The 20 July 2007 Oakland Earthquake

On 20 July 2007 a $M_{w}$ 4.2 earthquake occurred in Oakland, CA and was detected by the ElarmS NI processing. Initial event detection occurred 3 seconds after event origin, and the first magnitude estimate of M 4.1 was available one second later, or 4 seconds after origin. The magnitude estimate over the following 10 seconds varied between 4.0 and 4.3, eventually stabilizing on M 4.2 15 seconds after event origin.

Figure 2.33 shows the AlertMap at 4 seconds after origin, the time of the first magnitude estimate of M 4.1. Note the magnitude and location are recorded in the title of the map. The color scale represents the predicted modified Mercalli intensity (MMI) at all points based only on the event location and magnitude. At this time no stations have reported peak ground motions, so no bias correction has been applied to the empirical ground motion prediction equations (Newmark and Hall, 1982; Boore et al., 1997; Wald et al., 1999; Boatwright et al., 2003). The grey symbols represent stations which have reported triggers at this time, and the circular contours show the time in seconds until the onset of strongest ground motions (assuming a moveout speed of 3.75 km/s). At this time the epicenter location is somewhat south of the true location, and by looking at the map it becomes apparent that this is due to the fact that only stations to the north of the event have triggered at this time, leading to temporarily poor azimuthal coverage for this time period. This problem is eliminated in the following few seconds.

By the time 8 seconds have passed since the origin, the location has stabilized and several stations have reported observations of peak ground motion (Figure 2.34). The stations which have reported peaks are color-coded according to the corresponding MMI in this map, while stations currently experiencing peak ground shaking are filled in black. The observations of peak ground motion are used to compute a bias to the ground motion prediction equations, altering the predicted ground motions around the epicenter. Note that Oakland has already experienced peak ground motion (in fact it had already experienced the peak even 4 seconds after the origin), and San Francisco is just beginning to experience peak ground motion. However, much of the remaining Bay Area receives as much as 10 to 15 seconds warning. We find that the performance of ElarmS depends primarily on the density and distribution of stations around the source, so that the warning time for an event of this size would be comparable to that for a larger event ($M_{w} \sim 7$) if it were to nucleate in the same area.

An additional benefit to using ShakeMap algorithms to generate AlertMaps is this makes direct comparison between ElarmS predictions and the actual ShakeMap for each event. Figure 2.35 shows the ShakeMap for this event produced by USGS. A comparison between the ShakeMap and the AlertMap frames shows that even the first available ground motion predictions (Figure 2.33), which are based entirely on the event location and magnitude, are very similar to the observations shown in the ShakeMap. The availability of actual peak ground motion observations in Figure 2.34 helps to refine these predictions further.

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail:
© 2007, The Regents of the University of California