The decrease in amplification and downward shift in resonance frequency of soil deposits as excitation strength increase are the typical nonlinear effect. Beresnev and Wen (1995) analyzed surface and downhole acceleration records from the SMART1 and SMART2 strong-motion arrays in Taiwan. First, they used sediment-to-bedrockspectral ratio (SBSR) technique to estimate frequency-dependent amplification of the alluvial deposits at the SMART1 array, where the rock site is taken as a reference one. Secondly, they analyzed the site responses between the Pleistocene and recent sedimentary deposits at SMART2 array.
Dimitriu and Theodulidis (2000) studied nonlinear effect by utilizing the horizontal -to-vertical spectral ratio (HVSR) technique using SMART1 array data. There are four soil and one rock stations with 23 earthquakes events (ML 4.9-7.0) recorded. Mean HVSR curves of each station are calculated for two PGA ranges: less than 75 cm/
and higher than100 cm/
, representing weak and strong-motion respentively. At the soil stations,the weak (linear) and strong (nonlinear) responses are significantly different in distinct frequency ranges. One soil site (O07)shows significant negative correlation between resonance frequency and ground acceleration.
Major characteristics of such behavior agree well with other empirical studies and theoretical predictions. The velocity and amplification reduction effects are well compatible with the acceleration level increase, and the SBSR and HVSR techniques exhibit similar results by resonance frequency decreases with increasing excitation level.