SCEC Award Number 15177 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title Toward a Framework for Ground Motion Simulation Validation using Attenuation Relationships. Part 1: Calibration Between NGA-West2 Predictions, Physics-Based Synthetics, and Data
Investigator(s)
Name Organization
Ricardo Taborda University of Memphis
Other Participants Naeem Khoshnevis
SCEC Priorities 6e, 6c, 6a SCEC Groups GMSV, GMP, CME
Report Due Date 03/15/2016 Date Report Submitted 03/14/2016
Project Abstract
 This project aimed to develop the basis for a new framework for validation of ground motion simulations in the absence of data. Since this goes against the implication of validation referring to comparisons with records, the new framework should be such that it can be used to assess the level of (expected) accuracy for a given simulation even if records do not exist. The project’s primary objective was to propose and test a metric that could lend itself for comparing simulations against a well-accepted proxy for ground motion prediction (e.g., GMPEs) as a replacement to data. The project plan consisted of (I) proposing the validation formulation, (II) testing the formulation using simulations for which data do exist, and (III) correlating the comparisons with data against that of the proposed formulation when used with the chosen proxy (GMPEs). We completed steps I and II, and are in the process of completing step III. The outcomes of the project include the formulation of our new validation metric, which is based on the slope and amplitude misfit of a given intensity parameter (e.g., decay of PGV with distance); and results from initial tests using data from recorded earthquakes. While tests against GMPEs are still underway, our initial results indicate that the new metric has the potential to serve as a good predictor of the goodness-of-fit of simulations in the absence of data. This will help advance simula-tions, especially for validation of scenario earthquakes such as those used in physics-based PSHA models.
Intellectual Merit The research idea promoted by this project creates the foundation for a framework aimed to provide a validation alternative to current goodness-of-fit criteria and other ad-hoc validation methods, for validating simulations for which there is limited or no recorded data available. The intellectual merit of the project and the framework being developed resides precisely in that the new validation scheme will not depend on the existence of data, as opposed to current validation methods centered on comparisons with records from past earthquakes. The method will instead use well-accepted ground motion prediction equations as a proxy for the assessment of the accuracy of simulations. This concept will also help maximize the use of synthetic results over the full surface area of a simulation domain as opposed to being limited to the reduced number of locations (stations) for which there are observations. This is accomplished through the novel idea of treating single simulation datasets as rich sets of observations (data-points sample) and assume that these can be consider analog to observations from many events at a reduced number of stations (which is the concept at the core of developing GMPEs). The expectation is that the further development of this framework will lead to a methodology usable for validating scenario earthquakes. This will significantly contribute to SCEC projects promoted and helped by the GMSV activity group initiatives and contribute to the efforts of the UGMS committee, which in turn helps advance simulation initiatives important to SCEC scientists and engineers. A potential future implementation of the framework being developed by this work may also help the in other projects such as CyberShake and the Broadband Platform.
This project falls within proposal category B: theory and integration; it addresses SCEC’s fundamental problem 6: seismic wave generation and scattering: prediction of strong ground motions; and focuses on research priority 6e: collaboration with the engineering community in validation of ground motion simulations.
Broader Impacts The validation framework put forward by this project will help evaluate the expected accuracy of scenario ground motion simulations, which provides a basis for validating simulations such as those used in physics-based probabilistic seismic hazard analysis. Furthermore, because the framework being developed by seed funding from this project is based on concepts deeply regarded by the earthquake engineering community, completion of the project plan will translate into closer integration and ac-ceptance of simulations for engineering applications (e.g., building code provisions as intended by the SCEC UGMS committee).
On an educational front, this project has provided direct funds and a research opportunity for Shima Azizzadeh-Roodpish, a research assistant in the Center for Earthquake Research and Information (CERI) and Ph.D. candidate in doctoral program in Civil Engineering at The University of Memphis. Support from the project allowed Azizzadeh-Roodpish to attend the 2015 SCEC Annual Meeting, where she presented preliminary work leading to the development of the project activities.
Both PI Ricardo Taborda (Hispanic) and graduate student Azizzadeh-Roodpish (woman) belong to underrepresented groups in STEM fields.
Exemplary Figure Figure 1. PGV attenuation regressions (solid lines) as functions of distance (in log-log scale) for multiple events comparing results derived from data-points corresponding to stations for which there exist records (blue) and the synthetic data-points obtained at the same locations (red), contrasted with the regressions obtained from using the full simulation-domain gridded surface data-points (green line and gray dots). Validation metrics obtained from comparing station data and synthetic stations is to be used to calibrate validation the new validation metric for when comparing full simulation-domain results to GMPE predictions as if there were no records available.