Observation and synthesis of seismic wavefields in basin structures analysis of seismic wave amplification at Alfredton Basin, New Zealand

Abstract

An experiment was mounted at a small basin near Alfredton, northern Wairarapa, New Zealand, to investigate amplification of seismic waves. S-wave parts of seismograms from 15 earthquakes, ranging in size from magnitude 2.5 to 5, recorded by 12 or more of the 19 seismometers deployed, have been analysed in two ways: spectral ratio analysis and peak ground motion analysis, assisted by SH wave modelling. The amplification patters of seismic waves have been obtained for the basin.

What has been learnt about the basin includes the following:

1. The response of a single rock or firm ground site is not stable enough to be used as a reference for amplification analysis. Therefore, it was very important to use responses at several firm sites to synthesise an average reference response that is less random than the responses at individual firm sites.
2. There is significant amplification for vertical components in the basin. Differences in the vertical component between soil sites and reference firm sites suggest that a ratio of the horizontal to vertical components at a soil site cannot be taken as an indicator of the amplitude of the horizontal amplification at a soil site in a basin like this in which the vertical component is amplified significantly.
3. Seismic amplification is directional in the basin. The north-south components have been amplified more than the east-west components by a factor of about 2 in the ratio of peak ground motion. If a single factor is required for evaluating the amplification at a site like this, a synthesised polar horizontal component in the time domain, or an average component in the frequency domain, rather than the n or e component, is an appropriate choice. 
4. Directional amplification shown in spectral ratios is less remarkable than that shown in peak ground motion amplification factors. Therefore both together, rather than spectral ratio alone, can give a more complete description of the amplification response of a site.
5. Amplification is event-dependent. The amplification for local events is significantly larger than that for distant events. The explanation for this, being due to the effects of the cancellation of seismic signals from different paths, is supported by SH wave modelling results. It follows that data from events that are as local as possible should be included in site response evaluation so as to avoid possibly significant under-estimation of the peak amplification response of a site.