Geophysical Journal International

Transcription

Geophysical Journal International
Geophys. J. Int. (2014) 198, 1644–1652
GJI Seismology
doi: 10.1093/gji/ggu234
Application of teleseismic long-period surface waves from ambient
noise in regional surface wave tomography: a case study in western
USA
Yingjie Yang
ARC Centre of Excellence for Core to Crust Fluid Systems (CCFS) and GEMOC ARC National Key Centre, Earth and Planetary Sciences, Macquarie
University, Sydney, Australia. E-mail: yingjie.yang@mq.edu.au
Accepted 2014 June 17. Received 2014 May 20; in original form 2014 February 20
Key words: Interferometry; Surface waves and free oscillations; Seismic tomography.
1 I N T RO D U C T I O N
Traditional surface wave tomography uses earthquake surface waves
to image the structures of crust and upper mantle. However, because
most large earthquakes occur at plate boundaries, earthquake-based
surface tomography suffers from uneven distributions of earthquakes. Especially in regional teleseismic surface wave tomography, uneven distributions of earthquakes result in unfilled azimuthal
gaps in the ray coverage, which could result in smeared velocity
anomalies in tomography and make it difficult to recover azimuthal
anisotropy. In addition, due to strong attenuation and scattering
of short-period teleseismic surface waves caused by the heterogeneous crust, it is notoriously difficult to obtain a high-resolution
crustal model from earthquake surface waves. In the last decade,
the advent of ambient noise tomography (ANT) has revolutionized
1644
C
seismic tomography because it can overcome the above limitations
of earthquake surface wave tomography (Sabra et al. 2005; Shapiro
et al. 2005). This technique uses diffuse background ground motion,
which is conventionally called ‘ambient noise’ and mostly comes
from the interaction of ocean waves with the crust (e.g. Traer et al.
2012; Traer & Gerstoft 2014), to extract surface wave empirical
Green’s functions between a pair of stations by cross-correlating
continuous time series of ambient noise. Within a regional seismic array, all interstation surface wave dispersion measurements
can be measured and tomography can be performed to image the
underling lithospheric structures. ANT apparently has advantages
over earthquake-based tomography because surface wave dispersion
curves between station pairs can be measured without requiring the
occurrences of earthquakes and each station can be imagined as a
‘virtual’ earthquake in addition to being a receiver.
The Author 2014. Published by Oxford University Press on behalf of The Royal Astronomical Society.
Downloaded from http://gji.oxfordjournals.org/ at Macquarie University on October 27, 2014
SUMMARY
Since the emerging of ambient noise tomography (ANT) in 2005, it has become a routine
method to image the structures of crust and uppermost mantle because of its exclusive capability to extract short-period surface waves. Most of previous ANT studies focus on surface
waves at periods shorter than 40/50 s. There are only a few studies of long-period surface wave
tomography from ambient noise (longer than 50 s) in global scale. No tomography studies have
been performed using teleseismic long-period surface waves from ambient noise in a regional
scale, probably due to the two reasons that (1) energy of long-period ambient noise is weaker
and it is harder to retrieve good signal-to-noise ratio long-period surface waves from portable
stations with several years of ambient noise data and (2) long-period dispersion measurements
from ambient noise may have larger uncertainties than those at shorter periods (<40/50 s).
In this study, I investigate the feasibility of using teleseismic long-period surface waves from
ambient noise in regional surface wave tomography and also evaluate the accuracy of longperiod dispersion measurements at periods up to 150 s. About 300 USArray/Transportable
Array (TA) stations located in the Colorado Plateau and surrounding areas and 400 teleseismic stations relative to the TA stations are selected. Clear, strong, and coherent long-period
teleseismic surface waves at periods much longer than 50 s are observed in the teleseismic
cross-correlations between the TA stations and the teleseismic stations. Using long-period
dispersion curves from ambient noise, I generate phase velocity maps at 50–150 s periods and
then compare them with phase velocity maps from teleseismic earthquake data. The results
show that phase velocity maps from ambient noise data and earthquake data are similar at the
50–150 s period range, verifying the validity of using long-period surface wave from ambient
noise in regional surface wave tomography.
Use of the teleseismic surface waves from ANT
in the Colorado Plateau and surrounding areas and 400 permanent
stations distributed globally are selected. By performing teleseismic
cross-correlation of ambient noise between the TA stations and
the ‘global’ stations, I observe clear, strong, and coherent longperiod Rayleigh waves at periods longer than 50 s between stations
with thousands of kilometres of interstation distances. To further
investigate the accuracy of long-period surface wave dispersion
measurements from ambient noise and demonstrate the feasibility of
using them in regional surface wave tomography, I perform regional
teleseismic ANT and compare resulting tomographic dispersion
maps based on ambient noise with those based on earthquake data.
2 D ATA A N D C R O S S - C O R R E L AT I O N S
In this study, the region of the Colorado Plateau and surrounding
areas is selected as a case study. The Colorado Plateau is a large and
stable tectonic unit with an elevation of ∼1800–2000 m. Contrast
to the strongly deformed regions surrounding the Colorado plateau,
the plateau’s internal lithosphere is almost undeformed. In order
to understand the stability and the origin of the high elevation, a
number of seismic tomography studies have been carried out to
image the lithosphere structure. Strong lateral velocity variations
have been found with higher seismic velocities in the interior of the
plateau surrounded by lower seismic velocities in the peripheries
except in the northern boundary (e.g. Schmandt & Humphreys 2010;
Levander et al. 2011; Liu et al. 2011). Such a region serves as
a good testing site for investigating whether the ambient noisederived long-period surface waves are able to image the strong
heterogeneities of lithosphere.
About 300 TA stations located in the Colorado Plateau and surrounding areas (Fig. 1a) and about 400 permanent stations from
Global Seismographic Network (GSN) and International Federation of Digital Seismograph Networks (FDSN) distributed around
the globe (Fig. 1b) are selected. Teleseismic surface waves are extracted from teleseismic cross-correlation of ambient noise between
Figure 1. Left-hand panel: the locations of selected seismic stations (red triangles) from the USArrray/Transportable Array (TA) deployed in the Colorado
Plateau and surrounding regions. Right-hand panel: Azimuthal equidistant projection of the selected 400 teleseismic stations (blue circles). The plot is centred
at the geographic point of 35◦ N, 108◦ W, as marked by the red star.
To date, ANT has become a routine method to map the crustal
and uppermost mantle structures (e.g. Moschetti et al. 2007; Bensen
et al. 2009; Ekström et al. 2009, in USA; Yao et al. 2006, 2010;
Zheng et al. 2011; Zhou et al. 2012, in Asia). Both Rayleigh and
Love surface wave dispersion maps are commonly obtained with
spatial extents ranging from regional to continental scales. Most of
the existing ANT studies have focused on surface waves at periods
shorter than 40–50 s because ambient noise at this period range
is strong and accurate interstation dispersion curves can be easily measured. Two natural questions one may ask are (1) whether
longer-period surface wave (>50 s) can also be easily extracted
from ambient noise and (2) whether long-period dispersion curves
from ambient noise are accurate enough for tomography to constrain
upper-mantle structures.
Shapiro & Campillo (2004) have first shown that surface wave
dispersion curves up to 125 s period can be retrieved from crosscorrelations of ambient noise with interstation distances longer than
2000 km. However, there are only a few tomographic studies using
teleseismic long-period surface waves (>50 s) in global scale (e.g.
Nishida et al. 2009; Shen & Zhang 2012) and continental scales
(e.g. Bensen et al. 2009). No ANT studies have been performed using teleseismic long-period surface waves from ambient noise, like
100 s period, at a regional scale with an aperture of several hundred
kilometres. In global ANT studies, more than 10 yr of ambient noise
data (e.g. Nishida et al. 2009) are cross-correlated to generate sufficiently high signal-to-noise ratio (SNR) long-period surface waves.
It is not a problem if seismic stations used are permanent stations.
However, for regional tomography, most portable arrays are usually
deployed only for several years, which calls into question whether
teleseismic long-period surface waves can be extracted from mere
several years of ambient noise data.
In this study, I investigate the possibility of extracting teleseismic
long-period surface waves from ambient noise recorded at portable
seismic arrays by taking the USArray/Transportable Array (TA)
in western USA as a case study. About 300 TA stations located
1645
1646
Y. Yang
these ‘base’ stations and the 400 ‘remote’ global stations. These
300 TA stations work as ‘base’ stations covering a study region
where regional surface wave tomography is performed. Vertical
components of continuous ambient noise seismic data recorded
during 2007–2010 are collected and processed for both the ‘base’
and ‘remote’ stations.
The procedures of data processing of teleseismic crosscorrelation are similar to the conventional procedures described
in Bensen et al. (2007) except the stacking part. First, vertical component seismograms are filtered at a broad period band of 10–300 s
after the mean and trend are removed. Because different types of
seismic sensors are used among the TA stations and the ‘remote’
stations, instrument responses are removed from seismograms. The
filtered and instrument response removed seismograms are then
whitened in frequency domain and normalized using running average in time domain to normalize amplitude of ambient noise and
meanwhile suppress the amplitudes of earthquake signals. The procedures of whitening and normalization used in this study are different from the frequency–time normalization method developed by
Shen & Zhang (2012) to normalize seismograms in order to extract
long-period surface waves from ambient noise.
Cross-correlations of daily segments are then performed for
each possible station pair of a ‘base’ station and a ‘remote’ station. The daily cross-correlations are stacked to form monthly
cross-correlations. Different from the conventional method linearly
stacking the monthly cross-correlations to form the final crosscorrelations, in this study, a stacking method based on S-transform
(Stockwell et al. 1996; Schimmel et al. 2011) is adopted to stack
monthly cross-correlations to generate the final cross-correlations.
The stacking method has been proved to be effective to improve
SNR of cross-correlations (Schimmel et al. 2011; Ren et al. 2013)
compared to a linearly stacking method.
The finally stacked cross-correlations are folded to form the socalled symmetric components by further stacking the negative and
positive time lags of each cross-correlation. At last, a method of
frequency-time analysis (FTAN; Dziewonski et al. 1969; Levshin
et al. 1972; Levshin & Ritzwoller 2001; Bensen et al. 2007) is applied to the symmetric components of cross-correlations to obtain
phase velocity measurements at 50–250 s periods. The period dependent SNR of each cross-correlation is calculated by taking the
ratio between the maximum amplitude within a surface wave window defined by a group velocity window of 3–5 km s−1 and the rms
of trailing time-series following the surface wave time window. In
the subsequent surface wave tomography, only those phase velocity
measurements with SNR larger than 8 are retained for tomography.
Examples of teleseismic cross-correlations between a ‘remote’ station IC.MDJ located in northeast China, and the ‘base’ TA stations
located in the Colorado Plateau are plotted in Fig. 2. The distances
between the station IC.MDJ and the TA stations range from 7500 to
8500 km. Clear and strong long-period Rayleigh waves emerge in
Figure 2. Examples of teleseismic cross-correlations between a teleseismic station IC.MDJ located in northeast China and the selected TA stations. (a) The
great-circle paths between station IC.MDJ and the TA stations are plotted as grey lines. The cross-correlations between IC.MDJ and the TA stations are filtered
at three period bands: 50–100 s (b), 100–200 s (c), 200–300 s (d).
1647
the cross-correlations with a nearly linear move-out consistent with
the propagation of surface waves. Another example is for teleseismic cross-correlations between a ‘remote’ station II.SUR located
in South Africa and the TA stations plotted in Fig. 3. The distances
between them vary from ∼14 500 to ∼16 000 km, close to the half
circumference of the earth. Clear and coherent long-period surface
waves from the teleseismic cross-correlations still appear. One thing
worth mentioning and emphasizing here is that the time duration of
the continuous ambient noise data used in these teleseismic crosscorrelations is about 2 yr, the typical deployment duration of TA
stations.
Stacking over increasingly longer time series generally increase
the SNR of resulting long-period surface waves. One example of
increasing of SNR as a function of months of stacking between
station IC.MDJ and station TA.R20A is plotted in Fig. 4 along with
the cross-correlations stacked at different lengths of time-series.
One advantage of surface waves from ambient noise over earthquake surface waves is that the source term is approximately known
(Snieder 2004; Lin et al. 2008) given the effective distribution of
ambient noise is nearly homogenous when cross-correlation is performed over a long period of time, such as 2 yr, and time and
frequency domain normalization are adopted to normalize the amplitudes of noise sources (Bensen et al. 2007). Dispersion curves
from cross-correlations can be measured without the need to invert
for earthquake source term. One example of surface wave dispersion curve between the ‘remote’ station IC.MDJ and a ‘base’ station
TA.R20A (located in western Colorado) filtered at 50–250 s are
plotted in Fig. 5(b) with the corresponding cross-correlation waveforms filtered at various period ranges plotted in Fig. 5(a). Phase
velocity is calculated by measuring instantaneous phase at each
period and unwrapping the phase using reference phase velocities
between a station pair calculated from a Global Dispersion Model
GDM52 (Ekström 2011). The measured dispersion curve (black
line in Fig. 5b) is very close to the reference phase velocity curve
(red dashed line in Fig. 5b). To visualize the quality of long-period
surface waves, the 2-D FTAN diagram of normalized signal power
as a function of time and frequency are also plotted in the background in Fig. 5(b) along with the group velocity curve (blue line)
measured from the maximum amplitudes along the frequency axis.
Clear and focused signals of surface waves are apparently exhibited
in the 2-D FTAN diagram.
Figure 4. Example of increasing SNR of Rayleigh waves as a function of
months of stacking. (a) Cross-correlations at the specific months of stacking
between station IC.MDJ and station TA.R20A. All the cross-correlations are
bandpass filtered at a 50–250 s period band. (b) SNR of emerging Rayleigh
waves from stacking of 1, 3, 6, 12 and 24 months at 50–250 s periods.
Systematic comparison between the measured dispersion curves
from teleseismic cross-correlations and predicted dispersion curves
from the Global Dispersion Model GDM52 provides a means to
evaluate the accuracy of the dispersion curves from ambient noise
Figure 3. Same as Fig. 2 but for a teleseismic station II.SUR located in southern Africa. The cross-correlations between II.SUR and the TA stations are filtered
at one period band of 100–300 s.
1648
Y. Yang
Figure 6. (a) Azimuthal distribution of the teleseismic stations from which
high SNR long-period surface waves are obtained from teleseismic crosscorrelations with the TA stations. (b) Azimuthal distribution of selected
teleseismic events occurring 2007–2010 with Ms >5.5 and epicentral distances relative to the selected TA stations longer than 3000 km.
and will be performed in a future study. In this study, I mainly intend
to focus on the verification of the accuracy of long-period dispersion
measurements from ambient noise by comparing the tomographic
dispersion maps from ambient noise with those from earthquakes
and quantifying their differences.
Out of the total ∼400 selected ‘remote’ stations, about 50 per cent
yield high SNR long-period surface waves from the teleseismic
cross-correlations with the ‘base’ stations (Fig. 6a). In the regional
surface wave tomography discussed in the following section, these
‘remote’ stations are treated as ‘virtual’ teleseismic events.
3 R E G I O N A L T E L E S E I S M I C S U R FA C E
WAV E T O M O G R A P H Y
Traditionally, long-period surface waves (>50 s) are usually obtained from teleseismic events to image upper-mantle structures.
However, due to scattering or multipathing caused by lateral heterogeneities along long propagating distances between earthquakes
and a seismic array, teleseismic incoming waves into a regional
seismic array could be distorted, causing incoming azimuths deviating away from the great-circle azimuth and leading to complex
wavefields. To deal with this problem, Forsyth & Li (2005) has developed a method called two-plane-wave tomography (TPWT) by
modelling an incoming teleseismic wavefield using the sum of two
plane waves, each with initially unknown amplitude, initial phase
and propagation direction. The sensitivities of surface waves to
phase velocity heterogeneities for each plane wave are represented
by 2-D sensitivity kernels (Yang & Forsyth 2006a) based on Born
approximation (Zhou et al. 2004).
Because surface waves used in this method are from teleseismic
events with epicentral distances typically more than 3000 km, much
larger than the aperture of a study region (typical having a scale of
several hundred kilometres), the interference of two plane waves is
used to model an incoming wavefield. Data used in the tomography
are relative phases and amplitudes of surface waves among a seismic
array. Thus, the exact source mechanisms and locations of earthquakes have insignificant effects on tomographic results and can be
neglected. This is an advantage over some surface wave tomography methods requiring accurate information of earthquake sources.
However, the TPWT method requires both amplitude and phase
data of surface waves in order to resolve incoming wavefields and
phase velocity variations across a study region (e.g. Li et al. 2003;
Yang & Forsyth 2006b; Yang & Ritzwoller 2008). For teleseismic
surface waves extracted from ambient noise, accurate amplitudes
cannot be measured from cross-correlations because amplitudes in
cross-correlations are affected by azimuthal variations of strength
of noise sources, the duration of cross-correlations, the normalization method applied in data processing (Lin et al. 2011), and so on.
Thus, not both amplitude and phase measurements are available to
model an incoming wavefield with two plane waves.
With the sole measurements of phase, as an alternative means,
the incoming phase front of a teleseismic event rather than the incoming wavefield is modelled using a single plane wave with the
unknown propagation direction and initial phase. Based on experiences in TPWT, one major wave of the modelled two plane waves
for most teleseismic events dominates the wave amplitude with the
secondary one merely having a small amplitude of about 10–20
per cent of the primary one (Forsyth & Li 2005; Yang & Forsyth
2006b). Modelling phase fronts using one simple plane wave rather
than modelling the wavefield using two plane waves is sufficient
enough to represent phase information of a teleseismic event. To
verify this conclusion, the differences of phase velocities generated
by TPWT and one-plane-wave tomography are compared. For this
comparison, about 700 teleseismic events are selected from a total
number of ∼1000 earthquakes occurring during 2007–2010 with
Ms >5.5 and epicentral distances >3000 km. The distribution of
Figure 5. Left-hand panel: example of teleseismic cross-correlation between station IC.MDJ (in northeast China) and station TA.R20A (in western Colorado)
filtered at 50–100, 100–200 and 200–300 s period bands. Right-hand panel: resulting phase velocity dispersion curves (black line) measured from the crosscorrelation compared with the reference phase velocity dispersion curves calculated from Global Dispersion Model GDM52 (red dashed line) (Ekström 2011).
The 2-D FTAN diagram of normalized signal power as a function of time and frequency are also plotted in the background along with the group velocity curve
(blue line).
Figure 7. Comparison of phase velocity maps at 60 s period generated
by using two-plane-wave tomography (left-hand panel) and one-plane-wave
tomography (middle panel). The map of differences is plotted as percentage
relative to the regional average phase velocity of 3.86 km s−1 in the righthand panel.
4 R E S U LT S O F S U R FA C E WAV E
TOMOGRAPHY
As shown in Figs 2 and 3, long-period surface waves (>50 s) are
extracted from teleseismic cross-correlations of continuous ambient noise. For a dense regional array, a ‘remote’ station with several
thousand kilometres away is treated as ‘virtual’ teleseismic earthquake. Similar to a teleseismic event, the incoming phase front from
the ‘remote’ station is modelled as one-plane wave with unknown
propagating direction and initial phase. One-plane-wave tomography is applied to long-period surface wave phase data obtained from
cross-correlations between the ‘base’ TA stations and the ‘remote’
stations with interstation distances longer than 3000 km.
The study area of this work is large with an aperture over 1000 km,
which exceeds the limitation of one-plane-wave assumption in a
Cartesian and spherical coordinates. In addition, surface wave phase
fronts propagating across such a large area could have lateral variations which cannot be modelled by one single plane wave. Thus,
the study region is divided into four subregions, within each of
which the incoming wave front is modelled as one-plane wave. Total four plane waves are employed to represent the lateral variations
of phase fronts from each ‘remote’ station. Dividing the study region into more subregions may more accurately model the incoming
phase front. However, based on previous studies (Yang & Ritzwoller
2008; Yang et al. 2011), for teleseismic surface waves with epicentral distances longer than 3000 km, incoming wavefields propagating across a region with an aperture around 500–700 km can be
modelled by two plane waves well. Further division of subregions
into smaller regions dose not result in significant and systematic
changes of final phase velocity maps in tomography.
In regional surface wave tomography, we are interested in velocity
anomalies in order of 100 or several hundreds of kilometres. The
wavelength of a surface wave increases with period; for example, at
200 s period, the wavelength of a Rayleigh wave is ∼900 km, close to
the aperture of the whole area of this study. Therefore, it is difficult to
image lateral variations at a scale of a few hundreds of kilometres at
periods as long as 200 s. Thus, in this study, regional surface wave
tomography is only performed at 50–150 s periods even though
teleseismic surface waves from ambient noise at periods longer
than 150 s are still observed.
Resulting tomographic phase velocity maps at 50, 70, 100 and
140 s periods are plotted in Fig. 8 (left-hand column). At these periods, phase velocities are most sensitive to the shear velocities of the
upper mantle. Low velocities are observed near the western, eastern
and southern fringes of the Colorado Plateau. High velocities are
observed in the central and north Colorado Plateau. These high velocities appear connected with the high velocities in the Wyoming
craton. High velocities are also observed in central United States to
the east of the Colorado Plateau. Stronger velocity contrasts are seen
at intermediate periods of 50–70 s than at longer periods (>100 s)
across the east boundary of the plateau. The phase velocity maps
are similar to other surface wave tomography studies (e.g. Liu et al.
2011; Shen et al. 2013; Foster et al. 2014).
Here, instead of focusing on the interpretation of the phase velocity maps, I intend to evaluate the similarity of the phase velocity maps generated using ambient noise data and earthquake data.
Thus, earthquake-based phase velocity maps are also generated using teleseismic surface waves from earthquakes shown in Fig. 6(b).
The exactly same one-plane-wave tomography applied to teleseismic surface waves from ambient noise is applied to these earthquake
surface waves.
Both sets of phase velocity maps from ambient noise data and
earthquake data respectively are plotted together side by side for
comparison in Fig. 8. Features of phase velocity anomalies are similar between the earthquake-derived and noise-derived dispersion
maps, especially in the interiors of the study region where path coverage is dense for both data sets. Slightly larger differences at the
peripheries of the study regions are observed where uncertainties of
phase velocity maps also become larger because the path coverage
is not as good as that in the interior area. To quantify the differences,
the histogram of the phase velocity differences are presented in the
right-hand column of Fig. 8 and the mean and standard deviations
are calculated. Differences of phase velocities nearly follow a Gaussian distribution. The mean of differences is less than 10 m s−1 and
the standard deviation is around 30 m s−1 , less than 1 per cent of
phase velocities at these periods. There are no apparent variations of
phase velocity differences with period at the 50–150 s period range.
The standard deviations of phase velocity differences at 50–150 s
periods in this study are slightly larger than the standard deviations
of ∼20 m s−1 at intermediate periods of 25–40 s as shown by Shen
et al. (2013) using eikonal tomography for both ambient noise data
and earthquake data. This may be mainly due to the fact that a larger
number of data are used in the tomography of Shen et al. (2013)
5 DISCUSSION
By examining the distribution of ‘good’ ‘remote’ stations
which generate high SNR teleseismic surface waves from
cross-correlations with the ‘base’ TA stations (Figs 1b and 6a),
selected teleseismic events is exhibited in Fig. 6(b). Instantaneous
phases, amplitudes and SNRs of Rayleigh waves at various periods are calculated after instrument responses, means and trends of
seismograms are removed. Both one-plane-wave and TPWT are applied to the selected earthquake surface waves. One example of the
resulting phase velocity maps at 60 s period using one-plane-wave
modelling is plotted in Fig. 7 compared with the phase velocity map
using two-plane-wave modelling as well as the map of differences.
Phase velocity maps at 60 s period based on these two approaches
are similar with an offset of the mean at 0.5 m s−1 and a standard
deviation of differences at 16 m s–1 , only ∼0.4 per cent relative to
the regional average phase velocity of 3.86 km s−1 . The differences
at other periods are similar to those at 60 s period.
1649
1650
Y. Yang
Figure 8. Resulting tomographic phase velocity maps at 50, 70, 100 and 140 s periods based on ambient noise data (left-hand column) and teleseismic
earthquake data (middle column). The ambient noise- and earthquake-based phase velocity maps are plotted as perturbations relative to the same reference
phase velocities of individual periods, 3.81, 3.87, 3.97 and 4.18 km s−1 at 50, 70, 100 and 140 s periods, respectively. The histograms of differences of phase
velocity maps are plotted in the right-hand column with their means and standard deviations shown at the top left of each panel.
6 C O N C LU S I O N S
I have demonstrated that high SNR long-period surface waves can
be extracted from cross-correlations of two years of ambient noise
data from station pairs with teleseismic interstation distances. The
teleseismic long-period surface waves can be used in regional surface wave tomography by treating ‘remote’ stations as ‘virtual’
teleseismic earthquakes. Long-period phase velocity maps based
on ambient noise data are similar to those based on teleseismic
earthquakes, indicating the long-period dispersion curves from ambient noise are reliable. Long-period surface waves from ambient
noise are complementary to those from earthquakes and can be
included in regional and global surface wave tomography, significantly increasing both lateral and azimuthal path coverage, which
is essential to improving the imaging of high resolution heterogeneities and azimuthal anisotropy, especially at regions with large
gaps of azimuthal distributions of earthquakes.
AC K N OW L E D G E M E N T S
The author thanks two anonymous reviewers for constructive comments that have improved this paper. The facilities of the IRIS
Data Management System, and specifically the IRIS Data Management Center, are used for access to waveforms products required
in this study. This research is supported by an Australian Research
Council Discovery grant (DP120102372 and DP120103673) and
Future Fellowship (FT130101220). This is contribution 476 from
the ARC Centre of Excellence for Core to Crust Fluid Systems
(http://www.ccfs.mq.edu.au) and 946 in the GEMOC Key Centre
(http://www.gemoc.mq.edu.au).
REFERENCES
Bensen, G.D., Ritzwoller, M.H., Barmin, M.P., Levshin, A.L., Lin, F.,
Moschetti, M.P., Shapiro, N.M. & Yang, Y., 2007. Processing seismic
ambient noise data to obtain reliable broad-band surface wave dispersion
measurements, Geophys. J. Int., 169, 1239–1260.
Bensen, G.D., Ritzwoller, M.H. & Yang, Y., 2009. A 3D shear velocity
model of the crust and uppermost mantle beneath the United States from
ambient seismic noise, Geophys. J. Int. 177, 1177–1196.
Dziewonski, A., Bloch, S. & Landisman, M., 1969. A technique for the
analysis of transient seismic signals, Bull. seism. Soc. Am., 59, 427–444.
Ekström, G., 2011. A global model of Love and Rayleigh surface wave
dispersion and anisotropy, 25–250 s, Geophys. J. Int., 187(3), 1668–1686.
Ekström, G., Abers, G. & Webb, S., 2009. Determination of surface-wave
phase velocities across USArray from noise and Aki’s spectral formulation, Geophys. Res. Lett., 36, L18301, doi:10.1029/2009GL039131.
Forsyth, D.W. & Li, A., 2005. Array-analysis of two-dimensional variations
in surface wave phase velocity and azimuthal anisotropy in the presence
of multi-pathing interferece, in Seismic Earth: Array Analysis of Broadband Seismograms, pp. 81–98, eds Levander, A. & Nolet, G., Geophys.
Monogr., 157, AGU.
Foster, A., Ekstrom, G. & Nettles, M., 2014. Surface-wave phase velocities
of the western United States from a two-station ämethod, Geophys. J. Int.,
196, 1189–1206.
Levander, A., Schmandt, A.B., Miller, M.S., Liu, K., Karlstrom, K.E.,
Crow, R.S., Lee, C.-T.A. & Humphreys, E.D., 2011. Continuing Colorado
plateau uplift by delamination-styleconvective lithospheric downwelling,
Nature, 472, 461–465.
Levshin, A.L. & Ritzwoller, M.H., 2001. Automated detection, extraction,
and measurement of regional surface waves, Pure appl. Geophys., 158(8),
1531–1545.
Levshin, A.L., Pisarenko, V.F. & Pogrebinsky, G.A., 1972. On a frequency–
time analysis of oscillations, Ann. Geophys., 28, 211–218.
it is noted that there is lower percentage of ‘good’ stations located
in oceanic islands than those located in continents. The main reason for this is probably that local noise levels at island stations are
stronger than continental stations, resulting from local interaction
of oceanic waves with the island as also demonstrated by Lin et al.
(2006). The local noise obscures the propagation of coherent noise
signals from one station to another. It is also noted that a large
number of ‘remote’ stations with their corresponding full paths
passing across the Pacific Ocean, such as those stations in Australia and Antarctic, do not generate high SNR surface waves. As
is well known in cross-correlations, only those noise sources propagating from either end of a station pair constructively contribute
to the generation of interstation surface waves; while noise sources
in the middle part between a station pair obscure the propagation
of the coherent signals from the two ends (Snieder 2004; Snieder
et al. 2007). As various studies show that long-period ambient noise
are especially strong in the Pacific Ocean (e.g. Rhie & Romanowicz
2004, 2006; Traer et al. 2012), the energetic Pacific-originated noise
sources in the middle of a station pair could destructively contribute
to the coherent interstation surface waves and result in the low SNR
surface waves. Another observed feature is that most of long-period
cross-correlations are more symmetric, that is both positive and negative components have strong and comparable surface wave signals
(Figs 2 and 3), compared with cross-correlations at shorter periods
(10–40 s). The higher symmetry in cross-correlations at longer periods mostly likely results from the fact that long-period noise energy
is much less attenuated by the Earth and can propagate across an
entire continent from one end to the other, over much longer distances than shorter period noise, which contributes to more diffuse
distribution of coherent long-period ambient noise energy between
a station pair.
The high similarity between ambient noise-based and earthquakebased long-period phase velocity maps (>50 s) verifies the validity
of using long-period surface waves from ambient noise in imaging
lithospheric and sublithospheric upper-mantle structures. By combining teleseismic surface waves from both earthquake data and
ambient noise data, better lateral and azimuthal coverage of surface
wave paths can be achieved, which allows us to better image high
resolution heterogeneities of upper-mantle structures and improve
the ability to recover small-scale azimuthal anisotropy. Long-period
surface waves from ambient noise are complementary to earthquake
surface waves. Most of large earthquakes occur in oceanic plates either in subduction zones or in mid-ocean ridges and there are much
fewer earthquakes occurring in the interior of continents. Fortunately, there are a large number of stations in continents, which can
be used to create a large number of ‘virtual’ earthquakes by crosscorrelation of ambient noise. These ‘virtual’ earthquakes provide
‘fresh’ teleseismic surface wave data which can never be obtained
from earthquakes.
In the past decade, there has been rapid growth of large scale
seismic arrays deployed around the globe, such as USArray in
USA and CEArray in China, and meanwhile more and more permanent stations are also installed by many countries. Permanent
stations provide a global network of potential ‘virtual’ earthquakes
for portable arrays from which teleseismic surface waves can be
obtained for regional surface wave tomography as shown in this
study. All portable seismic arrays and permanent stations can also be
cross-correlated with each other, promising to provide a huge number of dispersion measurements, which can be easily incorporated in
global surface wave tomography to significantly improve data abundance and coverage, one of important aspects in advancing seismic
tomography.
1651
1652
Y. Yang
Shen, Y. & Zhang, W., 2012. Full-wave ambient noise tomography of the
Eastern Hemisphere, in Proceedings of the 2012 IRIS Annual Meeting,
Boise, Idaho, June 13–15, 2012.
Shen, W., Ritzwoller, M.H. & Schulte-Pelkum, V., 2013. A 3-D model of the
crust and uppermost mantle beneath the central and western US by joint
inversion of receiver functions and surface wave dispersion, J. geophys.
Res., 118, 1–15.
Snieder, R., 2004. Extracting the Green’s function from the correlation of
coda waves: a derivation based on stationary phase, Phys. Rev. E, 69,
046610.
Snieder, R., Wapenaar, K. & Wegler, U., 2007. Unified Green’s function
retrieval by cross-correlation; connection with energy principles, Phys.
Rev. E, 75, 036103.
Stockwell, R.G., Mansinha, L. & Lowe, R.P., 1996. Localization of the
complex spectrum: the S transform, IEEE Trans. Signal Process, 44,
998–1001.
Traer, J. & Gerstoft, P., 2014. A unified theory of microseisms and hum,
J geophys. Res., 119, 3317–3339.
Traer, J., Gerstoft, P., Bromirski, P.D. & Shearer, P.M., 2012. Microseisms
and hum from ocean surface gravity waves, J. geophys. Res., 117, B11307,
doi:10.1029/2012JB009550.
Yang, Y. & Forsyth, D.W., 2006a. Regional tomographic inversion of amplitude and phase of Rayleigh waves with 2-D sensitivity kernels, Geophys.
J. Int., 166, 1148–1160.
Yang, Y. & Forsyth, D.W., 2006b. Rayleigh wave phase velocities, smallscale convection and azimuthal anisotropy beneath southern California,
J. geophys. Res., 111, B07306, doi:10.1029/2005JB004180.
Yang, Y. & Ritzwoller, M.H., 2008. Teleseismic surface wave tomography in
the western U.S. using the Transportable Array component of USArray,
Geophys. Res. Lett., 35, L04308, doi:10.1029/2007GL032278.
Yang, Y., Shen, W. & Ritzwoller, M.H., 2011. Surface wave tomography
in a large-scale seismic array combining ambient noise and teleseismic
earthquake data, Earthq. Sci., 24, 55–64.
Yao, H., van der Hilst, R.D. & de Hoop, M.V., 2006. Surface-wave array
tomography in SE Tibet from ambient seismic noise and two station
analysis—I. Phase velocity maps, Geophys. J. Int., 166(2), 732–744.
Yao, H., van der Hilst, R.D. & Montagner, J.-P., 2010. Heterogeneity and
anisotropy of the lithosphere of SE Tibet from surface wave array tomography, J. geophys. Res., 115, B12307, doi:10.1029/2009JB007142.
Zheng, Y., Shen, W., Zhou, L., Yang, Y., Xie, Z. & Ritzwoller, M.H., 2011.
Crust and uppermost mantle beneath the North China Craton, northeastern
China, and the Sea of Japan from ambient noise tomography, J. geophys.
Res., 116, B12312, doi:10.1029/2011JB008637.
Zhou, L., Xie, J., Shen, W., Zheng, Y., Yang, Y., Shi, H. & Ritzwoller, M.H.,
2012. The structure of the crust and uppermost mantle beneath South
China from ambient noise and earthquake tomography, Geophys. J. Int.,
189, 1565–1583.
Zhou, Y., Dahlen, F.A. & Nolet, G., 2004. 3-D sensitivity kernels for surface
wave observables, Geophys. J. Int., 158, 142–168.
Li, A., Forsyth, D.W. & Fischer, K.M., 2003. Shear velocity structure and azimuthal anisotropy beneath eastern North America from
Rayleigh wave inversion, J. geophys. Res., 108(B8), 2362, doi:10.1029/
2002JB002259.
Lin, F., Ritzwoller, M.H. & Shapiro, N.M., 2006. Is ambient noise tomography across ocean basins possible?, Geophys. Res. Lett., 33, L14304,
doi:10.1029/2006GL026610.
Lin, F., Moschetti, M.P. & Ritzwoller, M.H., 2008. Surface wave tomography of the western United States from ambient seismic noise: Rayleigh
and Love wave phase velocity maps, Geophys. J. Int., 111, B12310,
doi:10.1111/j1365-246X.2008.03720.x.
Lin, F.C., Ritzwoller, M.H. & Shen, W., 2011. On the reliability of attenuation measurements from ambient noise cross-correlations, Geophys. Res.
Lett., 38, L11303, doi:10.1029/2011GL047366.
Liu, K., Levander, A., Niu, F. & Miller, M., 2011. Imaging crustal and upper mantle structure beneath the Colorado Plateau using finite-frequency
Rayleigh wave tomography, Geochem. Geophys. Geosyst., 12, Q07001,
doi:10.1029/2011GC003611.
Moschetti, M.P., Ritzwoller, M.H. & Shapiro, N.M., 2007. Surface wave
tomography of the western United States from ambient seismic noise:
Rayleigh wave group velocity maps, Geochem. Geophys. Geosys., 8,
Q08010, doi:10.1029/2007GC001655.
Nishida, K., Montagner, J.P. & Kawakatsu, H., 2009. Global surface wave
tomography using seismic hum, Science, 326(5942), 112.
Ren, Y., Stuart, G., Houseman, G., Grecu, G.B. & Hegedüs, E. South
Carpathian Project Working Group, 2013. Crustal structure of the
carpathian-pannonian region from ambient noise tomography, Geophys.
J. Int., 195, 1351–1369.
Rhie, J. & Romanowicz, B., 2004. Excitation of Earth’s continuous free
oscillations by atmosphere-ocean-seafloor coupling, Nature, 431, 552–
556.
Rhie, J. & Romanowicz, B., 2006. A study of the relation between ocean
storms and the Earth’s hum, Geochem. Geophys. Geosyst., 7, Q10004,
doi:10.1029/2006GC001274.
Sabra, K., Gerstoft, G., Roux, P., Kuperman, W.A. & Fehler, M.C., 2005.
Surface wave tomography from microseism in southern California,
Geophys. Res. Lett., 32, L14311, doi:10.1029/2005GL023155.
Schimmel, M., Stutzmann, E. & Gallart, J., 2011. Using instantaneous phase
coherence for signal extraction from ambient noise data at a local to a
global scale, Geophys. J. Int., 184, 494–506.
Schmandt, B. & Humphreys, E.D., 2010. Complex subduction and smallscale convection revealed by body wave tomography of the western U.S.
upper mantle, Earth planet. Sci. Lett., 297, 435–445.
Shapiro, N.M. & Campillo, M., 2004. Emergence of broadband Rayleigh
waves from correlations of the ambient seismic noise, Geophys. Res. Lett.,
31, L07614, doi:10.1029/2004GL019491.
Shapiro, N.M., Campillo, M., Stehly, L. & Ritzwoller, M.H., 2005. High
resolution surface wave tomography from ambient seismic noise, Science,
307, 1615–1618.

Geophysical Journal International

Transcription

Similar documents

theBoom "O"

• Power and Intensity • Doppler effect for (i) mechanical waves e.g.

Peace River - AEP - Government of Alberta

Waves and Wave Properties

Pantallas Acusticas Pared Verde (EN)

(El Método de Hidrodinámica de Partículas Suavizada)

August 10, 2016 Danny Page Valecraft Homes Ltd. / Jean

megawave - James Law Cybertecture

Sony CDX

1.68MB - Waves Inc.