1. Finite-Source Models of the December 22, 2003 Mw6
Finite-Source Models of the December 22, 2003 Mw6.5 San Simeon, CA Earthquake and Applications to ShakeMap
Douglas Dreger1, Peter Lombard1, Jack Boatwright2, David Wald3 and Lind Gee1
1. Berkeley Seismological Laboratory
2. U. S. Geologic Survey Menlo Park, CA
3. U. S. Geologic Survey Golden, CO

2. Introduction
The determination of finite-source parameters such as the orientation of the fault plane, its dimensions, the slip distribution, and a kinematic description of how the slip occurs in time may be determined in near-realtime, and provide information that can improve estimation of near-fault strong shaking in areas that do not have the appropriate type of instrumentation. Dreger and Kaverina (2000) outlined such a system and applied it to the 1999 Hector Mine earthquake. Since 1999 the Berkeley Seismological Laboratory has integrated automated finite-source analysis in its realtime analysis system, and has developed tools to assist in the review and updating of the finite-source results. This system was operating at the time of the December 22, 2003 Mw6.5 San Simeon earthquake, and the derived finite-source information was used to provide additional constraint to published ShakeMaps.
This poster describes the evolution of ShakeMaps and finite-source models for the San Simeon earthquake, how finite-source information may be used to improve ShakeMaps, and uses the finite-source inversion method to investigate the source process of the earthquake and the source-related influence on recorded strong ground motion.

3. CISN Rapid Instrumental Intensity Map (RIIM)
CISN Rapid Instrumental Intensity Map (RIIM). Instrumental intensity is computed from recorded and estimated (from attenuation relationships) peak ground acceleration and peak ground velocity (e.g. Wald et al., 1999a and 1999b). Available recording stations are shown as triangles, the epicenter is shown as the star, and the line extending from the epicenter is the length of the ruptured fault inferred from finite-source models. This map shows that the greatest shaking extends SE toward the town of Paso Robles.

4. Community Internet Intensity Map (CIIIM)
Community Internet Intensity Map (CIIIM). This map shows the intensity plotted in zip code areas based on felt reports compiled from responses to an online questionnaire. There is good correlation between the RIIM and the CIIM to the extent that the coarse zip area mosaic allows. The maximum intensity was VIII. There was moderate to severe damage to unreinforced masonry buildings in Paso Robles.

5. Evolution of CISN ShakeMaps for the San Simeon earthquake
To the right the progression of ShakeMap updates is illustrated. Panels A and B compare ShakeMaps prepared using ML 6.4 and MW 6.5, respectively. The closest and only station on the map is at Parkfield, approximately 60 km to the ENE. Panel C compares the ShakeMap taking the dimension of the source (from the finite-source analysis) into account, with no additional recorded motion data. Panel D shows the map with the Templeton, CA data added. Panel E adds the Cambria, CA data. Panel F is the current ShakeMap. It is noteworthy that the addition of the dimension of the ruptured fault greatly extends and increases the instrumental intensity values to the SE of the epicenter. The subsequently added Templeton data confirmed this level ground motion.

6. C A E F B D

8. Automatic Finite-Source Inversion and Near-fault strong ground motion simulation
Invert the representation theorem following Hartzell and Heaton (1983)
Invert broadband (0.01 – 5.0 Hz) data for line-source and planar-source models
Determine fault plane
Determine source kinematics
Integrate source kinematics to simulate near-fault strong ground motion
Combine predictions and observations to produce shakemap
Dreger and Kaverina, 2000

9. Seismic Moment Tensor Solution
Seismic moment tensor determined from Hz, three-component, displacement seismograms. The source depth was found to be 8 km.

10. Automatic Line-Source
Location Map NW SE
Line-source models extending in both directions away from the hypocenter for both possible nodal planes were used to invert broadband displacement waveform data recorded by the BDSN. Blue triangles on the map show stations that were used. There is a good fit to the three-component broadband data (black). The line-source models both showed that the main slip was located SE of the hypocenter, however it was not possible to determine the causative rupture plane.

11. Reviewed Line-Source Used To Update ShakeMap
The line-source model was updated by adding additional stations, and applying timing delays to adjust for velocity structure. The revised model shows that the slip extends more than 22 km from the hypocenter. The dimension of this slip model was used to adjust the distance calculation used in ShakeMap. The black line in the ShakeMap plots shows the location of the line-source model

12. Comparison of Final and Line-source Adjusted ShakeMaps
The line-source adjustment is seen to substantially improve the ShakeMap in the near-fault region, and this map compares very favorably with the current maps with and without the adjustment.

13. Combined GPS and Seismic Waveform Finite-Fault Inversion
Combined inversion of three-component seismograms at 9 stations (blue triangle), and 2 GPS sites (black arrows).
Seismic data comes from the BDSN, NSMP, CSMIP, and TriNet.
GPS data was provided by K. Hudnut. The length of the defor-mation vector at CRBT is 53.2 mm.
The map view projection of the slip shows that it extends SE approx-imately 25 km with a very narrow depth range.

14. Composite Finite-Source Slip Model
Slip in the composite model is shallower than the focus at 8km depth, and lies primarily in the 3-8 km depth range. The scalar seismic moment, average static stress drop, average slip and the peak slip are 6.4e+25 dyne cm, 25 bars, 89 and 327. The slip distribution is unusual for a shallow dip-slip event in that its lateral extent is nearly 5 times greater than the vertical extent. This model is in general agreement with Yagi’s teleseismic result where he found a scalar moment of 8.8e+25 and peak slip of 2.2 m.

15. The fit to the broadband (0
The fit to the broadband (0.02-5Hz), three-component data (black) with the composite model synthetics (red) is very good. The records illustrate that the source is rich in longer periods.

16. The slip distributions for each of the six time windows are compared
The slip distributions for each of the six time windows are compared. Slip is largely contained in the first four time windows, representing a rise time variability of 1-2.5sec.

17. The Directivity Focusing Was Relatively Mild
Source finiteness has been shown to be important in terms of the distribution of the strong ground motions. It has been suggested that directivity also played a role. Simulated PGV for the composite source model (A) is compared to simulated PGV for the same model with a vertical strike-slip fault geometry (B). The strike-slip rupture triples the simulated PGV in some areas, and the region with greater than 10 cm/s is nearly quadrupled. Thus, while directivity was an important influence to the San Simeon strong ground motions, it was a relatively mild directivity.
The Directivity Focusing Was Relatively Mild

18. Forward Prediction of Templeton, Ca Velocity RecordsEW NS Z
The Templeton records were not used in the source inversion due to unknown timing and site effects (NEHRP C/D). Neverthless the forward prediction of the velocity seismograms obtained by multiplying the rock synthetics by a factor of 1.62 (adjustment for 1.5 km/s to 400 m/s site correction) produces fairly good fit.

19. Conclusions
The combined inversion of GPS and seismic waveform data for the kinematic rupture process of the San Simeon earthquake reveals an elongate rupture over a narrow, shallow depth range. The peak slip of 327 cm is unusually large for a Mw6.5 event and contributed to the strength of the ground motions.
While directivity was also a factor in the strength of the ground motions comparative simulations between the dip-slip case and a hypothetical strike-slip case reveals that a Mw6.5 strike-slip event with identical slip distribution would result in about a 3-fold increase in peak PGV, and a 4-fold increase in the area experiencing greater than 10 cm/s. Considering that the 1952 Bryson event was strike-slip it is important to note that directivity in the 2003 San Simeon earthquake was relatively mild and not necessarily representative of the hazard due to Mw6.5 events in the region.
The automatic finite-source ground motion estimation method of Dreger and Kaverina (2000) was shown to provide important constraints on the extent of the ruptured fault, which significantly improved ShakeMap representations of near-fault strong ground motion.

20. References