. New relationships between
of Santiago Norte aquifer, Santiago–Chile, Master thesis. Waterloo Uni- Vs, thickness of sediments, and resonance frequency calculated by the
versity, Notario, Canada. H/V ratio seismic noise for the Cologne area (Germany), Bull. seism. Soc.
Kawase, H., 1996. The cause of the damage belt in Kobe:”the basin-edge Am., 92, 2521–2527.
effect”, constructive interference of the direct S-wave with the basin- Rodriguez, H.S. & Midorikawa, S., 2002. Applicability of the H/V spectral
induced diffracted Rayleigh waves, Seism. Res. Lett., 67, 25–34. ratio of microtremors in assessing site effects on seismic motion, Earthq.
Konno, K. & Ohmachi, T., 1998. Ground-motion characteristics estimated Eng. Struct. Dyn., 31, 261–279.
from spectral ratio between horizontal and vertical components of mi- Rovelli, A., Scognamiglio, L., Marra, F. & Caserta, A., 2001. Edge-diffracted

Downloaded from https://academic.oup.com/gji/article/176/3/925/560230 by guest on 27 July 2022

crotremor, Bull. seism. Soc. Am., 88, 228–241. 1-sec surface waves observed in a small-size intramountain basin (Colfior-
Lachet, C. & Bard, P.-Y., 1994. Numerical and theoretical investigations on ito, Central Italy), Bull. seism. Soc. Am., 91, 1851–1866.
the possibilities and limitations of Nakamura’s technique, J. Phys. Earth, Theodoulidis, N., Cultrera, G., De Rubeis, V., Cara, F., Panou, A., Pagani,
42, 377–397. M. & Teves-Costa, P., 2008. Correlation between damage and ambient
Lay, T. et al., 2005. The great Sumatra-Andamman earthquake of 26 decem- noise H/V spectral ratio: the SESAMEproject results, Bull. Earthq. Eng.,
ber 2004, Science, 308, 1127–1133. 6, 109–140.
Lebrun, B., Hatzfeld, D. & Bard, P.-Y., 2001. A site effect study in urban Toshinawa, T., Matsuoka, M. & Yamazaki, Y., 1996. Ground-motion char-
area: experimental results in Grenoble (France), Pure appl. Geophys., acteristics in Santiago, Chile, obtained by microtremor observations, in
158, 2543–2557. Proceedings of the 11th World Conference on Earthquake Engineering,
Lermo, J. & Chavez-Garcia, F., 1993. Site effect evaluation using spectral Acapulco, Mexico, 23–28 June, Paper 1764.
ratios with only one station, Bull. seism. Soc. Am., 83, 1574–1594. Uebayashi, H., 2003. Extrapolation of irregular subsurface structures using
Lermo, J. & Chavez-Garcia, F., 1994. Are microtremors useful in site re- the horizontal-to-vertical ratio of long-period microtremors, Bull. seism.
sponse evaluation?, Bull. seism. Soc. Am., 84, 1350–1364. Soc. Am., 93, 570–582.
Midorikawa, S., Riddell, R. & Cruz, E., 1991. Strong-ground accelerograph Uetake, T. & Kudo, K., 1998. The excitation of later arrivals in Ashigara
array in Santiago, Chile, and preliminary evaluation of site effects, Earthq. valley during earthquakes occurring in east part of Yamanashi prefecture,
Eng. Struct. Dyn., 20, 403–407. in Proceedings of the Second International Symposium on the Effects of
Monge, J., 1986. El sismo del 3 de marzo 1985, Chile (In Spanish), 2nd edn, Surface Geology on Seismic Motion, pp. 427–434, Balkema: Rotterdam,
Acero Comercial, Santiago de Chile, Chile. Yokohama, Japan.
Morales-Jerez, F.R., 2002. Definicion de aquiferos en la Cuenca del Rio Valenzuela, G.B., 1978. Suelo de fundacion del gran Santiago (In Spanish
Maipo (In Spanish), Master thesis. University of Chile, Santiago de Chile, with English abstract), Instituto de Investigaciones Geologicas, 33, 1–27.
Chile. Wyllie, L. et al., 1986. The Chile earthquake of March 3, 1985, Earthq.
Mucciarelli, M., Gallipoli, M., Di Giacomo, D., Di Nota, F. & Nino, E., Spectra, 2, 293–371.
2005. The influence of wind on measurements of seismic noise, Geophys. Woolery, E.W. & Street, R., 2002. 3D near-surface soil response from H/V
J. Int., 161, 303–308. ambient-noise ratios, Soil. Dyn. Earthq. Eng., 22, 865–876.
Nakamura, Y., 1989. A method for dynamic characteristics estimation of Yalcinkaya, E. & Alptekin, O., 2005. Contributions of basin-edge-induced
subsurface using microtremor on the ground surface, Q.R. Rail. Tech. surface waves to site effect in the Dinar bain, Southwestern Turkey, Pure
Res. Inst., 30, 25–30. appl. Geophys., 162, 931–950.


C 2008 IRSN, GJI, 176, 925–937
Journal compilation 
C 2008 RAS