Search

This study analyzed site effects including PGA and empirical transfer functions at 15 selected surface–downhole stations by using ground motion recordings of earthquakes with ML > 4 during 2012 and 2013 in Taiwan. In addition, we used all strong motion recordings of four large earthquakes (ML > 6) to generate intensity distribution maps and assess the differences in magnitude at the surface and downhole. The site amplification factors of the PGA were calculated using the ratio between the surface and downhole recordings. The mean PGA amplification factors ranged from 2 to an exaggerated value of 20 at different stations. In addition, the power law relationships between the PGAs at the surface and downhole were evaluated to understand how amplification varies as PGA increases. Strong ground motions with and without site effects throughout Taiwan could be observed by comparing intensity distribution maps generated using the surface and downhole accelerations from four large earthquakes with magnitudes > 6. Empirical transfer functions derived using the single-station and two-station methods at the same stations showed comparable dominant frequencies and amplification factors; however, the empirical transfer function derived using the two-station method showed clearer resonance peaks, not only at fundamental frequencies but also at higher mode resonance frequencies. The HHSR and the HVSR were highly similar, particularly at medium frequencies. This finding indicates that the HVSR can be used instead of the HHSR when only the surface recording is available. Moreover, the local magnitudes calculated using surface recordings were higher than those calculated using downhole recordings. The differences are attributed to the amplification caused by the sedimentary layers and resultant in 0.36, 0.46 and 0.49 on average for events with ML of > 6, 5–6, and 4–5. Furthermore, HHSRs at 5–10 Hz and 1.1–1.7 Hz were strongly correlated with PGA amplifications and ML differences, respectively.

The damping ratio is an important index used in soil nonlinearity studies and is mostly proportional to the shear strain increase. Previous researches indicated a frequency-independent damping in most cases. In this study, frequency-dependent damping was introduced from frequency-dependent Q calculated through the spectral ratio method of near-surface structures using the power spectrum of strong motion records in the Strong Motion Array in Taiwan Phase I (SMART1). The dense SMART1 recorded significant strong motions in the 1980s, which can be used to identify soil nonlinearity at near surfaces. A 40%–50% increase in frequency-dependent damping for SMART1 was identified, with strain increasing from 0.01% to 0.1% in near-surface regions. A large damping was also found in the shallow sediments with mean Vs below 600 m/s on the topmost 500 m layers in the SMART1 database at a frequency range of 3–8 Hz, which is independent of the magnitude scaling or near-field travel distance scaling relations.

This study extracted some P-wave features from the first few seconds of vertical ground acceleration of a single station. These features include the predominant period, peak acceleration amplitude, peak velocity amplitude, peak displacement amplitude, cumulative absolute velocity and integral of the squared velocity. The support vector regression was employed to establish a regression model which can predict the peak ground acceleration according to these features. Some representative earthquake records of the Taiwan Strong Motion Instrumentation Program from 1992 to 2006 were used to train and validate the support vector regression model. Then the constructed model was tested using the whole earthquake records of the same period as well as the 2010 Kaohsiung earthquake with 6.4 M L . The effects on the performance of the regression models using different P-wave features and different length of time window to extract these features are studied. The results illustrated that, if the first 3 s of the vertical ground acceleration was used, the standard deviation of the predicted peak ground acceleration error of the whole tested 15-years earthquake records is 20.89 gal.The length of time window could be shortened, e.g. 1 s, and the prediction error is slightly sacrificed, in order to prolong the lead-time before destructive S-waves reaches.

The shear wave velocity of shallow sediments is very important in seismic wave amplification and thus Vs 30 is a well-known parameter for site classification. Based on the relationship between Vs and soil indexes, the empirical regression equations were evaluated using more than 600 data sets from the Ilan area and the Taipei Basin. Multivariable analysis, which can increase the accuracy of regression equations, was used in this study. Three extrapolations were compared, and then the most accurate bottom-constant extrapolation was adopted to estimate Vs 30 at 16 boreholes, none of which reached a depth of 30 m. Ultimately Vs 30 was derived for 110 free-field strong motion stations, and the stations in northeastern Taiwan were reclassified according to the Vs 30-based NEHRP provisions. Regression equations of Vs and the extrapolation of Vs 30 were also applied to boreholes with only N -values and corresponding depths of less than 30 m for assessing the Vs 30 and NEHRP class.