PEER has just published Report No. 2013/22 titled “GEM-PEER Task 3 Project: Selection of a Global Set of Ground Motion Prediction Equations” as a new addition to the PEER Report Series. It was authored by Jonathan P. Stewart, John Douglas, Mohammad B. Javanbarg, Carola Di Alessandro, Yousef Bozorgnia, Norman A. Abrahamson, David M. Boore, Kenneth W. Campbell, Elise Delavaud, Mustafa Erdik, and Peter J. Stafford.
Ground-motion prediction equations (GMPEs) relate a ground-motion parameter (e.g., peak ground acceleration, PGA) to a set of explanatory variables describing the earthquake source, wave propagation path and local site conditions. In the past five decades many hundreds of GMPEs for the prediction of PGA and linear elastic response spectral ordinates (e.g., pseudospectral acceleration, PSA) have been published. The Global Earthquake Model (GEM) Global GMPEs project, coordinated by the Pacific Earthquake Engineering Research Center (PEER), brought together ground-motion experts from various institutions around the world to develop recommendations on what GMPEs should be used by GEM when conducting global seismic hazard assessments. The GEM-PEER Project has seven tasks, as listed below:
Task 1a: Defining a Consistent Strategy for Modeling Ground Motions
Task 1b: Estimating Site Effects in Parametric Ground Motion Models
Task 2: Compile and Critically Review GMPEs
Task 3: Selection of a Global Set of GMPEs
Task 4: Include Near-Fault Effects
Task 5: Build an Inventory of Recorded Waveform Databases
Task 6: Design the Specifications to Compile a Global Database of Soil Classification
This report presents the methodology used in, and results of, Task 3, Selection of a Global Set of GMPEs. The reports of the other tasks of the GEM-PEER project are published by the GEM foundation and posted at http://www.globalquakemodel.org.
For Task 3, a transparent and objective procedure was followed to reach the final recommendations. The procedure includes examination of the multi-dimensional (e.g., magnitude, distance and structural period) predicted ground-motion space, examination of functional forms, and evaluation of published quantitative tests of GMPE performance against observational data that was not used for their derivation.
The final selected GMPEs are a subset of the pre-selected models compiled during a previous task (Task 2) of this project. The recommendations for the prediction of PGA and PSA are:
- For stable continental regions: the models of Atkinson and Boore , Pezeshk et al. , Silva et al.  (double corner with saturation) [but with its aleatory variability (standard deviation) model replaced by the variability model of EPRI , and Toro et al.  as modified by Toro  to include finite-source geometric attenuation effects;
- For subduction zones: the models of Abrahamson et al. [2012, submitted], Atkinson and Boore , and Zhao et al. ;
- For active shallow crustal regions: the models of Akkar and Bommer , Chiou and Youngs , and Zhao et al. ; and
- For the three special regimes (volcanic zones, deep non-subduction Vrancea-type earthquakes and earthquakes with oceanic travel-paths), we recommend additional work be undertaken since the uncertainties in ground-motion prediction for these regimes is considerable.
Some of the recommended GMPEs have no site amplification term (i.e., three of the models for stable continental regions) or have linear site amplification terms. In either of those cases, we recommend application of the GMPEs for reference rock site conditions with modification by a nonlinear site term. The nonlinear site terms would ideally be region-specific, but in the absence of that information, we provide regime-specific recommendations. We consider the use of site amplification functions having nonlinearity to be important because linear models can significantly over-estimate ground motions for relatively large
magnitude/close distance conditions, which often control the results of probabilistic seismic hazard analysis.
We emphasize that the goal of this project is to select a set of GMPEs for global hazard
analysis; therefore, the number of selected GMPEs may be less than what might be used for sitespecific
analysis and/or development of national hazard maps. We also note that GMPE
development is a continuously evolving research area, and new and/or updated GMPEs are
published regularly as more empirical and simulated data become available and our knowledge
of ground-motion hazard expands. Thus, the set of GMPEs recommended here should not be
viewed as a long-term recommendation and should be re-evaluated on a regular basis.