GSJ 2004

The relationship of geodetic strain to seismicity on the island arc of Japan

Ali O. Oncel

Geological Survey of Japan

cc

Previous Research A) Fault Complexity/Magnitude

Intrerrelationship B) Seismic Cluster/Magnitude

Interrrelationship C) GPS-strain and Seismicity

Contents of Presentation

FRACTAL ANALYSIS OF FAULTING(Oncel, Wilson and Nishizawa, JGR-2001)

Correlation coefficients between b and DS computed along lines 1-3 are contoured to illustrate the spatial distribution of positive and negative correlation regions (areas 1-3) throughout Japan

4 16 64

Scene 1

D = 1.32

D = 0.93

r ~ 7

10

r

100

ND = 0.91

D = 1.31

r ~ 5

10

100N

1 10r

If pattern has fractal properties then

N = Cr-D

N :the number of occupied boxesr: the length of the box (r).

Seismic b-value(Oncel, Wilson and Nishizawa, JGR-2001)

Correlation coefficients between b and DS computed along lines 1-3 are contoured to illustrate the spatial distribution of positive and negative correlation regions (areas 1-3) throughout Japan

b=2.303/(Mmean-Mmin+0.05)

Space-time variability of b-value can be effected by several

physical factors such as:

• Material heterogeneity• Applied shear stress level • Coulomb failure stress• Thermal gradient • Fault complexity

Local scale 0.5 and 1.6. Regional scale equal to 1.

MAGNITUDE DISTRIBUTION

log

N

0

1

2

3

4

0 2 4 6 8

MAGNITUDE

23-31°E

b-value

Fractal Analysis of Seismicity(Oncel, Wilson and Nishizawa, JGR-2001)

• The spectrum of the generalized fractal dimensions (Dq, q = 0,1,2,....) of

seismicity data is determined by Correlation Integral.

• Estimated from the linear portion of the log-log plot of Cq versus distance, can be

used to evaluate the distribution for multifractal behavior.

• D2 Regional scale

• D15 Local scale

• D2>D3>……>D15 heterogeneous

• D2=D3=……=D15 homogeneous

D15

D2

15 1.1014 1.1113 1.1112 1.1211 1.1210 1.13 9 1.14 8 1.16 7 1.18 6 1.20 5 1.23 4 1.27 3 1.34 2 1.44

q Dq

Multiple FractalN=3912 q=2-15

DISTANCE R (KM)

CO

RR

EL

AT

ION

IN

TE

GR

AL

C( R

>r)

DISTANCE DISTRIBUTION

)1/(1

1

1

1

)(1)(

q

N

j

q

jq N

rRN

NrC

N

Line 2

Line 1

Line 3

Shikoku

Honshu

200km

200k

m

Correlations of seismotectonic variables (Oncel, Wilson and Nishizawa, JGR-2001)

ISTL ISTLISTL

MTL

MTL

b-value

D D

b-valueb-value

D

Active faults in Japan digitized from the 1:200,000 active fault maps produced by the Research Group for Active Faults of Japan (1991).

31

34

37

40

132 135 138 141 144

EURASIAN P

LATE

PACIFIC PLATE

Honshu Arc

Southwest

No

rth

ea

st

Ja

pan

Japan Arc

M7.0, 1600-1997

5.7M6.9, 1885-1997

4.1M5.6, 1926-1997

Completeness

JAPAN SEA

PHILIPPINE SEA PLATE

Mapping the correlations of seismotectonic variables

(Oncel, Wilson and Nishizawa, JGR-2001)

Correlation coefficients between b and DS computed along lines 1-3 are contoured to illustrate the spatial distribution of positive and negative correlation regions (areas 1-3) throughout Japan

-1.00

-0.80

-0.60

-0.40

-0.20

0.00

0.20

0.40

0.60

0.80

Area 1

Area 2

Area 3

5.00 to 6.00

6.00 to 7.00

1998-2003

M=6.22003.07.26

M=5.52003.07.26

Tohoku Events: Recent seismic events in the Tohoku region (M=5.5 and 6.2) were located in a positive correlation (Area III) noted to be anomalously quiescent. These intermediate magnitude events caused considerable damage and were characterized by higher than normal intensity for earthquakes of this magnitude.

Negative Correlations :Observed in the areas accommodates rupture on interconnected faults of larger total surface and therefore larger seismic moment.

Positive Correlations: Observed in the areas where stress is released by lower magnitude seismicity on smaller fault strands.

AcceleratingSeismicity

Correlation Dimension

Temporal variations of Seismicity (Oncel et al.,PAGEOPH-1996, Oncel and Wilson, BSSA-2001)

Negative Correlation

Temporal variations of Seismicity(Oncel et al.,Nonlinear Geophys-1995, Oncel and Wilson.BSSA-2001)

Increased b and decreased DC suggest that the rise in the level of low magnitude seismicity and high intensity clustering along the western portion of NAFZ did not completely release stress transferred into this segment of the fault zone.

This combination of factors - westward migration along with increased levels of low magnitude seismicity and higher intensity seismic clustering - are indicators of increased seismic risk in the area.

PositiveCorrelation

NegativeCorrelation

Low-magnitude events(Preshocks)

Large-magnitude events

Spatial variations of Seismicity (Oncel et al., Tectonophysics-1996, Oncel and Wilson, BSSA-2001)

33 38B la ck Sea

Aegea n Sea

NAFZ

12

3

45

6

7

The spatial variation of b and DC reveals a change in the dynamics of plate interaction along the length of NAFZ. However, the significance of those changes is not clearly understood. The areas of higher b and lower DC (more clustered seismicity) may be the result of creeping parts of fault zone. Whereas higher DC (less clustered seismicity) and lower b may be related to asperities along the fault (Öncel and Wyss, 2000).

(r=-0.85)

GPS and Earthquake MechanismReilinger, Toksoz and Barka, GSA-2000

We know that changes in the recent deformation observed by GPS measurement reveals dependence on fault type and velocity field. Velocity is lower for lateral motion while it is observed to be higher for vertical motion. Relationships between GPS derived deformation rate and Seismic Hazard has not, as yet, been proposed.

GPS and Hazard Parameters Modified after (Oncel and Wyss, 2000)

Changes in GPS measurement seems to be partially related to the fault patches of seismic (asperity) and aseismic (creep).

Especially, GPS velocity vectors is smaller in northern strand of NAFZ indicating higher seismic hazard while observing larger in its southern strand.

-400 -200 0 200 400

km

-200

0

200

400

600

800

km

- 8

- 6

- 4

- 2

0

2

4

6

8

1 0

1 2

1 4

1 6

1 0 y e a r A r e a S t r a i n s

Contour Map of the 10 year Area Strain Data(Wilson, Kano, and Nishizawa, in revision, PAGEOPH)

-400 -200 0 200 400

km

-200

0

200

400

600

800

km

-80

-70

-60

-50

-40

-30

-20

-10

0

10

20

30

40

100 Year A rea S tra in

Contour Map of the 100 year Area Strain Data(Wilson, Kano, and Nishizawa, in revision, PAGEOPH)

Correlations of seismotectonic variables

N

Line 2

Line 1

Line 3

Shikoku

Honshu

200km

200k

m

Station locations from which horizontal crustal strain measurements were made.

Active faults in Japan digitized from the 1:200,000 active fault maps produced by the Research Group for Active Faults of Japan (1991).

200k

m

200km

kilo

met

ers

kilometers

The Main Objective: to evaluate the relationship of geodetic strain to hazard parameters such as earthquake seismicity distribution, rate and magnitude.

Research Plan:

Correlation will be investigated for the two different periods of short (1883-1994) and long (1985-1994).

Correlation between strain data and hazard parameters will be computed locally along transects similar to those shown in Figures 3 and 4 of (Oncel, Wilson and Nishizawa, 2001).

What we propose?

Interrelationship between seismotectonic variables and geodetic strain

The occurrence of positive, negative or zero correlation between b and horizontal area strain will provide insight into the relationship between strain (e) and seismicity (b).

A variety of different relationships seem likely:

a) Positive correlation: could indicate anomalous behavior in the form of 1) gradual stress release in the form of relatively small magnitude earthquakes on the one hand or 2) stress buildup followed by sudden release of larger amounts of stored energy.

b) Negative correlation: could arise through increases of b associated with decreases in e or vice versa. Negative correlations seem to best fit our expectations.

c) No correlation: are likely to occur in transitions from positive to negative correlation or in areas where the fluctuations of individual parameters are random and statistically insignificant