Abstract
The consideration of soil nonlinearity is important for the accurate estimation of the site response. To evaluate the soil nonlinearity during the 2008 Ms8.0 Wenchuan Earthquake, 33 strong-motion records obtained from the main shock and 890 records from 157 aftershocks were collected for this study. The horizontal-to-vertical spectral ratio (HVSR) method was used to calculate five parameters: the ratio of predominant frequency (RFp), degree of nonlinearity (DNL), absolute degree of nonlinearity (ADNL), frequency of nonlinearity (fNL), and percentage of nonlinearity (PNL). The purpose of this study was to evaluate the soil nonlinearity level of 33 strong-motion stations and to investigate the characteristics, performance, and effective usage of these five parameters. Their correlations with the peak ground acceleration (PGA), peak ground velocity (PGV), average uppermost 30-m shear-wave velocity (VS30), and maximum amplitude of HVSR (Amax) were investigated. The results showed that all five parameters correlate well with PGA and PGV. The DNL, ADNL, and PNL also show a good correlation with Amax, which means that the degree of soil nonlinearity not only depends on the ground-motion amplitude (e.g., PGA and PGV) but also on the site condition. The fNL correlates with PGA and PGV but shows no correlation with either Amax or VS30, implying that the frequency width affected by the soil nonlinearity predominantly depends on the ground-motion amplitude rather than the site condition. At 16 of the 33 stations analyzed in this study, the site response showed evident (i.e., strong and medium) nonlinearity during the main shock of the Wenchuan Earthquake, where the ground-motion level was almost beyond the threshold of PGA > 200 cm/s2 or PGV > 15 cm/s. The site response showed weak and no nonlinearity at the other 14 and 3 stations. These results also confirm that RFp, DNL, ADNL, and PNL are effective in identifying the soil nonlinearity behavior. The identification results vary for different parameters because each parameter has individual features. The performance of the PNL was better than that of DNL and ADNL in this case study. The thresholds of ADNL and PNL are proposed to be 2.0 and 7%, respectively.Graphical .
Highlights
It is well known that seismic waves can be amplified by surface soil layers that have a strong impedance contrast with deep bedrock
The objective of this study was to evaluate the level of soil nonlinearity of 33 strong-motion stations using five parameters and to identify the peak ground acceleration (PGA) and peak ground velocity (PGV) thresholds beyond which the site response behaved nonlinearly during the Wenchuan main shock
“–” means that site-predominant frequency (Fp) and frequency of nonlinearity (fNL) cannot be identified a The average uppermost 30-m shear-wave velocity (VS30) data were taken from the Next Generation Attenuation (NGA)-West2 database b Stations 51MXT and 62WIX are located at rock sites, and the other stations are located at alluvial sites based on station construction reports a b
Summary
It is well known that seismic waves can be amplified by surface soil layers that have a strong impedance contrast with deep bedrock. The objective of this study was to evaluate the level of soil nonlinearity of 33 strong-motion stations using five parameters and to identify the PGA and peak ground velocity (PGV) thresholds beyond which the site response behaved nonlinearly during the Wenchuan main shock These parameters include DNL, frequency of nonlinearity (fNL), and percentage of nonlinearity (PNL), defined by Régnier et al (2013), and the ratio of the predominant frequency (RFp) and absolute degree of nonlinearity (ADNL), as defined in this study. Evidence of soil nonlinearity during the Wenchuan main shock The values of RFp, DNL, ADNL, fNL, and PNL for each site were calculated (Table 1) Based on these parameters, sites showing nonlinear behavior were identified; the correlations between these parameters and PGA, PGV, VS30, and Amax were investigated. The empirical relationship between PNL and Amax was regressed (Fig. 9d), indicating that PNL has a moderate positive correlation with Amax, similar to DNL and ADNL
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
