Advancing physics-based ground motion modelling of the 2016 Kaikoura Earthquake: Modelling basin-edge effects in Wellington and implications for seismic design

Executive Summary

Ground motions in the Wellington region observed during the 2016 Kaikoura earthquake exhibited appreciable amplitudes at long vibrations periods (T > 1 s) that resulted in relatively larger demands to multi-storey structures and consequent damage. Preliminary research by the authors and others has already illustrated that site amplification in the Wellington CBD region has occurred in past events - in particular, the 2013 Seddon and Lake Grassmere events, and therefore can be inferred to occur in future events also.

This report documents further research that has been undertaken to examine the systematic site effects (including so-called ‘basin-effects’) that exist in Wellington city as observed through ground-motion observations in multiple historical earthquakes, as well as informed through ground-motion simulations. In particular, 19 historical earthquakes above M = 5 in the Wellington region were used to understand systematic site amplification relative to a nearby reference station. It is illustrated that the amplifications consistently exceed those prescribed through the use of NZS1170.5 (code-based) site amplification factors - implying that the seismic hazard at these locations (all other things equal) is underestimated.

A second approach, examining the systematic difference between observed and predicted ground motions from empirical ground-motion models (GMMs), was used to construct non-ergodic site amplification factors which can be used in so-called non-ergodic probabilistic seismic hazard analysis. It is illustrated that at moderate-to-long vibration periods, both the ’reference site’ and ’non-ergodic’ site amplification approaches yield similar results.

Finally, the empirically observed results from ground-motion observations are used to examine the predictive capability of physics-based ground motion simulations with a preliminary version of a Wellington sedimentary basin model. It is illustrated that the simulations capture the basic features of basin amplification, however further fine-tuning is needed to capture the spatial variation in ground motion. These simulations highlight a hypothesis that the depth of sedimentary material overlying basement rock (including highly weathered material) is deeper than current inferences; and further illustrate the need for non-invasive and invasive data collection to provide improved constraint on the basin geometry and geophysical properties.

---

Technical Abstract

This report addresses three approaches to examine so-called ‘basin-effects’ that result in stronger-than-expected ground motion amplitudes at moderate-to-long vibration periods in Wellington region. Two of the approaches make direct use of observed ground motions to develop empirical models, while a third approach uses numerical simulations that are subsequently compared with data. The results of these approaches highlight:

1. Site amplification predicted using the NZS1170.5 (code-based) prescriptions underestimates the level of amplification that is systematically observed in 19 historical earthquakes above magnitude;
2. Two empirical approaches for estimating systematic site effects yield consistent results at moderate-to-long vibrations periods, and these form a basis for the development of short-term empirical adjustments to seismic hazard analysis results for the Wellington city region to account for such effects;
3. Ground-motion simulations provide significant potential to treat basin-effects in a holistic manner in Wellington (as well as other NZ locations), and are able to capture the essential features of basin amplifications, but would benefit significantly by additional data collection to constrain the geometry and geophysical properties of the Wellington basin.