SCEC Award Number 18142 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title Continued Development of OpenSHA/UCERF3 in Support of Operational Earthquake Forecasting, Hazard Assessment, and Loss Modeling
Investigator(s)
Name Organization
Thomas Jordan University of Southern California Edward Field United States Geological Survey Kevin Milner University of Southern California
Other Participants
SCEC Priorities 4c, 4d, 5a SCEC Groups EFP, GM, CME
Report Due Date 03/15/2019 Date Report Submitted 05/02/2019
Project Abstract
 Substantial progress was made comparing the RSQSim full-cycle physics-based earthquake simulator to UCERF3 (published in Shaw et al. 2018), and computing deterministic ground motions with RSQSim rupture slip-time histories. We found strong agreement in hazard maps computed with the UCERF3 and RSQSim models. We also compared the RSQSim ruptures with the UCERF3 multi-fault rupture plausibility criteria, which found 80% agreement. Initial CyberShake PSHA calculations with RSQSim were completed in 2018. Previously, CyberShake has combined empirically-based ERFs (UCERF2, accessed through OpenSHA) with kinematic rupture generators to produce three-dimensional deterministic physics-based PSHA estimates. In 2018, we added capabilities in OpenSHA to use RSQSim catalogs as input to CyberShake, producing the first ever three-dimensional end-to-end physics-based PSHA assessment. These RSQSim-based CyberShake hazard curves were the first PSHA estimate computed without assuming any statistical distribution in the earthquake rate model, rupture generation, or ground-motion estimation.

We also rewrote much of the UCERF3-ETAS code in order to make it accessible to outside researchers. We held a training session at the 2018 SCEC Annual Meeting which was attended by 6 researchers, who learned to run the model locally and on HPC resources. The code, scripts, and examples are published on GitHub (https://github.com/opensha/ucerf3-etas-launcher). Additional progress was made on the USGS Fast-ETAS operational aftershock forecasting model, powered by OpenSHA. This included a new ability to efficiently compute hazard maps from ETAS forecasts directly on the users machine in minutes.
Intellectual Merit One of the three bullets in SCEC’s mission statement is to “Integrate information into a comprehensive, physics-based understanding of earthquake phenomena.” To that end, we computed the first probabilistic seismic hazard study using exclusively three-dimensional, physics-base models (RSQSim and CyberShake) in this reporting period. Validation efforts are underway to further evaluate the results, but we have demonstrated the feasibility of a fully-physics-based PSHA approach.
Broader Impacts OpenSHA, and it’s implementation of the UCERF3 models, continues to be a valuable tool for the SCEC community. OpenSHA is used by engineers, researchers, and students. OpenSHA is also used in conjunction with CyberShake to generate seismic hazard maps and to generate data products for the UGMS project. We held two training sessions at the SCEC Annual Meeting on OpenSHA products. The first was part of the greater “SCEC Software Products Workshop” which was attended by 37 members of the SCEC community. We also redesigned, documented, and released a version of the UCERF3-ETAS model for wider use. This included a small workshop for 6 researchers at the SCEC AM who learned to use the model and run simulations on HPC resources.
Exemplary Figure Figure 1
First probabilistic seismic hazard curves computed only with deterministic, physical models (RSQSim). This study uses rupture descriptions (including slip time histories) from RSQSim directly as input to the CyberShake 3-D PSHA simulation platform. Prior Cybershake studies have used kinematic rupture generators to extend empirical earthquake rupture forecasts. Initial hazard curves for the RSQSim catalog used in Shaw et al. (2018) is shown above, computed at the USC CyberShake site. The 3-D CyberShake-RSQSim curve is in black, 1-D BroadBand Platform-RSQSim in gold, and an empirical ground motion model (Abrahamson et al. 2014) in blue.
Credits: Kevin Milner (USC)