Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

17
Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA

Transcript of Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Page 1: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Forecasting Magnitude from Fault Geometry

Bill Ellsworth, USGS Menlo Park, CA

Page 2: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Magnitude – Area RelationsM = log(A) + k

• Wells and Coppersmith (W&C, 1994) widely used in hazard analysis.

• Good agreement between W&C and kinematic rupture models derived from seismic waves.

• Application of W&C to WG02 fault model overpredicts historical seismicity rate.

• WG02 adopted 3 relations for large earthquakes:

M = 3.98 + 1.02 log(A) (W&C) M = 4.2 + log(A) (Ellsworth) M = 3.03 + 4/3 log(A) (Hanks & Bakun)

where A = Length x Width x R (seismic coupling factor)

Page 3: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

• Length (L): easy• Width (W): difficult;

disagreement between seismic and geodetic rupture models

• Aseismic slip factor (R): shallow creep – do-able; brittle-ductile transition – hard

• Trade-off between W and R: M = log(L) + log(WR) + k

Magnitude – Area RelationsM = log(LWR) + k

Page 4: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

M(A) and the Earthquake Cycle

Page 5: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Seismologists observethe coseismic rupture

M = log(LW) + k

Page 6: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Forecast modelsmust account for the

total slip budget

M = log(LWR) + k

Page 7: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

In this exampleR = 0.7

orlog(R) = -0.15

Page 8: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

If the coseismic ruptureis described by

M = log(LW) + 4.0

the forecast rupture is

M = log(LWR) + 4.15

Page 9: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

WG02 Approach to Determining W and R

Define W as the depth of the brittle-ductiletransition determinedfrom seismicity andthermal data

Use geodetic data to determine R given W

Page 10: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Depth of Seismicity and Depth to Brittle-Ductile transition in the San Francisco Bay Area

Colin Williams USGS, Menlo Park

Page 11: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.
Page 12: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

WG02 Fault L, W and R Values

R factor accounts for creep but not for aseismicslip at the brittle-ductile transition

If great earthquakes rupture into the brittle-ductiletransition W and R will be incorrect

Page 13: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

2002 Mw 7.9 Denali Fault, Alaska Earthquake

Page 14: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Denali Aftershocks LocationsRathkoviski et al.

Page 15: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

East (km)

No

rth (

km

)

-20 -10 0 10 20

-20

-10

010

20

1 m

Near-Fault Displacements from GPS Surveyand Rupture Depth of a Uniform Dislocation

Distance to Denali Fault (km)

Fau

lt P

ara

lle

l S

lip

(m

)

-20 -15 -10 -5 0 5 10 15

-3-2

-10

12

3

Displacement (m)

De

pth

of

Fau

ltin

g (

km

)

3 4 5 6 7 8 9

51

015

20

0.90.99

Best model: 5.95 m slip from 0 to 11.6 km

Fault displacement vectors along the

Trans-Alaskan Pipeline

Page 16: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Comparing Depth of Rupture EstimatesObtained from Geodesy and Seismology

Depth range ofseismic rupture models

Depth range ofgeodetic models

Page 17: Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.

Concluding Remarks

• Different approaches to the definition of W lead to different M(A) relations.

• The trade-off between W and R will be difficult to resolve with available data.

• The discrepancy between M(A) relations derived from earthquake slip models and those derived from earthquake cycle considerations can be explained by the R-factor.

• If rupture in large magnitude earthquakes routinely extends below the depth of complete locking the R-value in the current Working Group model will need to be modified if the W&C relation is used.