«1. US Geological Survey, Menlo Park, California 2. University of California, Berkeley 3. California Polytechnic State University, San Luis Obispo 4. ...»
Preliminary Report on the 22 December 2003 M6.5 San
Simeon, California, Earthquake
Jeanne L. Hardebeck1, John Boatwright1, Douglas Dreger2, Rakesh Goel3, Vladimir
Graizer4, Kenneth Hudnut5, Chen Ji6, Lucile Jones5, John Langbein1, Jian Lin7, Evelyn
Roeloffs8, Robert Simpson1, Keith Stark5, Ross Stein1, John C. Tinsley1.
1. US Geological Survey, Menlo Park, California
2. University of California, Berkeley
3. California Polytechnic State University, San Luis Obispo
4. California Geological Survey, Sacramento
5. US Geological Survey, Pasadena, California
6. California Institute of Technology, Pasadena
7. Woods Hole Oceanographic Institution, Woods Hole, Massachusetts
8. US Geological Survey, Vancouver, Washington In press at Seismological Research Letters, accepted January 7, 2004.
Introduction The MW6.5 San Simeon earthquake struck the central California coast on December 22, 2003 at 19:15:56 UTC (11:15:56 am local time.) The epicenter was located 11 km NE of the town of San Simeon, and 39 km WNW of Paso Robles (Figure 1), as reported by the California Integrated Seismic Network (CISN, the California region of the Advanced National Seismic System (ANSS)). The mainshock nucleated at 35.702N, 121.108W and a depth of 7.1 km, and the rupture propagated unilaterally to the SE. The strong directivity of the rupture resulted in a concentration of damage and aftershock activity to the SE of the hypocenter. The worst earthquake damage occurred in Paso Robles, where 1 two people died in the collapse of an unreinforced masonry building. The accurate and rapid earthquake information provided in near real-time by the CISN/ANSS to the Governor’s Office of Emergency Services made it possible to focus emergency response in the source area, although the earthquake was felt from San Francisco to Los Angeles.
The San Simeon earthquake occurred on a reverse fault striking NW and most likely dipping to the NE. Although motion along the Pacific-North American plate boundary in California is dominantly strike-slip, there is a small compressional component through central California. Repeated thrust earthquakes like the San Simeon event accommodate this compression, and build the Coast Ranges. Other recent thrust earthquakes in central California include the 1983 Coalinga (M6.4), and the 1985 Kettleman Hills (M6.0) earthquakes. Prior earthquakes in the vicinity of the San Simeon event include a M5-6 earthquake in 1853, a M5.7 earthquake in 1906, and the ML6.2 Bryson earthquake of 1952 (Figure 1) (McLaren & Savage, 2001.) The San Simeon earthquake occurred on a previously unknown blind thrust fault. No surface rupture associated with the earthquake has been identified. A number of roads, including state Highway 46, buckled due to the earthquake, but this deformation appears mainly to be failure of road fill due to ground shaking, and not the result of tectonic surface rupture. Extrapolation of the fault plane to the surface would roughly align with the surface trace of the Oceanic Fault, but this is thought to be a vertical strike-slip fault.
Two models for the kinematics of the region have previously been proposed. The first is a fault propagation fold model developed by Namson & Davis (1990) for the San Lucia
geometry is similar to, although more steeply-dipping than, the main blind thrust of this model, implying that this model may be applicable to the San Simeon region as well.
The second is the model of McLaren & Savage (2001), in which the region is dominated by strike-slip faulting with shortening on high-angle reverse faults. This model also may be applicable, although the dip of the San Simeon mainshock is shallower than predicted.
The San Simeon earthquake was followed by a vigorous aftershock sequence, with 165 events above M3 reported by the CISN within the first week of the mainshock. Although the event triggered many aftershocks, it did not significantly impact the seismicity rates of other nearby faults such as the San Andreas Fault and the San Simeon-Hosgri fault zone. The only triggered seismicity seems to be a few small events within the mainshock coda at the Geysers geothermal area, north of San Francisco. The San Simeon earthquake did, however, trigger shallow creep on the San Andreas Fault at Parkfield and hydrologic changes in hot springs in Paso Robles.
Mainshock Source Modeling The mainshock was first modeled as a spatial and temporal point source, using regional data from the CISN. The seismic moment tensor and the best source depth were determined by fitting seismic waveform data from the Berkeley Digital Seismic Network (BDSN/CISN). The focal parameters for the mainshock were found to be strike=290°, dip=58°, rake=78° and depth 8 km with a scalar seismic moment of 6.0e+25 dyne cm.
structure (Figure 1.) The P-wave first-motion solution found by the Northern California Seismic Network (NCSN/CISN) was strike=297°, dip=56°, rake=97°, very similar to the moment tensor solution.
A finite source model was also determined from the regional BDSN/CISN stations.
Broadband, three-component, displacement waveforms from 6 stations were inverted using the method of Dreger and Kaverina (2000) to determine the distribution of fault slip. The finite-fault modeling assumes a planar fault striking 290°, and dipping 58° to the NE. Although there is a slight preference for the NE dipping plane, the difference in fit using either of the moment tensor nodal planes is not significant. The aftershock distribution suggests the NE dipping plane, discussed below (Figure 2). The rake is held fixed at the moment tensor value of 78°. The fault dimensions are 44 km along strike, and 22 km along dip with 2 km by 2 km subfaults. The hypocenter is located at a depth of 8 km in the center of the fault. The fault dimensions are oversized for an Mw6.5 event to allow the data to determine the direction of the rupture. Slip positivity and derivative minimization smoothing (e.g., Hartzell and Heaton, 1983; Dreger and Kaverina, 2000) were employed to stabilize the inversions.
Assuming a single slip time window, the slip rise time and rupture velocity were found by performing inversions over a range of values. The data are rich in low frequencies, and we found that a rise time of 3 seconds and a rupture velocity of 2.1 km/s best fit the data. With this simplified initial fault model the slip was found to extend to the SE
nature of the slip is consistent with the shallow nature of the aftershocks, which are generally shallower than the mainshock’s 8 km depth (see below.) The peak and average slip were found to be 131 and 33 cm. The scalar seismic moment was found to be
5.7e+25 dyne cm, consistent with the long-period moment tensor result. The average slip and area of the main slip patch yields a static stress drop of 12.6 bars. With this simplified fault model, reasonably good fits to the data were obtained (Figure 3).
A second finite slip model for the 2003 San Simeon earthquake was developed from teleseismic P waveforms, obtained from IRIS/DMC data center (Figure 4). Two fault planes were first constructed using the NCSN/CISN hypocenter location and the CMT solution of SCSN/CISN and then modified to achieve a better waveform fit. A finite fault inversion using the method of Ji et al.  indicates that both nodal planes could fit the data well but that the NE-dipping plane (strike=307o, dip=50o) fits slightly better.
Modeling with either possible fault plane clearly shows that rupture propagated southeastward. The SE directivity can be seen by comparing the waveforms at stations PAYG and MAJO: the waveform at PAYG is much more compact than at MAJO. Since the ray path to the PAYG is roughly SE, and that to MAJO is roughly NW, only southeast propagation could explain the difference. The slip is concentrated around ~10 km depth, with a peak of ~50 cm, and is nearly pure dip-slip.
Both finite-fault models described above produce similar, fairly simple slip patterns, although the teleseismic model has lower slip. The majority of slip takes place on a small
much slip at the surface, suggesting a blind thrust fault. Both finite-fault models indicate that this earthquake had a significant component of directivity to the SE, providing an explanation for the relatively high levels of damage in Paso Robles and the high peak acceleration recorded in Templeton, both located to the SE of the epicenter.
Strong Motion, ShakeMap, and Community Internet Intensities The mainshock was recorded by three strong motion instruments in the near-source region. All three of these instruments are operated by the California Geological Survey (CGS) under the CSMIP and CISN programs. The distribution of peak motions indicates that the ground motion was strongly conditioned by the main-shock rupture directivity to the southeast (Figure 6.) The instrument near Cambria, only 13 km south of the epicenter, recorded a peak acceleration of 18% g, while the instrument at the crest of the San Antonio Dam, 22 km NNE of the epicenter, recorded 22% g (although dam crests are considered structural rather than free-field sites). The largest ground acceleration, 48% g, was recorded in Templeton, 38 km SSE of the epicenter but much closer to the SE end of the aftershock zone and the probable rupture area. The main shock was also recorded by CSMIP/CISN instruments further to the southeast, in San Luis Obispo (17% g) and Park Hill (15% g), as well as by instruments in Parkfield (3% g), Coalinga (3% g), and Simmler (7% g). The Parkfield Array, operated by CGS, recorded peak accelerations that ranged from 4% to 23% g, very similar to the range observed for the 1983 M6.5 Coalinga
the Array station closest to Paso Robles.
A preliminary comparison of the peak acceleration data for this event with that predicted by a standard relationship is useful. A plot of peak acceleration versus distance (log-log) for the records obtained to date is shown in Figure 5. The distances range from 12 km, for the Cambria station, to many stations at distances of over 250 kilometers. For reference, the Boore-Joyner-Fumal (BJF97, Boore et al., 1997) attenuation relationship is shown. (Coefficients for a reverse fault and an average shallow Vs of 700 m/sec were used; the thin line indicates distances beyond the suggested limit of the authors, 80 km).
The data shows reasonable agreement with BJF97 in its applicable range. Beyond that, higher attenuation with distance than predicted by the extrapolated BJF97 curve is indicated. These new data, and other recent data from digital instruments, allow extending the existing relationships to greater distances.
The point above the curve at about 40 km is Templeton, which had 0.48g, the largest value recorded in the earthquake; higher-than-expected acceleration is consistent with directivity-increased shaking in the rupture direction. The two closest stations, Cambria and San Antonio Dam, both plot below the curve, consistent with directivity-reduced values in the direction away from the rupture.
The absence of stations near the rupture initially limited the accuracy of the automatically-produced CISN ShakeMap. When a line-source model for the earthquake
the surface projection of the line-source. With the line-source included, ShakeMap adequately predicted the intensity in the near-field, including the MMI = 7-8 suffered by Paso Robles and Atascadero. The Templeton and Cambria records were retrieved and incorporated a few hours after the earthquake, further improving the ShakeMap. This highlights the importance of expanding the real-time data collection of strong motion stations to improve the usefulness of ShakeMap to emergency response.
The nearly unilateral rupture and consequent directivity in the earthquake contributed to the extensive building and chimney damage in Paso Robles and Atascadero. The largest CIIM (community internet intensity map) intensity reported for the San Simeon earthquake was the MMI = 7-8 in Paso Robles, Templeton, and Atascadero. MMI = 6 intensities were reported as far south as Santa Maria, and MMI = 5-6 intensities in Lompoc and Point Conception, some 100 km SSE of the earthquake. In the opposite direction, however, King City reported only an MMI = 5 intensity at a distance of 50 km from the epicenter. The elongation of the MMI = 6 region to the southeast corroborates the conclusion that there was substantial directivity in the main shock.
Geodetic Observations The earthquake produced measurable static displacements at the 14 continuously operating Global Positioning System (GPS) stations located closest to the event (Figure 7). All but one of the stations are northeast of the rupture, in the inferred hanging wall,
southwest, was observed at station CRBT (Camp Roberts). The cluster of stations near Parkfield all recorded southwestward movement of approximately 1.5±0.5 cm. The stations are too far away from the mainshock to constrain a detailed source model. As of one week after the earthquake, no post-seismic motion could yet be discerned at CRBT or any of the other nearby stations, nor was there any indication of transient deformation at sites along the San Andreas Fault.
In addition the co-seismic offsets, the GPS stations near Parkfield record at 1-second intervals. Using the method described by Bock et al. (2000), positions of each site relative to the master site, POMM, are estimated at each sample. The positions for the north component are shown in Figure 8. For the site, CRBT, which is closest to the mainshock, the peak to peak displacement is 17 cm and the co-seismic displacement is easily seen just after the arrival of the P-wave.
Geologic Field Observations The San Simeon area was searched both on the ground and by helicopter for signs of surface rupture due to the earthquake. No features that could be ascribed to coseismic surface faulting were found. Almost all the earthquake ground effects that were observed are best ascribed to rockfalls and landslides or to the settlement or slumping of man-made fills. Liquefaction was mapped in the Salinas River channel, in parts of Oceano on the coast, and west of Paso Robles.