The site effect has been a polular topic in the field of earthquake engineering for a long time. Much effort has been made in studis on site effect by means of numerical simulation and theoretical computation. However, some methods based on strong-motion recordings are recently recognized and widely used, due to their simplicity and no need of many assumptions any more. In the 2008 Wenchuan earthquake, the China National Strong Motion Observation Network System (NSMONS) captured a large number of strong-motion recordings at free field both in the main shock and aftershocks. This makes a good chance for studying the site effect in the Wenchuan earthquake. In this paper, based on a database of these recordings, it takes a systematical study on the site classification, generalized inversion of site response and identification of nonlinearity of site response. That manages to extend the application of strong-motion recordings in the field of scientific research and engineering practice, and promote the development of strong-motion observation in China. The contents are summarized as follows:
(1) The merits and drawbacks of three common-used HVSR (Horizontal-to- Vertical Spectral Ratio) methods of site classification are summarized. Then an improved HVSR method using entropy weight theory is proposed. The site classes of 54 permanent stations and 66 temporary stations in the Wenchuan earthquake are classified by using these four methods. The comparisons between the results given by each method shows that the new method has a more scientific rationality.
(2) An updated classification criterion for six site classes is proposed and used in the process of site classification of temporary strong-motion stations in the Wenchuan earthquake. The results comfirms that it can raise the classification accuracy and effectively solve the problem that HVSR method can not be used for the sites where HVSR curves appear multiple peaks or no any peak. Finally, a preliminary analysis of the correlation between station elevation and its natural period TG identified by HVSR indicates that TG is generally smaller at higher elevations in Sichuan area.
(3) A total of 602 strong-motion recordings obtained by 28 stations from 96 aftershocks of the Wenchuan earthquake are selected as a dataset. The source, path and site effect of these recordings are separated by the generalized inversion technique (GIT). The inversion results have been verified as reliable by comparing the site response at station 62WUD using the GIT method and the the standard spectral ratio (SSR) method. For all 28 stations, the site predominant frequency Fp and the average site amplification in different frequency bands of 1.0-5.0 Hz, 5.0-10.0 Hz and 1.0-10.0 Hz are calculated based on the inversion results. Compared with the results from the HVSR method, it shows that the HVSR method can reasonably estimate the site predominant frequency but underestimates the site amplification. The linear fitting between the average site amplification for each frequency band and the Vs20 (the equivalent uppermost-20 m shear wave velocity) shows a good correlation.
(4) A distance measurement called the asperity distance DAsp is proposed to reasonably characterize the source-to-site distance for large earthquakes such as the Wenchuan earthquake, and is verified to be much better than the other three kinds of distance measurements, including the rupture distance (DRup), fault distance (DFau) and hypocentral distance (DHyp). Then it’s used for generalized inversion of site responses in the main shock of Wenchuan earthquake.
(5) The S transform is suggested to be used for identifying the nonlinearity of site seismic response. Combined with GIT method and HVSR method, it is practiced to identify the site responses of 33 strong-motion stations behave linearity or nonlinearity in the main shock of Wenchuan earthquake. The result shows that the site response appears obvious nonlinearity when PGA larger than 300 cm/s2 or PGV larger than 20 cm/s. In this regard, the sites of station 051AXT, 051GYS, 051GYZ, 051JYC, 051JYD, 051JYH, 051JZW, 051MXN, 051SFB and 062WUD also behave significant nonlinear responses. Compared with the results given by GIT method and HVSR method in the frequency domian, it is verified that the S transform can effectively identify the nonlinerity behavior of site seismic response in the time domain, and have an ablitity to determine the particular time when the nonlinearity occurs.
(6) The correlations between the ratio of Fp under weak motions and strong motion (RFp) and ground motion levels such as PGA and PGV, and site coefficients such as Vs20 and Vs30 are analyzed respectively. The results show RFp has a strongly positive correlation with PGA and PGV, but is uncorrelated with Vs20 and Vs30. A new definition called ADNL (absolute degree of nonlinearity) is proposed to characrize the degree of nonlinearity of site seismic response. Then the ADNL of each site and other parameters such as fNL (frequency of nonlinerity), DNL (degree of nonlinearity) and PNL (percentage of nonlinearity) are calculated. The empirical regression analysis between these parameters and PGA represents that each one has a strong correlation with PGA.