Boatwright and Bundock (2005) produced a new Modified Mercalli intensity map in the form of a ShakeMap for the 1906 San Francisco earthquake reviewing all intensities reported in the Lawson report (Lawson et al, 1908; Figure 2.30A). We use this as a reference for the quake's actual ground motion to compare with ShakeMaps produced using subsets of the real time strong motion stations in Northern California from the BK, NC and NP networks. Ground motion for each site is taken from Graves (2006) synthetic ground motions for a 1906-type event ( 7.8, epicenter 2 mi W of San Francisco, rupture from San Jose to Cape Mendocino) and fed into ShakeMap.
Figure 2.30B shows the ShakeMap which would result if all 160 real time stations in the NCSS were able to transmit their data reliably to the datacenter where the ShakeMap is produced. Two things stand out in comparison with Figure 2.30A: The high level of ground shaking around the Bay Area and the extremely low estimate of shaking along the northern segment of the San Andreas Fault, where there are few stations. The latter factor can be somewhat alleviated by introducing the extent of fault rupture (Figure 2.30C). This information may be available within an hour of the quake, either from finite fault inversions or from the distribution of aftershocks. The estimated relatively slow decay of ground motion with distance from the fault is caused by the attenuation relationship currently used in ShakeMap (Boore et al, 1997).
Figure 2.30B may be considered a "best case scenario", with all real time stations reporting. The 1906 earthquake produced strong shaking. If the data center at the USGS Menlo Park were incapacitated, no data from the NC and NP networks would be available. In this case, the only data to constrain the initial ShakeMap would come from the 27 BK strong motion sites (Figure 2.30D).