SCEC Award Number 12047 View PDF
Proposal Category Collaborative Proposal (Integration and Theory)
Proposal Title US-Japan Collaboration on Strong Ground Motion Prediction Techniques
Investigator(s)
Name Organization
Paul Somerville URS Corporation Arben Pitarka URS Corporation Jeff Bayless URS Corporation
Other Participants Professor Hiroshi Kawase, Kyoto University - leader of Japanese collaboration team, funded by the Japan Society for the Promotion of Science
SCEC Priorities 6e SCEC Groups GMP, GMSV, EEII
Report Due Date 03/15/2013 Date Report Submitted N/A
Project Abstract
 1. Implementation of the Irikura Recipe on the SCEC Broadband Strong Motion Simulation Platform
We received a Linux version of the code in which the GUI components had been stripped out. In particular, there was no Linux version of the GUI that is used to construct the earthquake source model, including the discrete asperities. It was therefore necessary to convert the GUI code into Linux. We also needed to adapt the Irikura Recipe to use the standard source representation used on the SCEC Broadband Platform, which is distinctly different from the source representation required as input by the Irikura Recipe.
2. US-Japan Collaboration on Strong Ground Motion Prediction Techniques
We have selected strong ground motion recordings of California earthquakes from the PEER database, and are analyzing average attenuation characteristics of the data using the same regression formula used by our Japanese counterparts. We will compare the regression coefficients that we obtain with those obtained by our Japanese counterparts using Japanese data in the same distance and magnitude range, and compare biases and standard deviations. We have developed earthquake rupture models for a suite of the scenario earthquakes which for magnitudes up to 7.0 are the same ones that are being used in the SCEC Broadband Validation Project to compare the simulations against ground motion prediction equations, which are fairly well constrained by data. We are now preparing to transfer the rupture models of these earthquakes to our Japanese counterparts for their use in their simulation methods, and to receive the rupture models from our Japanese counterparts for us to use in our ground motion simulation method.
Intellectual Merit The objective of this project is to compare strong motion simulation methods in California and Japan. Each side is simulating the ground motions of scenario earthquakes using their method but with the source parameters developed by the other side for the same scenario. This exchange of source parameters is important because SCEC’s experience has been that differences in source characterization are the main causes of differences in ground motion simulations performed using different simulation methods (SDSU, UCSB, and URS) on the SCEC Broadband Ground Motion Simulation Platform. We are also implemeting the Irikura Recipe for strong motion simulation, developed in Japan, on the SCEC Broadband Strong Motion Simulation Platform.
Broader Impacts The SCEC Broadband Platform’s SRF file contains a full description of the spatial and temporal evolution of rupture on the fault, which is partly stochastic. In contrast, the Irikura Recipe, which is used in Japan, is based on a deterministic model of asperities embedded within a background zone. This project has expanded the scope of the SCEC Broadband Strong Motion Simulation Platform to include the Irikura Recipe, and increased the understanding of differences in approaches to strong ground motion simulation between California and Japan.
Exemplary Figure Fig.1. Goodness of fit of recorded response spectra, corrected to hard rock site conditions, to ground motions simulated using the Irikura Recipe, for 40 stations of the 1989 Loma Prieta earthquake. The red line shows the mean residual, and the yellow bands show the 90% confidence interval of the mean. This shows that there is no significant bias in the prediction of the recorded response spectra for periods between 0.01 and 0.6 seconds and 2 to 3 seconds, and the simulations underpredict the recorded response spectra at other periods. The green bands show the standard deviation of a single measurement.