High Resolution Surface Wave Tomography from Ambient Seismic Noise

1
High Resolution Surface Wave Tomography from
Ambient Seismic Noise
Mike Ritzwoller, Nikolai Shapiro, Greg Bensen,
Michel Campillo, Laurent Stehly University of
Colorado at Boulder Universite Joseph Fourier,
Grenoble
Earth is a noisy place!
2
Principal Conclusions

• Earths Hum (bass section normal mode studies
  250 s) possesses tenor, alto, and soprano
  sections (lt 5 sec to gt 100 sec) useful for
  surface wave tomography.
• 2. Surface wave Green functions are extracted by
  cross-correlating long noise sequences at two
  stations.
• 3. Ambient noise tomography improves the
  resolution of the crust and upper mantle.
  (USArray)
• 4. Source of this noise (our signal) appears to
  be oceanic.
• 5. This noise is not truly ambient, but
  possesses a strong directional component from
  off-shore storms.

R. Weaver, Science, 2005
3
Outline

• Simulations to illustrate the idea behind using
  ambient noise to study surface wave dispersion.
• Short period surface wave tomography (7 - 18 s)
  across California.
• Current tomography from a prototype ANSS
  continental backbone network at intermediate and
  long periods (16 - 150 s).
• Source location(s)? (Mostly from four winter
  months.)
• First observations of crustal P-waves (1 Hz) and
  Rg (0.5 Hz) from the Scripps team Roux,
  Gerstoft, Sabra, Kuperman, Fehler.

4
The Idea of the Method
Set-up of the Simulation 2 stations
5
The Idea of the Method
Add a source
6
The Idea of the Method
Add a source Zoom around receivers
7
The Idea of the Method
1
2
2
time (sec)
1
time (sec)
8
The Idea of the Method
Add another source randomly
1
2
2
time (sec)
1
time (sec)
9
The Idea of the Method
Two sources in line with the stations
1
2
2
time (sec)
1
time (sec)
10
The Idea of the Method
2
time (sec)
1
2
1
time (sec)
Green function for propagation between the
stations.
cross- correlation
lag (sec)
11
The Idea of the Method
Map the tendency for constructive
interference between nearby events.
grad (differential travel
time) min along outside the receivers
max between receivers
12
The Idea of the Method
Azimuthally homogeneous distribution of sources
1000s of sources, over 30 days
2
time (sec)
1
time (sec)
13
The Idea of the Method
Azimuthally homogeneous distribution of sources
1000s of sources, over 30 days
cross-correlation
lag (sec)
14
The Idea of the Method
Azimuthally homogeneous distribution of sources
1000s of sources, over 30 days
cross-correlation
lag (sec)
theoretical Green function
15
The Idea of the Method
Azimuthally inhomogeneous distribution of sources
1000s of sources, over 30 days
cross-correlation
lag (sec)
lag (sec)
lag (sec)
theoretical Green function
16
Outline

• Simulations to illustrate the idea behind using
  ambient noise to study surface wave dispersion.
• Short period surface wave tomography (7 - 18 s)
  across California.
• Current tomography from a prototype ANSS
  continental backbone network at intermediate and
  long periods (16 - 150 s) -- and indication of
  source location in 4 winter months (Nov 03 - Feb
  04).
• Source location(s)? (Mostly from four winter
  months.)
• First observations of crustal P-waves (1 Hz) and
  Rg (0.5 Hz) from the Scripps team Roux,
  Gerstoft, Sabra, Kuperman, Fehler.

17
Transportable Array (August, 2004) 62
stations
18
Example Rayleigh Wave Dispersion Curves
19
correlations computed over four different
three-week periods
PHL - MLAC 290 km
band- passed 15 - 30 s
20
correlations computed over four different
three-week periods
PHL - MLAC 290 km
band- passed 15 - 30 s
band- passed 5 - 10 s
repetitive measurements provide uncertainty
estimations
21
Short period ( 6 - 20 sec) surface wave
tomography across California
One month stack Aug 03 Measurements
retained SNR gt 10 /- lags consistent 62
stations BDSN, TriNet, Anza, TA 0.25 deg
grid Background model CUB Ray tomography Maps
at 7.5, 15, 18 sec.
22
Short period ( 6 - 20 sec) surface wave
tomography across California
One month stack Aug 03 Measurements
retained SNR gt 10 /- lags consistent 62
stations BDSN, Trinet, Anza, TA 0.25 deg
grid Background model CUB Ray tomography Maps
at 7.5, 15, 18 sec.
23
Resolution (ray-theoretic)
24
dispersion maps
high resolution tomography of the Californian
crust from ambient seismic noise
Central Valley
Ventura basin
Imperial Valley
LA basin
25
dispersion maps
high resolution tomography of the Californian
crust from ambient seismic noise
Sierra Nevada
Sacramento basin
Franciscan formation
Peninsular Ranges
Salinean block
San Joaquin basin
26
Comparison Between Traditional Teleseismic
Tomography and Tomography Based on Ambient
Seismic Noise
27
Outline

• Simulations to illustrate the idea behind using
  ambient noise to study surface wave dispersion.
• Short period surface wave tomography (7 - 18 s)
  across California.
• Current tomography from a prototype ANSS
  continental backbone network at intermediate and
  long periods (16 - 150 s) -- and indication of
  source location in 4 winter months (Nov 03 - Feb
  04).
• Source location(s)? (Mostly from four winter
  months.)
• First observations of crustal P-waves (1 Hz) and
  Rg (0.5 Hz) from the Scripps team Roux,
  Gerstoft, Sabra, Kuperman, Fehler.

28
ANSS-like Backbone broad-band, continental
scale application
29
ANSS-like Backbone broad-band, continental
scale application
Nov. 2003
30
CMB
Nov. 2003, G. Bensen
Example Receiver Gather 1 month of
data 33-67 s period
33-67 s
31
CCM
Nov. 2003, G. Bensen
33-67 s
32
HRV
Nov. 2003, G. Bensen
33-67 s
33
Broadband tomography across the US
Four month stack Nov 03 - Feb 04 Measurements
retained SNR gt 10 /- lags consistent 125
stations 1 deg grid Background model CUB Ray
tomography 10 - 20 sec 16 sec 20 - 50 sec
20 sec 33 - 66 sec 50 sec 70 - 150 sec 100
sec
34
16 s Rayleigh wave group velocity measurements
from ambient seismic noise (Nov., 2003-Feb.,
2004 4508 paths)
35
16 sec reference dispersion map (CUB global
tomographic model)
36
16 sec inversion of noise-based measurements
(Var. reduction 65)
37
20 s Rayleigh wave group velocity measurements
from ambient seismic noise (Nov., 2003-Feb.,
2004 4121 paths)
38
20 sec reference dispersion map (CUB global
tomographic model)
39
20 sec inversion of noise-based measurements
(Var. reduction 63)
40
50 s Rayleigh wave group velocity measurements
from ambient seismic noise (Nov., 2003-Feb.,
2004 1516 paths)
41
50 sec reference dispersion map (CUB global
tomographic model)
42
50 sec inversion of noise-based measurements
(Var. reduction 28)
43
100 s Rayleigh wave group velocity measurements
from ambient seismic noise (Nov., 2003-Feb.,
1004 May-June, 2004 1000 paths)
44
100 sec reference dispersion map (CUB global
tomographic model)
45
100 sec inversion of noise-based measurements
(Var. reduction 25)
46
Outline

• Simulations to illustrate the idea behind using
  ambient noise to study surface wave dispersion.
• Short period surface wave tomography (7 - 18 s)
  across California.
• Current tomography from a prototype ANSS
  continental backbone network at intermediate and
  long periods (16 - 150 s) -- and indication of
  source location in 4 winter months (Nov 03 - Feb
  04).
• Source location(s)? (Mostly from four winter
  months.)
• First observations of crustal P-waves (1 Hz) and
  Rg (0.5 Hz) from the Scripps team Roux,
  Gerstoft, Sabra, Kuperman, Fehler.

47
Whats and wheres the source of the Ambient
Noise?
The story for 5-10 sec microseisms is straight
forward ..
Source is from the coast and seasonal variability
is small.
48
Whats and wheres the source of the Ambient
Noise?
At longer periods the story is less straight
forward. Hint comes from the azimuthal dependence
of SNR.
20 - 50 sec, Nov 03 - Feb 04
49
Whats and wheres the source of the Ambient
Noise?
At longer periods the story is less straight
forward. Hint comes from the azimuthal dependence
of SNR.
20 - 50 sec, Nov 03 - Feb 04
20 - 50 sec, Nov 03 - Feb 04
50
Whats and wheres the source of the Ambient
Noise?
At longer periods the story is less straight
forward. Hint comes from the azimuthal dependence
of SNR.
20 - 50 sec, Nov 03 - Feb 04
SNR is more strongly a function of inter-station
azimuth than inter-station distance.
NW toward N. Pacific
NE toward N. Atlantic
51
Whats and wheres the source of the Ambient
Noise?
To really address this question Can use the
asymmetry of of the cross-correlations.
Nov 03 - Feb 04
SNR
SNR
52
Whats and wheres the source of the Ambient
Noise?
20-50 sec, Nov 03 - Feb 04
Asymmetry of the cross-correlation is a general
features of high SNR observations.
53
Whats and wheres the source of the Ambient
Noise?
33-66 sec, Nov 03 - Feb 04
Take high SNR observations. Plot lines between
stations where the cross-correlation is
strongly asymmetrical. Lines divide into two
group No. Pacific (red) No. Atlantic
(blue)
54
Whats and wheres the source of the Ambient
Noise?
33-66 sec, Nov 03 - Feb 04

• Invert for source locations
• consistent with inter-station
• azimuths.
• winter season
• source outside the US
• off-shore
• Pacific Gulf of Alaska,
• Bering Sea
• Atlantic somewhere in
• N. Atlantic
• probably continental shelf,
• not deep sea.

misfit (degrees)
55
Significant Wave Heights Aug - Dec, 1999

• Storms are seasonal and hemi-
• spheric -- stronger in the
• S.H. than the N.H.
• Southern storms are circum-
• polar.
• Note hurricanes hitting E.
• US in the Fall.
• Northern storms are at inter-
• mediate latitudes, but deliver
• much energy to continental
• shelves.
• Non-seasonal storms are stronger
• in the S.H. than the N.H.

From Bob Detrick
56
Whats and wheres the source of the Ambient
Noise?
Seasonal variability example 10 - 20 sec for
California stations ..
Unlike 5-10 sec microseisms strong seasonal
variability
57
Outline

• Simulations to illustrate the idea behind using
  ambient noise to study surface wave dispersion.
• Short period surface wave tomography (7 - 18 s)
  across California.
• Current tomography from a prototype ANSS
  continental backbone network at intermediate and
  long periods (16 - 150 s) -- and indication of
  source location in 4 winter months (Nov 03 - Feb
  04).
• Source location(s)? (Mostly from four winter
  months.)
• First observations of crustal P-waves (1 Hz) and
  Rg (0.5 Hz) from the Scripps team Roux,
  Gerstoft, Sabra, Kuperman, Fehler.

58
P waves Parkfield Network

• P-wave 5.5 km/s
• Rayleigh 2.5 km/s

San Andreas Fault
0.1-1.3 Hz

• ZZ component
• 30 seismic stations
• Max range 11 km
• 1 month average
• All station pairs 40-60 deg (main noise
  direction)

0.1-0.5 Hz
Spectrogram at 8.8km
Range (m)
0.7-1.3 Hz
Roux,P, P Gerstoft, K Sabra, WA Kuperman and, MC
Fehler, P-waves from cross- correlation of
seismic noise, submitted GRL, 2005
59
Conclusions

• Ambient noise tomography is a promising method
  that provides a much needed alternative to
  traditional teleseismic surface wave tomography.
• improved lateral resolution
• better constraints on crustal structure
• assessment of uncertainties.
• Many potential applications.
• improved crustal/mantle models
• hazard assessesment
• nuclear monitoring (location, shorter period
  Msmb).
• 3. The method remains in its enfancy -- more
  remains unknown than known about the method
• the nature, variability and source of the ambient
  noise
• optimizing the observational and inverse
  techniques.

60
(No Transcript)
61
P-waves Travel Times Measurements
0.7-1.3 Hz

• Extract P waves travel times from time-derivative
  of ambient noise correlations 0.7Hz-1.3Hz
• Good agreement with ray code prediction.

Range (m)
Time (s)
Ray code prediction
() crossing SAF
(O) East of SAF (fast)
() West of SAF (slow)
Roux,P, P Gerstoft, K Sabra, WA Kuperman and, MC
Fehler, P-waves from cross- correlation of
seismic noise, submitted GRL, 2005
62
Some Key Questions -- practical theoretical

• Methods to optimize azimuthal homogeneity of the
  ambient noise?
• Optimal time-series length?
• Bandwidth?
• Love waves? Overtones?
• Source of the seismic signal in the ambient
  wavefield e.g., seasonal variability?
• Conditions under which meaningful Green functions
  can be recovered i.e., when wont the method
  work?
• Nature and shape of the sensitivity kernel?

63
Comparison with Earthquake Records
64
Cross-correlations from ambient seismic noise at
US stations
frequency-time analysis of broadband
cross-correlations computed from 30 days of
continuous vertical component records
Shapiro Campillo, GRL, 2004.
65
The Idea of the Method
Azimuthally inhomogeneous distribution of sources
1000s of sources, over 30 days
cross-correlation
lag (sec)
theoretical Green function
66
Canada
67
Green Functions by Cross-Correlating Ambient
Noise in Antarctica?
Record section Cross-correlate 1 month of
ambient noise, Z
20 sec period Rayleigh wave
Bandpass centered on 20 sec