SCEC Award Number 21122 View PDF
Proposal Category Collaborative Proposal (Integration and Theory)
Proposal Title Investigation of Systematic Differences between CyberShake Studies 15.12 and 20.X Using Ground Motion Modeling Techniques Relevant to Seismic Hazard
Investigator(s)
Name Organization
Christine Goulet University of Southern California Xiaofeng Meng University of Southern California
Other Participants Scott Callaghan
SCEC Priorities 4b, 4c, 4d SCEC Groups GM, EEII, Seismology
Report Due Date 03/15/2022 Date Report Submitted 03/15/2022
Project Abstract
INTERIM REPORT:
We proposed to evaluate how much CyberShake simulations improved with the changes implemented in 2019, compared with the empirical data and empirically-derived GMMs (eGMMs). Unfortunately, the CyberShake simulation platforms encountered unexpected bugs after migrating onto a new supercomputing cluster. The CyberShake team spent lots of time and effort identifying and fixing the bugs, which significantly delayed the completion of the new CyberShake study we intended to use; it is now scheduled for mid 2022. However, an alternate CyberShake study was completed in early January 2022, which allows to perform a similar type of analysis, but we have not had enough time to complete the work before the SCEC report deadline. Hence we requested a no-cost extension.
Since the completion of the new CyberShake study (CS21.12), we developed a GMM for the study. We compared the median predictions of CS21.12 GMM with those from a prior CyberShake study in the same region (CS15.4), as well as those from empirical GMMs (Figure 1). We also compared the total variances and their decompositions from CS21.12 against CS15.4 and empirical datasets
In the next a few months, we plan to adapt a new spatially-varying coefficient (SVC) modeling approach to extract individual source, path and site effects from the CyberShake simulations. Our plan is to use these techniques to extract technically defensible components of the simulations. Our proposed work therefore serves as a proof of concept and a path forward for the integration of simulations into empirically-based PSHA.
Intellectual Merit Interim report - no-cost extension requested as stated in Abstract
Broader Impacts Interim report - no-cost extension requested as stated in Abstract
Exemplary Figure Figure 1. Differences of median ground motion predictions for 5s RotD50 between CS15.4 datasets and empirical GMMs (Left), and CS21.12 and empirical GMMs (Right). Colors denote differences in ground motions in logarithmic scale.