← Back to Sessions

How Well Does Poissonian PSHA Approximate Hazard Implied by Clustered Seismicity? a Synthetic Study

Session: Beyond Poisson: Seismic Hazards and Risk Assessment for the Real Earth

Type: Oral

Date: 4/19/2021

Presentation Time: 02:45 PM Pacific

Description: 

Classical Poissonian probabilistic seismic hazard assessment (PSHA) determines rates of mainshocks by removing foreshocks and aftershocks. However, earthquakes and ground motions cluster in multi-generational cascades of triggered seismicity. Here, we assess how well classical declustered Poissonian (DP) and nondeclustered (NDP) PSHA, and single-generation Sequence-based PSHA (SPSHA, Iervolino et al., 2014) approximate the hazard levels implied by the Epidemic Type Aftershock Sequence (ETAS) model. For simplicity, we focus on the time domain. We simulate a 5-million-year ETAS catalog of seismicity and ground motions. We then take these as quasi-real, use the above methods to assess hazard and compare with the ETAS-implied hazard. The degree to which traditional PSHA and SPSHA approximate ETAS-PSHA depends on probability levels and engineering demands. At design-oriented probability levels (2% and 10% in 50 years), DP-PSHA under-estimates, and NDP-PSHA and SPSHA over-estimate, ETAS-PSHA by less than 7%, suggesting the approximations work well at these levels. For probability levels larger than ~45% in 50 years, all methods over-estimate ETAS-PSHA: DP-PSHA by at most 7%; NDP-PSHA and SPSHA by at least 8% and 12%, respectively. This is due to higher median and mode seismicity rates estimated by the methods than implied by ETAS. Greater differences appear in the hazard curves of multiple exceedances, important for cumulative damage assessments. At and below 2% in 50 years, the probabilities of at least 5 and 10 exceedances are underestimated by more than 25% by traditional PSHA and SPSHA. Finally, we present a new simple model of the ensemble-averaged decay of hazard after large quakes that shows it can take years to centuries for the conditional hazard after M6.5 to M9 events to return to the long-term average. We conclude that refined PSHA should consider the influence of ETAS-like earthquake cascades for select engineering demands.

Presenting Author: Shaoqing Wang

Student Presenter: Yes

Authors