CSNI Workshop – November 15-19, 2004 - Tsukuba, Japan

Using Borehole Data for Ground Motion Prediction:

Separating Source, Path, and Site Effects

Jamison Steidl, Dominic Assimaki, Kenichi Tsuda, and Ralph Archuleta

University of California at Santa Barbara, USA

Ground Motion Deconvolution

Estimation of Source and Path Parameters Using Borehole Data

2003 Miyagi-okiEarthquake SequenceKik-Net Borehole Stations28 Aftershocks M3-5 range used

Source and Path Inversion19 Kik-Net Borehole StationsStation Selection• Mainshock PGA ≥ 300 Gal• 19 Kik-net, 16 K-net Stations• Site Response after Source

InversionData Selection (S-wave portion)• 10-sec: Aftershocks • 20-sec: Mainshock

Source/Path Inversion: Method

Observed Spectrum (i-th event at j-th station):

*Source Spectrum (Boatwright 1978, Brune 1970)

*Attenuation Factor (Frequency Dependent)

Unknown Parameters

1( ) ( ) exp( ) ( )( )

iji j

s

fRO f S f R G f

Q f Vπ−= −

γfQo ⋅=Q(f)

γ,,, QofcMo

i4

i

2

i Mo)fc

f(1

)f2()Const()f(S+

π=

Source/Path Simultaneous Inversion1) Iterative Determination of Path (using 6 reference events)

2) Determination of Source (using all individual events)

Residual converges

Moγ,,Qofc

Residuals from Inversion= Borehole Response

Solve for

Solve for

Solve for fcMo, fixed are,γQo

Attenuation & corner fixed (0.5 - 1.0 Hz)

Moment fixed (0.5 – 20 Hz)

Inversion Results: Source

Inversion Results: Path Attenuation Model (Quality Factor)

Result0.74Q(f) 116f=

Borehole Residuals after fitting spectrum at 19 borehole stations

Ground Motion Deconvolution

Site

Site Response Results

19 Kik-Net Stations

Standard Surface/Borehole Spectral Ratio vs. Surface/Predicted Spectrum

Near Surface Attenuation Properties and Soil Behavior

Near-surface soil attenuation inversion using borehole array data from Garner Valley, California.Hybrid global/local inversion of shear-wave velocity profile, attenuation profile, and density using borehole data from Kik-Net and SCEC.Nonlinear effects at Kik-Net stations

Garner Valley (GVDA)NRC Sponsored Research Facility

Location in high seismicity region of Southern California

Garner Valley Instrumentation

Attenuation Analysis: Inversion Results at Garner ValleyFrequency dependent (red) and independent (black) attenuation with depth.

Array Simulation at Garner ValleyFrequency Dependent Q inversion

New NSF funded SFSI monitoring Project at Garner Valley

Future Analysis at GVDA

Simulation from Bedrock borehole motion to structural response, incorporating SFSI effects.

Garner Valley SFSI Test Facility

Ancestral Lake Bed – Soft Sediments (~20 m) over weathered granite (~70 m) and crystalline granite.

Uni-axial Accelerometer

Tri-axial Accelerometer

Rotation Sensor

Pressure Cell and Sensor Displacement Transducer

SFSI Monitoring at Garner Valley

: Natural Earthquake

Native Soil

Water Table 0-3 mPore Pressure

Accelerometer

F1

F2

F3

5m

5m

F2 : NEES@UCLA Shaker

F3 : NEES@UTexas Shaker

F1 : Permanent Shaker

Iterative Hybrid Inversion

Forward Model• Haskell-Thompson method using SH motion in

borehole as input to starting soil model.Iterative Inversion for Vs, Q, & Rho• Global (Genetic)

• Cross correlation between synthetic and observed motion in wavelet (time) domain

• Local (non-linear least squares)• Non-linear least-squares fit to empirical transfer

function in frequency domain.

0 1 2 3 4 5 6 7 8 9 10-15

-10

-5

0

5

10

15

Accelaration [gals]

Tria l Synthe t icRe cording

0 1 2 3 4 5 6 7 8 9 10-4

-2

0

2

4

Re cordingTria l Synthe t ic

0 1 2 3 4 5 6 7 8 9 10-10

-5

0

5

10

0 1 2 3 4 5 6 7 8 9 10-1

-0 .5

0

0 .5

1

0 1 2 3 4 5 6 7 8 9 10-0 .0 1

-0 .0 0 5

0

0 .0 0 5

0

0 .0 1

t [s e c]

Acceleration [gals]

Kik-Net Station IWTH04

Starting velocity model from Kik-Net web site

0 1000 2000 3000

0

10

20

30

40

50

60

70

80

90

100

110

Shear Wave Velocity [m/ s]

Dphm

Best fit model

Initial vector

0 0.1 0.2

0

10

20

30

40

50

60

70

80

90

100

110

Attenuat ion [1/ Q]

1 2 3 4

0

10

20

30

40

50

60

70

80

90

100

110

Density [kg/ cm 3]

Kik-Net Station IWTH04

0 5 10 15 2 0

250

5 0 0

750

10 0 0

125 0

150 0

1750FFT (Surface Re cording)FFT (Tria l S ynt he t ic )FFT (Bore hole Re coding)

0 5 10 15 200

10

20

3 0

4 0

5 0

f [Hz]

Surface/Borehole transfer function

The ore t ica l us ing t rial ve c t orEmpirica l s it e amplifica t ion

2 0 2 1 2 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9 3 0-3 0

-20

-10

0

10

2 0

3 0

Acceleration [gals]

Surface re cordingSynt he t ic (Input @ -10 9 m)

2 0 2 1 2 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9 3 0-3 0

-20

-10

0

10

2 0

3 0Abs olut e bore hole re cording (-10 9 m)

2 0 2 1 2 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9 3 0-3

-2

-1

0

1

2

3

t [s e c]

Re la t ive bore hole re cording (-10 9 m)

Simulated and Observed:Ground Motion Details (top) Average Site Response (bottom)

22 22.5 23 23.5 24 24.5 25-40

-20

0

20

40

T i [ ]

Acceleration[gals]

Surface Observa t ionBest Fit Model

Opt imiza t ion 2.0s window

2 4 6 8 10 12 140

10

20

30

F [H ]

Surface/Borehole transfer function Best Fit Model

Average linear sit e response from 28 events (10s)

Example 2: SCEC Borehole Data

Long Beach Water District (LBW)

SCEC Borehole Site LBWSoil Property Inversion Results

Input at 350 meters depthComparison of empirical vs. simulated surface motion and site response.

MainshockAnalysis

0 5 10 150

5

10

15

20

25

3 0

f [Hz]

FT (Surface) / FT (-109m)

Empirica l NL tra ns fe r func t ion ( -10 9 m)The ore t ica l us ing "be s t fit " e q. line ar mode lEmpirica l line a r tra nsfe r funct ion (-10 9 m)

Frequency (Hz)

Time (s)

2 0 2 5 3 0 3 5-10 0 0

-50 0

0

5 0 0

10 0 0

Acceleration [gals]

Synt het ic (Input @ -10 9 m)Surface Re cording

2 0 2 5 3 0 3 5-10 0 0

-50 0

0

5 0 0

10 0 0

Absolut e bore hole recording (-10 9m)

2 0 2 5 3 0 3 5

-10 0

0

10 0

t [s e c]

Relat ive bore hole recording (-10 9m)

Acceleration [gals]

Acceleration [gals]

Mainshock – Nonlinear Behavior

22 23 24 25 26 27 28 29 30-1000

0

1000

Surface Recording

SHAKEAcceleration [gals]

22 23 24 25 26 27 28 29 30-1000

0

1000

[ ]

Surface Recording

Eq. Linear w/ G/ Gmax(f),Q(f)Acceleration [gals]

0 5 10 150

5

10

15

20

25

3 0

f [Hz]

FT (Surface) / FT (-109m)

Nonline ar s it e re s ponseElas t ic s it e re spons eEq Line ar w/ G/Gmax(f) , Q( f)SHAKE

Frequency (Hz)Time (seconds)

Mainshock – Nonlinear Simulation

1-D wave seismic Wave propagation(Bonilla, 2000)Total Stress Analysis

(no pore pressure build up)

Conclusions

Source:• Borehole data provide a clearer picture of the

earthquake source.Path:• Borehole data is useful for examining crustal

attenuation models.Site:• Borehole arrays are useful for improving simulation

techniques for ground motion modeling of local soil conditions.

• Borehole data provides critical observations for modeling dynamic soil behavior at large strain.