Empirical ground motion prediction

Sponsors: Rutherford Discovery Fellowship (RSNZ), Marsden Fund (RSNZ), NZ Earthquake Commission (EQC), NZ Natural Hazards Research Platform (NHRP)

Collaborators: Misko Cubrinovski, Matthew Hughes, Jack Baker (Stanford)

Postgraduate researchers: Varun Joshi, Robin Lee, Karim Tarbali

Ground motions observed in the 4 September 2011 Darfield earthquake

The prediction of strong ground motions from large earthquakes is of principal importance in identifying seismic design loadings for structures in seismic regions. This work is focused on both improving New Zealand-specific empirical ground motion prediction models, and extending ground motion prediction capabilities worldwide via the consideration of uncommon ground motion intensity measures which correlate well with seismic response of engineered structures.

NZ-specific GMPE: Based on the foriegn ground motion prediction models, a New Zealand-specific pseudo-spectral acceleration model was developed in mid-2010 (Bradley, 2010 and Bradley, 2013) (see Software and Data for matlab codes). The subsequent Canterbury earthquakes have yielded an unparalleled dataset of near-source ground motions. Comparison of these observed ground motions with the Bradley2010 model illustrates that it provides a significantly improved prediction than conventional (i.e. McVerry et al. 2006) NZ-specific ground motion models (Bradley, 2012). The Canterbury earthquakes are being used to scrutinize various aspects of empirical ground motion prediction models, including: (i) forward directivity modelling; (ii) sedimentary basin effects; (iii) nonlinear surficial site response; and (iii) spatial correlation of ground motions.

Forward directivity: Observations of near-source directivity and empirical models for its consideration within ground motion prediction have been examined based on data from the 2010-2011 Christchurch earthquakes (Joshi, 2014). These include the probability of pulse occurrence, response spectral amplification due to forward directivity, pulse period, and pulse peak ground velocity. The explicit consideration of near-source directivity in NZ-specific PSHA has also been examined.

Spatial ground motion maps have also been developed over the Canterbury region based on the use of ground motion prediction equations and strong motion station recordings (See Software and Data).

Non-ergodic ground motion prediction: Using repeated ground motion observations of major events in the 2010-2011 Canterbury earthquake sequence the effect of non-ergodic ground motion prediction for Christchurch has been examined (Bradley, 2015).

Ground motion directionality: Ground motion prediction equations (GMPEs) often provide predicted levels of ground motion intensity for a single definition of bi-directional ground motion. As a result, it is useful to understand the ratios of the various directionality definitions so that empirical models can be developed to easily convert between available and desired definitions. Data from the Canterbury earthquakes have been used to assess the adequacy of existing directionality models and also the possibility of site-specific directionality (Bradley and Baker, 2015).

Damping reduction factors: Damping modification factors (DMFs) as a result of near-source forward directivity, basin-induced surface waves, and surficial soil response have been examined (Bradley, 2015). It is illustrated that ‘spectral peaks’ in the 5% damped response spectra have systematically different DMFs than suggested by conventional empirical formulas, and as a result, conventional DMFs will lead to a systematic over-prediction of the damped spectra for at vibration periods where source- and site-effects lead to amplification of response spectral ordinates.

Spectrum intensity-based GMPEs: This research has also developed theoretical relationships which can be used to predict three ground motion intensity measures: (Housner's) Spectrum Intensity (SI), Acceleration Spectrum Intensity (ASI), and Displacement Spectrum Intensity (DSI); from pseudo-acceleration prediction equations. The benefit of ASI, SI, and DSI is that they represent the average severity of ground motion for short, moderate, and long period vibration periods, respectively (Bradley et al. 2009, Bradley 2010, Bradley 2011). These ground motion intensity measures also have a smaller uncertainty than pseudo-spectral accelerations.