Operational earthquake forecasting during the M6.4 Searles Valley and M7.1 Ridgecrest sequence using the UCERF3-ETAS model—evaluation and lessons learned

Kevin R. Milner, Edward H. Field, William H. Savran, Thomas H. Jordan, Morgan T. Page, & Maximilian J. Werner

Submitted August 9, 2019, SCEC Contribution #9401, 2019 SCEC Annual Meeting Poster #271 (PDF)

Poster Image: 
By 11:10 am on July 4, 2019 (36 minutes after the M6.4 Searles Valley earthquake), the first UCERF3-ETAS aftershock simulations were running at the University of Southern California’s High-Performance Computing Center. UCERF3-ETAS (Field et al., 2017), an extension to the Third Uniform California Earthquake Rupture Forecast, is the first comprehensive fault-based epidemic-type aftershock sequence model. It produces ensemble simulations of aftershock sequences both on and between explicitly modeled UCERF3 faults to answer a key question in earthquake forecasting: What are the chances that an earthquake that just occurred will turn out to be the foreshock of an even bigger event? At 11:39 am, the first results were posted to the Southern California Earthquake Center’s scientific response coordination page: the UCERF3-ETAS model predicted a 3% chance of triggering an aftershock larger than M6.4 within the first week.

Such an event occurred 34 hours later, this time a much more powerful (M7.1) earthquake. Aftershock probabilities were recalculated and provided to the U.S. Geological Survey and the California Earthquake Prediction Evaluation Council (CEPEC) for use in assessing the probabilities of a larger shock. As seismic activity propagated southeastward to the Garlock Fault, a particular concern was the possibility that the Ridgecrest sequence could trigger a large earthquake on this major fault. Standard short-term forecasting models in current use by the USGS and CEPEC do not explicitly consider the proximity of seismic activity to major faults like the Garlock. In contrast, the UCERF3-ETAS model has been explicitly designed to produce such probabilities.

Development of the UCERF3-ETAS code base during the past year allowed us to rapidly prepare Ridgecrest simulations. Moreover, new tools were quickly developed in the weeks following the M7.1 event, including the incorporation of 3-D finite rupture models, which allowed us to account explicitly for the distance between the observed rupture surfaces and neighboring faults. We will share lessons learned from this real world exercise including sensitivity of results to poorly resolved initial finite rupture models, forecast evaluations using methods defined by the Collaboratory for the Study of Earthquake Predictability, and future model developments.

Key Words
UCERF3,ETAS,WGCEP,OEF,Forecasting,Ridgecrest

Citation
Milner, K. R., Field, E. H., Savran, W. H., Jordan, T. H., Page, M. T., & Werner, M. J. (2019, 08). Operational earthquake forecasting during the M6.4 Searles Valley and M7.1 Ridgecrest sequence using the UCERF3-ETAS model—evaluation and lessons learned. Poster Presentation at 2019 SCEC Annual Meeting.


Related Projects & Working Groups
Ridgecrest Earthquakes