THE BIGGEST POSSIBLE EARTHQUAKES: A MAXIMUM RUPTURE MODEL FOR THE SOUTHERN SAN ANDREAS AND SAN JACINTO FAULTS, CALIFORNIA SINCE A.D. 500

We produce a rupture history model for the southern San Andreas Fault (SAF) and San Jacinto Fault (SJF) that produces the longest possible earthquakes by connecting ruptures across neighboring sites with contemporaneous earthquakes. This “maximum rupture model” utilizes paleoearthquake event timing from 30 sites spanning 710 km of the faults from latitudes of Parkfield, California to the southern Salton Sea for the last 1500 years for most of the modeled domain. Shorter records limit the model domain on the Cholame, Carrizo, and southern Coachella sections of the SAF and the San Bernardino, and southern Clark/Coyote Creek sections of the SJF. In the modeled domain, the average rupture equates to Mw7.3, and ruptures occur on average every 30 years, but five gaps longer than 100 years are also permitted. The recurrence is thus not unlike the historic record that has four events in 1800, 1812, 1857, and 1918 and the absence of ruptures in the subsequent 100 years. Accounting for reductions in the model domain through time, calculated moment release is fairly constant but given rupture age uncertainties, periods of increased moment release are permitted. The model includes two ruptures, 192 and 254 km long, that connect the San Bernardino, Banning, and Coachella sections of the SAF through San Gorgonio Pass. The most recent of these ruptures ca. A.D. 1309 has a >4.2 m displacement in San Gorgonio Pass (Heermance and Yule, 2017). Based on timing alone, the model also permits as many as seven ruptures during which the SAF and SJF at Cajon Pass could have coeval rupture, and two of these would exceed 400 km in length (Mw 8.0) if they in fact represented a single event. Given the rupture rates on each section around the fault junction at Cajon Pass, however, at least 30% of the Mojave section ruptures do not also rupture both the SAF and SJF south of Cajon Pass. Empirical data from other faults and dynamic rupture models on rupture continuation at major fault splays support that full dual rupture is very unlikely. More likely, and also important for seismic hazard models, is that these epochs with similar-aged events on the major fault splays illuminate a signal of cascades or regional clustering.