## How Much Farther? Estimating Rupture Length Probabilities After a Rupture Has Started

Steven G. Wesnousky, & Glenn P. BiasiSubmitted August 15, 2018, SCEC Contribution #8656, 2018 SCEC Annual Meeting Poster #041

The USGS ShakeAlert Earthquake Early Warning (EEW) system is designed to detect and alert within a few seconds of earthquake initiation. We investigate the probability that the rupture will grow, and develop estimates for eventual rupture length and magnitude. The condition that the rupture is under way distinguishes our estimates from a-priori rupture probabilities such as the Uniform California Earthquake Rupture Forecast v.3 (UCERF3) or that one might infer from a Gutenberg-Richter (GR) relationship. We investigate on three primary lines. For all models, we assume a UCERF3-like fault discretization with subsections of a few km in length, and that the EEW earthquake that has started roughly fills one subsection. A reference model that assumes equal probability ruptures of any length (2, 3, 4, … subsections) is appealing for its simplicity, but does not scale well with fault length. Applying a GR relation makes relative probabilities of length and magnitude look more realistic, but Field et al., SRL, 2017 question application of the GR on faults, and there are no fault physics inherent to the GR. We consider using UCERF3 time-independent probabilities of length and magnitude as proxies for the conditional probability case given an EEW initiation. If used directly, median predicted lengths for examples on the San Andreas and San Jacinto faults are unrealistically long (>350 km, M 7.8+). An ad hoc reduction by the number of subsections in the rupture improves the prediction, but still results in unrealistic probabilities of a long rupture. Bends and steps in rupture provide a physical basis for relative probabilities among fault ruptures, with support in empirical observation and fault dynamical modeling. We present probabilities for these physical “challenges” to rupture and show that they can be applied to produce conditional EEW length and magnitude predictions without ad hoc applications of GR or UCERF3.

**Key Words**

seismic hazard, earthquake early warning

**Citation**

Wesnousky, S. G., & Biasi, G. P. (2018, 08). How Much Farther? Estimating Rupture Length Probabilities After a Rupture Has Started. Poster Presentation at 2018 SCEC Annual Meeting.

**Related Projects & Working Groups**

Earthquake Forecasting and Predictability (EFP)