The Mw 5.1 San Juan Bautista, California Earthquake of 12 August 199 R. Uhrhammer, L. Gee, M. Murray, D. Dreger and B. Romanowicz Seismological Laboratory University of California Berkeley, CA 9472 submitted to Seismological Research Letters October 2, 1998 Introduction An Mw 5.1 earthquake occurred at 1410 UTC on 12 August 1998 on the San Andreas Fault (SAF), 12 km SSE of San Juan Bautista (SJB), at a depth of 9.2 km. This is the largest earthquake to have occurred in the northern creeping-to-locked transition zone of the SAF in the past 50 years. It is of interest because it occurred at the northwestern edge of a gap in the seismicity, it is the largest earthquake in the San Francisco Bay Area to have exercised the U.C. Berkeley (UCB) automated Rapid Earthquake Data Integration system (REDI, Gee et al., 1996a), and it was large enough to have produced a detectable signal at the nearest station of the continuously recording GPS BARD network (Bay Area Regional Deformation Network). Seismo-tectonic setting The SJB region extends from the northern end of the creeping section of the SAF to the southern end of the surface rupture that occurred during the great 1906 San Francisco earthquake (Figure 1). It is a region of frequent moderate earthquakes and thus also a place where the processes of stable sliding (creep) and unstable sliding (earthquakes) can be studied. Sixty km SE of SJB the observed creep rate is approximately 90% of the 33 mm/yr SAF slip rate. Towards the NW the creep rate decreases gradually to essentially zero by the latitude of Watsonville, ~20 km NW of SJB (Schultz et al., 1982; Buford and Harsh, 1980; Thatcher, 1990). We estimate the seismic moment rate for this portion of the SAF to be on the order of 6.3 10^{21} dyne-cm/yr, based on data from the past 47 years, while it is 1.2 10^{23} dyne-cm/yr in the Loma Prieta area, where there is no creep. Assuming that the latter moment rate represents a slip rate of 33mm/yr, we then obtain a seismic slip rate of about 3mm/year in the creeping zone, indicating that the seismic slip rate complements the creep rate (30mm/year) on that portion of the SAF. Historical Seismicity and Seismic Potential The SJB section of the SAF has been very active historically with a background rate of 30 events per year at the M 1.8 threshold. For comparison, the rate in this region (per 1000 km$^2$) is 4.9 times the average background rate of seismicity for the entire Central Coast Ranges (CCR). Since 1951, 20 earthquakes of M 4.5 and larger have occurred within 25 km of the 1998 mainshock epicenter as shown in Table 1. The largest of these events, a $M_{L}$ 5.5 that occurred on 06 Jan 1986, was centered 17 km to the ENE on the Quien Sabe Fault (Figure 2a). The $M_{L}$ 5.0 event that occurred on 20 Jan 1960, located 2.2 km to the South, was closest to the 12 Aug 1998 event. Note that there is an apparent deficit of large ($M_{L}$ 5.7+) earthquakes in the region. Indeed, in 50 years, assuming that the Gutenberg-Richter relation is valid, we would expect 2 earthquakes of $M_{L}$ 5.7+ to have occurred in the region. The potential for a M 6+ earthquake to occur on the SJB segment of the SAF is low, except perhaps in the inferred "Cienega" seismic gap (Wyss and Buford, 1985) where anomalously low seismicity (relative to the adjacent SAF segments) prevails. A cross-section of the seismicity (Figure 2b) clearly shows that the 12 August 1998 event occurred at the NW end of this seismic gap. Rapid Response Although a relatively minor earthquake, the only reported injuries were a teenager falling from a bunk bed and a painter falling from a ladder (San Francisco Chronicle, 8/13/98), the 12 Aug 1998 San Juan Bautista earthquake illustrates several elements of the U.C. Berkeley/U.S. Geological Survey (UCB/USGS) joint earthquake notification system for northern and central California (Gee et al., 1996b). The joint system relies, on the one hand, on data from the dense 400+ component short period Northern California Seismic Network (NCSN) and Earthworm-Earlybird processing environment (Johnson et al., 1995), which provide rapid detection and location of earthquakes in central and northern California, and, on the other, on data from the broadband and strong-motion sensors of the Berkeley Digital Seismic Network (BDSN), Romanowicz et al., 1997) and REDI analysis software Gee et al., 1996a), which provide reliable estimates of local magnitude and moment magnitude, moment tensor, as well as peak ground motions. The processing centers at Menlo Park and Berkeley are connected by a dedicated frame-relay link through which waveform and parameteric data are exchanged in near real-time (Gee et al., 1996b). The first alert of the event at the BSL came when the threshold signal detector at BDSN station BKS triggered (except, of course, for those staff members who felt the earthquake) at 22 seconds after the origin time. BKS is located on the UCB campus, and the threshold detector at this station is monitored as a crude indicator of the occurrence of an event of interest. Triggering at a velocity of 10 microns/sec, the "seismic alarm" notifies BSL personnel by pager and serves as a "wake-up call" (literally, in this case, since it was 7:10 a.m.) that an event is in progress. A "quicklook" automatic location from the Earthworm system was produced at the USGS in Menlo Park within 30 seconds, based on P-wave detections at the nearest 25 stations of the NCSN. A final automatic solution with a coda magnitude of 5.3 was transmitted to the REDI processing system at UCB 217 seconds after the event occurred and posted on the Web page (http://quake.usgs.gov/recenteqs). The REDI processing system then determined local magnitude, ground motions, and the seismic moment tensor, based on data from the broadband and strong motion BDSN stations. Using the preliminary magnitude as a guide, the REDI system processed waveforms from 14 BDSN stations to determine a local magnitude of $M_{L}$ 5.35. The Earthworm location and the REDI local magnitude were paged 272 seconds after the event and the magnitude information was transmitted to the USGS. Data from BDSN strong motion instruments were processed and peak ground accelerations (PGA) that exceeded 0.5\% g were distributed by pager. Although the REDI system has had the capability to distribute ground motion data since 1995, this event is one of the few to have exercised the system. The observed PGA of 12.6% g is consistent with that predicted by attenuation relations such as those of Boore et al., 1997}. In the final step of automated processing for this event, the seismic moment tensor and moment magnitude were determined (Pasyanos et al., 1996). The automatic moment magnitude of $Mw$ 5.1, the strike-slip mechanism (130/89/-164) and the centroid depth of 8 km are in good agreement with the final human reviewed solution of magnitude 5.1, mechanism (129/85/-171) and depth of 8 km (Figure 3). All REDI processing was complete a total of 11 minutes after the origin time of the event. In spite of its small size, this earthquake was widely felt. The automatic earthquake information produced by the UCB/USGS system provided valuable information to emergency response managers. For example, the Peninsula Commute Service, which operates trains on the San Francisco Peninsula, was able to decide that only 1 train of the 14 running from Gilroy to San Francisco needed to be stopped while the track was inspected. Source Parameters The aftershocks are distributed predominantly to the NW of the mainshock epicenter (Figure 2), implying that the rupture propagated unilaterally towards the NW along the SAF. The broadband wavefield complexity, due primarily to the highly variable lateral structure in the CCR and the size of the event, render recovery of finite source effects not practical. The near-field and broadband regional waveform inversion solutions (Figure 3) are consistent with a right-lateral strike-slip event occurring on the SAF. The discrepancy of 0.25 units between the $M_L$ and $Mw$ magnitudes measured for this event is fairly typical for events of this magnitude in northern California. Indeed, the average value found for $M_L-Mw$ (Uhrhammer et al., 1996) is 0.26 $\pm$ 0.19 over the last 10 years. The difference can be explained by the fact that these two estimates of magnitude are based on measurements in different parts of the source frequency spectrum (below 0.1 Hz for $Mw$ and above 1Hz for $M_L$). Hypocenter Location The hypocenter from the joint notification system (2 km SW of SAO; automatic: 36.7548$^o$N, -121.4637$^o$E; reviewed 37.7533$^o$N, -121.4618$^o$E) is somewhat inconsistent with the observed ground displacement at BDSN station SAO and the moment tensor solution. SAO is located on the SW side of the San Andreas Fault zone, about 2 km from the trace. The observed 1.96 mm fault parallel ground displacement at SAO (Figure 4and the moment tensor solution (Figure 3) require that the hypocenter be located ENE of SAO. The reported location was computed using a 1-dimensional velocity model and station corrections developed for this region at the USGS Menlo Park (Fred Klein, personal communication). However, strong velocity contrasts exist across the San Francisco Bay Area faults, and it is well known that they are responsible for the hypocenters plotting systematically several km off the traces of the main faults (Figure 2; also Mayer-Rosa, 1973; Pavoni, 1973). For example, Boore and Hill, 1973} document a velocity change from 6 km/s on the west side of the SAF to 5 km/s on the east side. We relocated the hypocenter using an empirically derived two-dimensional velocity model and the newly developed adaptive migrating grid search location algorithm BWRELP (Broadband Waveform Regional Earthquake Location Program, Dreger et al., 1998}) which utilizes a combination of phase onset times and azimuth data. Combining phase onset times from SAO as well as NCSN stations located within approximately 30 km of the hypocenter and the $266^o$ P-wave phase onset azimuth at SAO, the event was located using a best-fitting velocity model with a 40\% velocity contrast across the fault (with the SW side fast). The resulting hypocenter is located 1.08 km E of SAO (36.766$^o$N, -121.434$^o$E, depth 9.5 km) and 1.4 km SW of the SAF trace. No station adjustments were used in conjunction with the 2-D velocity model for determining the hypocenter. Note that the 40\% contrast in velocity across the fault, determined empirically to obtain the best fit to the phase and azimuth data, is larger than that found by Boore and Hill, 1973, but is in good agreement with what one would expect based on estimating the rigidity ratio across the fault using a value of rigidity for granite in the Gabilan Range at SAO of ~2. 10^{11}$ dynes/cm$^2 (Carmichael, 1989}; Uhrhammer et al., 1996}) and of ~1.4 10^{11}$ dynes/cm$^2 for sedimentary rocks in the Diablo Range on the NE side of the fault (Carmichael, 198}). This solution is compatible with the observed fault parallel ground displacement at SAO and with the $62^o$ SW dip of the revised BDSN moment tensor solution. Thus, the reported epicenter was biased about 3 km to the SW (the fast side of the SAF) by the unaccounted for 40\% velocity contrast across the SAF in the vicinity of SJB. This velocity contrast also explains the observed trend in the hypocenters, plotting up to 4 km SW of the SAF trace in the vicinity of SJB (Figure 2}). Seismicity in SJB Sequence Region This earthquake sequence is the largest sequence that has occurred along this segment of the SAF since 1951. At a threshold magnitude of 1.8, the $M_L$ 5.35 mainshock had two foreshocks, a $M_L$ 3.14 event 7.6 minutes prior and a $M_L$ 2.80 event 1.3 hours prior. The foreshock and mainshock foci are located within a 600 m radius sphere. The sequence (through September 23rd) has had 51 aftershocks of magnitude 1.8 and higher, and the largest aftershock had $M_L$ 3.31. This sequence is fairly typical of the sequences which occur on the SAF in the region: (1) foreshocks have been observed in every sequence of 6 or more events in the region; (2) the largest aftershock was about 2 $M_L$ units below the size of the mainshock (typically 1.4-1.7 $M_L$ units below size of mainshock for strike-slip events in the CCR); (3) the rate of aftershock occurrence decayed exponentially with time. Broadband Near-field Waveforms and GPS Static Displacement The STS-1 broadband sensors at BDSN station SAO were driven full scale by this event. Figure 4 shows a comparison of the STS-1 raw signals and the equivalent signals derived from the FBA-23 strong motion accelerometers. The peak horizontal ground velocities inferred from the accelerometers were about twice the clipping level of the broadband sensors, although the broadband sensors recovered from the non-linear effects within a couple of seconds of the onset of the S-wave. The near-field effects are strongly apparent between the P-wave and S-wave onsets. Observations at SAO show that an FBA (Force Balance Accelerometer) located near the fault can resolve static ground displacements of this amplitude. Deconvolving the SAO FBA recordings to ground displacement and rotating to obtain SAF parallel and transverse ground motion, we find that SAO had a static offset of 1.96 ($\pm$ 0.061) mm to the NW. Static displacement to the NW (Figure 5) is consistent with a source on the SAF with the strike-slip mechanism indicated by the moment tensor solution. The transverse pulse width is 0.63 seconds, implying a rupture dimension of about 3.6 km and a corresponding displacement across the fault of order 18 cm for the $Mw$ 5.1 event. One of the permanently installed Global Positioning System (GPS) stations of the BARD network (SAOB), which is collocated with the seismic sensors at SAO, was measurably displaced by the August 12, 1998 earthquake. This is the first time a station in the BARD network has measured a displacement caused by an earthquake on the San Andreas fault system. GPS measurements of the location of SAOB relative to another BARD station 122 km southeast near Parkfield (PKDB) have a day-to-day scatter of about 2 mm in the horizontal prior to the earthquake (Figure 6). Because both stations are located several km southwest of the San Andreas fault, fault-parallel displacements primarily change the length of the baseline between the two stations. Figure 6 shows estimated coseismic offsets in the north, east, and length components of the baseline, assuming that rate of change of each component is the same before and after the earthquake. No significant vertical offset was detected, and no other BARD sites were measurably offset relative to PKDB, indicating that only SAOB was affected by the earthquake. SAOB moved to the north 2.6 $\pm$ 0.5 mm and to the west by 4.3 $\pm$ 0.6 mm, resulting in a total increase in baseline length of 4.7 $\pm$ 0.6 mm. This GPS-derived coseismic static displacement, which represents an average over the 6-week interval following the earthquake, is more than double that inferred from the more instantaneous FBA measurements at SAO. Continued aseismic slip on the fault following the earthquake (e.g., Heki et al., 1997}) is one possible explanation for this difference. Measurements at the nearby SJT borehole tensor strainmeter (e.g., Gladwin et al., 1994}), which show an instantaneous 0.5 microstrain coseismic offset followed by an additional 0.5 microstrain increase over the next 12-day interval, are consistent with this interpretation. The GPS measurements also suggest an increase in displacement following the earthquake, but the scatter in the observations is too high to confidently estimate the temporal variation. Discussion The moderate August 12, 1998 San Juan Bautista earthquake has illustrated the capabilities currently in place in northern California for rapidly reporting source parameters of significant earthquakes in the San Francisco Bay Area, provided by the combination of data from the dense short period NCSN network, the broadband and strong-motion BDSN network and the GPS BARD network. Improvements in the system, which we are currently working on at UCB, include increasing the speed of processing, expanding the capability of the REDI system, and enhancing the reliability of data distribution. In the current operating environment, 11 minutes were required to generate the automatic location, magnitude, peak ground motions and seismic moment tensor. In order to improve processing speed, we are working to deliver rapid magnitude estimates from the determination of near-field ramps (Uhrhammer et al., 1993; Gee et al., 1997). When completed, we anticipate that preliminary locations and magnitudes will be available within 1 minute after the event, in contrast to the current 4-6 minutes. For some applications, preliminary information, even if unreliable, is of great value in the initial assessment of an earthquake's impact (e.g., Savage et al., 1998). To expand the capability of the REDI system, we are developing methods for the automatic estimation of finite-fault parameters. These methods, based on analysis of waveform data (Dreger and Romanowicz, 1994; Dreger, 1997) and continuously recorded GPS data Murray et al., 1996) will allow the determination of the rupture plane length, width, and slip. Useful in themselves, these parameters may then be used to estimate strong ground shaking following a major earthquake. This is particularly important for regions that do not have a sufficient density of strong-motion instruments to allow direct determination of "shake-maps". Finally, we are working to improve the reliability of post-earthquake information in Northern California. Both the BSL and USGS facilities are located in the earthquake-prone San Francisco Bay Area, which renders them vulnerable to a major event. We have recently established a separate processing facility at the California Office of Emergency Services in Sacramento. This site operates independently of the UCB facility and receives data from a subset of BDSN (currently 6 stations). It is capable of detecting and processing events of magnitude 3.4 and higher, thus providing a "backup" to the UCB/USGS notification system. Acknowledgements We wish to acknowledge the fruitful collaboration with our colleagues at Menlo Park, especially David Oppenheimer, in the development of the joint UCB/USGS earthquake notification system. We thank Ray Baxter for helping with the GPS data processing and Hrvoje Tkalcic for helping out with the moment tensor data processing and helping out with the response to media following this earthquake. The REDI program is partially supported by USGS/NEHRP grant 1434-HQ-97-GR-03169. This is Contribution Number 98-08 of the Berkeley Seismological Laboratory. References Boore, D., and D. Hill (1973). Wave propagation characteristics in the vicinity of the San Andreas Fault, in Proceedings of the Conference on Tectonic Problems of the San Andreas Fault System, Kovach, R., and A. Nur (Editors), Stanford University Press, 215-224. Boore, D.M., W.B. Joyner, and T.E. Fumal (1997). Equations for Estimating Horizontal Response Spectra and Peak Acceleration from Western North American Earthquakes: A Summary of Recent Work, Seism. Res. Lett, 68, 128-153. Burford, R.O. and P.W. Harsh (1980). Slip on the San Andreas fault in central California from alinement array surveys, Bull. Seism. Soc. Am., 70, 1223-1261. Carmichael, R.S. (1989). CRC Practical Handbook on the Physical Properties of Rocks and Minerals, CRC Press, Boca Raton, pp. xiii and 741. Dreger, D. (1994). Empirical Green's function study of the January 17, 1994, Northridge mainshock ($Mw$=6.7), Geophys. Res. Lett, 21, 2633-2636. Dreger, D. (1997). The large aftershocks of the Northridge earthquake and their relationship to mainshock slip and fault-zone complexity, Bull. Seism. Soc. Am, 8, 1259-1266. Dreger, D., and B. Romanowicz (1994). Source characteristics of events in the San Francisco Bay region, U.S. Geol. Surv. Open-File Report, 94-176, 301-309. Dreger, D., R. Uhrhammer, M. Pasyanos, J. Franck and B. Romanowicz (1998). regional and far-regional earthquake locations and source parameters using sparse broadband networks: a test on the Ridgecrest sequence, Bull. Seism. Soc. Am, (in press). Gee, L.S., D.S. Neuhauser, D.S. Dreger, M.E. Pasyanos, B. Romanowicz, and R.A. Uhrhammer (1996a). The Rapid Earthquake Data Integration System, Bull. Seism. Soc. Am., 86, 936-945. Gee, L.S., A. Bittenbinder, B. Bogaert, L. Dietz, D. Dreger, B. Julian, W. Kohler, A. Michael, D. Neuhauser, D. Oppenheimer, M. Pasyanos, B. Romanowicz, and R. Uhrhammer (1996b). Collaborative Earthquake Notification in Northern California (abstract), EOS, Trans. Am. Geophys. U., 77, F451. Gee, L.S., D.S. Neuhauser, R.A. Uhrhammer, S. Fulton, and B. Romanowicz (1997). Getting REDI for Early Warning (abstract), EOS, Trans. Am. Geophys. U, 78, F45. Gladwin, M.T., R.L. Gwyther, R.H.G. Hart, and K.S. Breckenridge (1994). Measurements of the strain field associated with episodic creep events on the San Andreas fault at San Juan Bautista, California, J. Geophys. Res., 99, 4559-4565. Heki, K., S. Miyazaki, and H. Tsuji (1997). Silent fault slip following an interplate thrust earthquake at the Japan Trench, Nature, 386, 595-598. Johnson, C., A. Bittenbinder, B. Bogaert, L. Dietz, and W. Kohler (1995). Earthworm: A flexible approach to seismic network processing, IRIS Newsletter, XIV, (2), 1-4. Mayer-Rosa, D. (1973). Travel-time anomalies and distribution of earthquakes along the Calaveras Fault Zone, California, Bull. Seism. Soc. Am., 63, 713-729. Murray, M., D.S. Dreger, D.S. Neuhauser, L.S. Gee, P. Segall, and B. Romanowicz (1996). Rapid Finite Fault Determination from Geodetic Data (abstract), EOS, Trans. Am. Geophys. U., 77, F147. Pasyanos, M., D. Dreger, and B. Romanowicz (1996). Toward real-time estimation of regional moment tensors, Bull. Seism. Soc. Am., 86, 1255-1269. Pavoni, N. (1973). A structure model for the San Andreas Fault Zone along the northern side of the Gabilan Range, in Proceedings of the Conference on Tectonic Problems of the San Andreas Fault System, Kovach, R., and A. Nur (Editors), Stanford University Press, 259-267. Romanowicz, B., L. Gee, M. Murray, D. Neuhauser and R. Uhrhammer (1997). Real time access to multiparameter geophysical observatories in northern California, IRIS Newsletter, XVI (1), 6-8. Savage, W., N. Abrahamson, and M. Maclaren (1998). Useful products for electric utilities from integrated regional seismic networks (abstract), Seism. Res. Lett, 69}, 166. Schultz, S.S., G.M. Mavko, R.O. Burford, and W.D. Stuart (1982). Long-term fault creep observations in Central California, J. Geophys. Res., 87, 6977-6982. Thatcher, W. (1990). Present-day crustal movements and the mechanics of cyclic deformation, The San Andreas Fault System, California, U.S. Geol. Surv. Professional Paper, 1515, pp. 283. Uhrhammer, R. (1993). Rapid seismic moment tensor estimation using near-field waveforms recorded by broadband seismographs (abstract), EOS, Trans. Am. Geophys. U., 74, F400. Uhrhammer, R.A., S.J. Loper, and B. Romanowicz (1996). Determination of Local Magnitude Using BDSN Broadband Records, Bull. Seism. Soc. Am., 86, 1314-1330. Wyss, M. and R.O. Burford (1985). Current Episodes of Seismic Quiescence along the San Andreas Fault between San Juan Bautista and Stone Canyon, California: Possible Precursors to Local Moderate Mainshocks, U.S. Geol. Surv. Open-File Report, 85-754, 367-426. Table I M 4.5+ events within 25 km of the 08/12/98 SJB earthquake. This list includes the locations and magnitudes for the 1951-1968 time period recently obtained in the framework of the Historical Earthquake Relocation Project (HERP)} Note: HERP is a project that grew out of the needs of the Working Group on California Earthquake Probabilities 1999 to determine the probability of large earthquakes occurring in the San Francisco Bay Region (SFBR). The project involves transcription of the data on the original station reading sheets at UCB into a computer readable format and re-analysis of the location and magnitude of earthquakes (including formal estimates of the uncertainties), that have occurred in the SFBR, using modern analytical algorithms. While the reading sheets exist from 1910 onward, amplitude data, required for the calculation of $M_L$, was not systematically reported prior to 1951. From 1984 onward, the data are already in a computer readable format. Thus the HERP is concentrating on re-analysis of SFBR earthquakes from 1951 through 1983. 1951/10/31 20:58:19.65 36.9267 -121.3923 15.48 4.71 1959/05/26 15:58:01.08 36.7291 -121.5028 12.67 4.61 1959/12/29 02:32:54.25 37.0383 -121.3002 9.98 4.60 1960/01/20 03:25:52.22 36.7338 -121.4634 11.10 4.96 1961/04/09 07:23:15.41 36.7235 -121.2019 6.97 5.13 1961/04/09 07:25:43.79 36.8466 -121.2255 7.93 5.18 1963/09/14 19:46:17.11 36.8752 -121.6343 6.54 4.96 1969/10/27 10:59:42.75 36.7820 -121.3873 11.19 4.60 1970/03/31 07:02:28.27 36.8497 -121.4103 11.52 4.70 1972/09/04 18:04:40.64 36.6360 -121.2520 5.57 4.70 1972/10/03 06:30:02.23 36.8043 -121.5170 6.44 4.80 1974/11/28 23:01:24.56 36.9202 -121.4663 6.11 5.20 1980/04/13 06:15:56.28 36.7838 -121.5097 8.51 4.70 1981/01/07 11:42:32.99 36.8725 -121.6172 8.94 4.50 1982/08/11 07:46:42.92 36.6502 -121.2757 10.93 4.60 1986/01/26 19:20:50.93 36.8040 -121.2847 8.72 5.50 1986/05/31 08:47:55.59 36.6373 -121.2497 5.75 4.70 1988/02/20 08:39:57.24 36.7957 -121.3117 9.61 5.10 1990/04/18 13:38:10.15 36.9218 -121.6437 5.95 4.70 1998/08/12 14:10:25.11 36.7533 -121.4618 9.20 5.35 Figure Captions Figure 1: Map of the San Andreas fault zone in central California. The location of the 08/12/98 earthquake is identified with a star. The southern extent of the 1906 rupture along the San Andreas is marked with a thick line and the location of broadband seismic instruments at SAO, SCZ, MHC, and PKD are identified by squares. The site of creep measurements along the San Andreas and Calaveras faults are marked by circles, with the average rate in mm/yr. Figure 2: a) Closeup view of the location of the 08/12/98 earthquake. 20 years of background seismicity are plotted with "+". The filled star indicates the preliminary location of the mainshock and the filled circles are aftershock locations. Earthquakes located with simple 1-D models in this area locate west of the mapped trace of the San Andreas. The revised location of the main shock, obtained by modeling a 40\% velocity contrast across the fault, moves the epicenter closer to the mapped fault trace (open star). The box indicates the events plotted in the along fault (b) and across fault (c) depth sections. Figure 3: A) Automatic moment tensor solution. The fit between displacement data (0.01 to 0.05 Hz) and synthetics (dashed) computed using the GIL7 model (Dreger and Romanowicz, 1994 are shown. B) Human reviewed solution. In this case higher frequency (0.02 to 0.10 Hz) data are used at MHC and two new stations, CMB and KCC, were added to improve azimuthal coverage. Figure 4: Comparison of the broadband ground velocity inferred from the STS-1 broadband seismometers (dashed line) and the corresponding ground velocity inferred from the FBA-23 strong motion accelerometers. The FBA-23 signals were deconvolved to ground acceleration and then convolved with the corresponding STS-1 response. The STS-1 seismometers signals are clipped on all three components at approximately $\pm$ 10 mm/sec, and the peak ground velocities, inferred from the FBA-23 accelerometers, are up to three times as large as the STS-1 clipping level. The STS-1 signals recover rapidly from the clipping and again track the FBA-23 inferred ground velocities. The observed differences, up to approximately 15 percent, in the inferred ground velocities (where the STS-1 signals are not obviously clipped) can be attributed to non-linear response within the broadband seismometer feedback circuitry and to differences in the absolute calibrations and in the instrumental noise levels at frequencies above 5 Hz. Figure 5:As a result of the occurrence of the $Mw$ 5.1 earthquake, a ground displacement of 1.96 $\pm$ 0.061 mm to the NW and parallel to the SAF was observed at SAO. The fault parallel displacement was obtained by deconvolution and rotation of the FBA-23 strong motion accelerometers signals. SAO is located on the SW side of the SAF approximately 2 km from the trace. The P-wave onset, near-field ramp and S pulse are indicated on the figure. The S-wave pulse had a duration of 0.62 seconds and an amplitude of 6 mm. The 1.96 $\pm$ 0.061 mm offset was estimated from the difference in the displacement before the P-wave onset (dashed line from 1-1.5 sec) and after the S-wave pulse (dashed line from 3.5-5.5 sec). Note that the observed offset to the NW is compatable with the RLSS BDSN moment tensor solution for an event occurring on the SAF and that it is not compatable with the reported epicenter 2 km SW of SAO. As discussed in the text, this discrepancy is attributed to an approximately 3 km SW bias in the reported epicenter that is caused by a large difference in the average velocities across the SAF near SJB where the SW side of the SAF is fast. Figure 6: GPS estimates of the coseismic displacement at the SAOB permanent GPS station. Shown are daily estimates from early May to late September, 1998 of the north, east, and length components of the baseline from PKDB to SAOB. Scatter is about the mean and the error bars are one standard deviation. The vertical dotted line is the time of the August 12 earthquake. The offset was estimated assuming a constant linear rate of change in each component before and after the earthquake, shown by the solid line. The offset implies that SAOB was displaced northwest by 4.7 $\pm$ 0.6 mm by the earthquake, averaged over a 6-week interval.