Full-3D Tomography of the Crustal Structure in Southern California Using Earthquake Seismograms and Ambient-Noise Correlagrams

En-Jui Lee, Po Chen, Thomas H. Jordan, Philip J. Maechling, Marine Denolle, and Gregory C. Beroza 1 SCEC, USC ; 2 University of Wyoming ; 3 Sripps , UCSD ; 4Stanford University

1 112 43

0 2 4 6 8 10 12 14 16 18 20 22 24 26

x 105

Iteration Number

EBP

EFP

ABP

AFG

AFP

SI AW SI

↓ add ANGF

↓ CMT

↓ CMT

↓ CMT

↓ CMT

E#=100

E#=135

E#=160

Bar

s: N

umbe

r of

Mea

sure

men

ts

Cur

ve: R

WM

Red

uctio

n (%

)

0

0.6

1.2

1.8

2.4

3

3.6

4.2

4.8

5.4

6

10

20

30

40

50

60

70

80

90

100

−120˚ −118˚ −116˚ −114˚

32˚

33˚

34˚

35˚

36˚

37˚

38˚Earthquake

4,671 windows

−120˚ −118˚ −116˚ −114˚

32˚

33˚

34˚

35˚

36˚

37˚

38˚ANGF (high&low)

3,176 windows

−120˚ −119˚ −118˚ −117˚ −116˚

34˚

35˚

36˚

(a)Isostatic Gravity Anomaly

1

2

3

4

5

6

7

8

−70−60−50−40−30−20−10

0

10

20

30mgal

−120˚ −119˚ −118˚ −117˚ −116˚

34˚

35˚

36˚

(b) Z2.5 of CVM−S4

SN

FZ

WWFGF

SAF

SAF

SJF

ECSZ

APF

1

2

3

4

5

6

7

8

−120˚ −119˚ −118˚ −117˚ −116˚

34˚

35˚

36˚

(c) Z2.5 of CVM−S4.26

2

3

5

1

4

7

8

60

1

2

3

4

5

6

7

8

Depth (km)

−120˚ −119˚ −118˚ −117˚ −116˚

34˚

35˚

36˚

(d) Z2.5 of CVM−H11.9

2

3

4

7

8

1

5

6

Abstract We have constructed a high-resolution model for the Southern California crust, CVM-S4.26, by inverting more than half-a-million waveform-mis�t measurements from about 38,000 earthquake seismograms and 12,000 ambient-noise correlagrams. The inversion was initiated with the Southern California Earthquake Center’s Community Velocity Model, CVM-S4, and seismograms were simulated using K. Olsen’s staggered-grid �nite-di�erence code, AWP-ODC, which was highly optimized for massively parallel computation on super-computers by Y. Cui et al. We navigated the tomography through 26 iterations, alternating the inversion sequences between the adjoint-wave�eld (AW) method and the more rapidly converging, but more data-intensive, scattering-integral (SI) method. Earthquake source errors were reduced at various stages of the tomographic navigation by inverting the wave-form data for the earthquake centroid-moment tensors. All inversions were done on the Mira supercomputer of the Argonne Leadership Computing Facility. The resulting model, CVM-S4.26, is consistent with independent observations, such as high-resolution 2D refraction surveys and Bouguer gravity data. Many of the high-contrast features of CVM-S4.26 conform to known fault structures and other geological constraints not applied in the inversions. We have conducted several other validation experiments, including checking the model against a large number (>28,000) of seismograms not used in the inversions. We illustrate this consis-tency with the excellent �ts at low frequencies (≤ 0.2 Hz) to three-component seismograms recorded throughout Southern California from the 17 Mar 2014 Encino (MW4.4) and 29 Mar 2014 La Habra (MW5.1) earthquakes, and we show these �ts to be much better than those ob-tained by two community velocity models in current use, CVM-S4 and CVM-H11.9. We con-clude by describing some of the novel features of the CVM-S4.26 model, which include un-usual velocity reversals in some regions of the mid-crust.

Figure 1. (a) Red solid line with circles: the relative waveform mis�t (RWM) for a set of waveforms selected for monitoring the improvements of our model. Vertical bars: the number of mis�t measurements used in each iteration. Colors of the vertical bars indicate the di�erent types of the mis�t measurements used. Vertical dash lines separate iterations carried out using the AW method from those carried out using the SI method. Black arrows indicate the iterations in which CMT inversions were performed or ambient-noise correlagrams were included. The number of earthquakes used in each iteration is shown following “E#=.” (b) Source-receiver paths for the 4671 earthquake waveforms used for monitoring the RWM reduction. Black stars: earthquake epicenters; red triangles: broadband stations. (c) Interstation paths for the 3176 Rayleigh waves on the ambient-noise correlagrams (Appendix A) used for monitoring the RWM reduction. Black triangle: stations used as virtual sources; red triangles: stations used as receivers. The source-receiver paths for these selected waveforms are shown as green lines in Figures 1b and 1c.

Figure 2. Maps of (a) isostatic gravity anomaly and (b~d) Z2.5 maps for CVMs in Southern California. The basin structures not well represented in the initial model, CVM-S4, are numbered: (1) Santa Maria Basin, (2) Southern San Joaquin Basin (SSJB), (3) Owens Valley (OV), (4) Indian Wells Valley (IWV), (5) Santa Barbara Channel (SBC), (6) Santa Monica Basin, (7) Antelope Valley (AV), and (8) San Bernardino Basin (SBB). Some faults are marked on (b): San Andreas Fault (SAF), White Wolf Fault (WWF), Sierra Nevada Fault Zone (SNFZ), Arroyo Parida Fault (APF), Garlock Fault (GF), Eastern California Shear Zone (ECSZ), and San Jacinto Fault (SJF).

Figure 3. (a) S wave velocities at (top) 2 km, (middle) 10 km, and (bottom) 20 kmdepths in (left) CVM-S4, (middle) CVM-S4.26, and (right) CVM-H11.9. The color bar on the upper left corner of each plot shows the range of the color scale with red indicating relatively slow S wave velocities and blue indicating relatively fast S wave velocities. Green dash line boxes on left column indicate the tomography region of Tape et al. [2009, 2010]. Black solid lines show major faults in Southern California. Numbers shown in the top row, middle column indicate locations of some geological features. 1: Southern San Joaquin Basin (SSJB); 2: Santa Maria Basin; 3: Salinian basement outcrop; 4: Cuyama Basin; 5: Ring-dike complexes; 6: Long-Valley Caldera; 7: Big Pine Volcanic Field; 8: Owens Lake; 9: IndianWell’s Valley (China Lake); 10: Searles Lake; 11: Santa Barbara Channel; 12: Antelope Valley; 13: Santa Monica Basin; 14: Catalina Schist outcrop; 15: San Bernardino Basin; 16: Peninsular Ranges Batholith.

−20

−10

0

WWF GF ʼAA

CVM−S4.26

−20

−10

0

WWF GF ʼAA

CVM−S4

−20

−10

0

0 20 40 60

WWF GF ʼAA

CVM−H11.9

LF GH−HF ʼBB

CVM−S4.26

LF GH−HF ʼBB

CVM−S4

0 20 40 60

LF GH−HF ʼBB

CVM−H11.9

3.5

SSF SGF SAF ʼCC

CVM−S4.26

SSF SGF SAF ʼCC

CVM−S4

0 20 40 60 80

3.5

SSF SGF SAF ʼCC

CVM−H11.9

4

WF SMFZ SAFSGF ʼDD

CVM−S4.26

WF SMFZ SAFSGF ʼDD

CVM−S4

0 20 40 60

WF SMFZ SAFSGF ʼDD

CVM−H11.9

5

ʼEE

CVM−S4.26

ʼEE

CVM−S4

0 20 40 60

ʼEE

CVM−H11.9

−20

−10

0

3.5

BF SAF JVF ʼFF

CVM−S4.26

−20

−10

0

BF SAF JVF ʼFF

CVM−S4

−20

−10

0

0 20 40 60

BF SAF JVF ʼFF

CVM−H11.9

2.25

2.50

2.75

3.00

3.25

3.50

3.75

4.00

4.25Vs (km/s)

−120˚ −118˚ −116˚

34˚

36˚

15481673

SOL

ADO

BOR

BTP

CIA

CLC

EDW2

IRMLAF

MCTMOP

SDD

SMR

SPG2

TUQ

RXH

0 25 50

SDD−Z

SDD−R

SDD−T

CVM−S4

0 25 50

CVM−S4.26

0 25 50

CVM−H11.9

25 50

ADO−Z

ADO−R

ADO−T

25 50 25 50

30 60 90 120

CLC−Z

CLC−R

CLC−T

CVM−S4

30 60 90 120

CVM−S4.26

30 60 90 120

CVM−H11.9

60 90 120

RXH−Z

RXH−R

RXH−T

60 90 120 60 90 120

60 90 120

SPG2−Z

SPG2−R

SPG2−T

60 90 120 60 90 120

60 90 120

IRM−Z

IRM−R

IRM−T

60 90 120 60 90 120

60 90 120

TUQ−Z

TUQ−R

TUQ−T

60 90 120 60 90 120

60 90 120 150

SMR−Z

SMR−R

SMR−T

60 90 120 150time (sec)

60 90 120 150

30 60

LAF−Z

LAF−R

LAF−T

CVM−S4

30 60

CVM−S4.26

30 60

CVM−H11.9

30 60

BTP−Z

BTP−R

BTP−T

30 60 30 60

30 60

EDW2−Z

EDW2−R

EDW2−T

30 60 30 60

30 60

SOL−Z

SOL−R

SOL−T

30 60time (sec)

30 60

30 60

CIA−Z

CIA−R

CIA−T

30 60 30 60

30 60 90

MOP−Z

MOP−R

MOP−T

30 60 90 30 60 90

30 60 90

BOR−Z

BOR−R

BOR−T

30 60 90 30 60 90

30 60 90

MCT−Z

MCT−R

MCT−T

30 60 90time (sec)

30 60 90

Community Velocity Models (CVM-S4, CVM-S4.26, & CVM-H11.9)

Iterative Inversions

Basin Structures

Figure 4. Cross sections of CVMs, including the CVM-S4, CVM-S4.26, and CVM-H11.9. On the velocity pro�les, the black lines represent projected fault surfaces from the SCEC Community Fault model (CFM) (14); the purple lines are from the Carena & Suppe’s fault model; brown lines are interpreted de-collement from LARSE surveys. The velocity contrasts on the cross-sections of CVM-S4.26 show high consistencies with the 3D fault models. The num-bers on the map indicate the (1) GF, (2) SNFZ, (3) SAF, (4) Elsinore Fault, and (5) SJF. Faults labeled on cross-sections for reference are Lenwood Fault (LF), Gravel Hills-Haper Fault (GH-HF), Santa Susana Fault (SSF), San Gabriel Fault (SGF), Whittier Fault (WF), Sierra Madre Fault Zone (SMFZ), Banning Fault (BF), and Johnson Valley Fault (JVF).

Fault Structures

Waveform Comparisons of La Habra Event (Not Used in the Inversion)Figure 5. Examples of observed (black) and synthetic (red) seismograms for the La Habra earthquake. For each station, observed and synthetic seismograms of the vertical, radial and transverse components are shown from top to bottom and synthetics computed using CVM-S4, CVM-S4.26 and CVM-H11.9 are shown from left to right. The beachball shows the CMT solu-tion used for computing the synthetics. Yellow star: epicenter location; white triangles: 3-component broadband stations whose seismograms are shown.

References & Acknowledgements

Lee, E., Chen, P., and Jordan, T. H. (2014). Testing Wave-form Predictions of 3D Velocity Models Against Two Recent Los Angeles Earthquakes, Seismological Re-search Letters, (in press) Lee, E., Chen, P., Jordan, T. H., Maechling, P. B., Denolle, M. A. M., and Beroza, G. C. (2014). Full-3D Tomography (F3DT) for Crustal Structure in Southern California Based on the Scattering-Integral (SI) and the Adjoint-Wave�eld (AW) Methods, Journal of Geophysical Re-search, (in press). Tape, C., Liu, Q., Maggi, A., Tromp, J., (2009). Adjoint tomography of the southern California crust. Science 325, 988–992. Support for this work was provided by Southern Cali-fornia Earthquake Center (SCEC).

CVM−S4 Vs @ 2 km

−120˚ −118˚ −116˚ −114˚

33˚

34˚

35˚

36˚

37˚

38˚

A

A’

D

D’

C

C’

F

F’

E

E’

B

B’

2.42.62.83.03.23.43.6

CVM−S4.26 Vs @ 2 km

−120˚ −118˚ −116˚ −114˚

1

2

3

4

6

7

8

910

11

12

13

14

15

5

16

2.42.62.83.03.23.43.6

CVM−H11.9 Vs @ 2 km

−120˚ −118˚ −116˚ −114˚

33˚

34˚

35˚

36˚

37˚

38˚

2.42.62.83.03.23.43.6

CVM−S4 Vs @ 10 km

−120˚ −118˚ −116˚ −114˚

33˚

34˚

35˚

36˚

37˚

38˚

3.03.23.43.63.84.04.2

CVM−S4.26 Vs @ 10 km

−120˚ −118˚ −116˚ −114˚

3.03.23.43.63.84.04.2

CVM−H11.9 Vs @ 10 km

−120˚ −118˚ −116˚ −114˚

33˚

34˚

35˚

36˚

37˚

38˚

34˚

36˚

38˚

3.03.23.43.63.84.04.2

CVM−S4 Vs @ 20 km

−120˚ −118˚ −116˚ −114˚

33˚

34˚

35˚

36˚

37˚

38˚

3.4

3.6

3.8

4.0

4.2

CVM−S4.26 Vs @ 20 km

−120˚ −118˚ −116˚ −114˚

3.4

3.6

3.8

4.0

4.2

CVM−H11.9 Vs @ 20 km

−120˚ −118˚ −116˚ −114˚

33˚

34˚

35˚

36˚

37˚

38˚

3.4

3.6

3.8

4.0

4.2

(b) (c)

(a)