| The site effect of ground motion has always been a key problem to be considered in the seismic fortification of engineering industry.The nonlinear response of soil under strong motion has also become a research hotspot in the field of seismic engineering.At present,there are relatively few studies on site nonlinearity considering time-varying characteristics.Therefore,based on the strong motion data recorded by Ki K-net network in Japan,this paper studies the site nonlinear time-varying characteristics by using deconvolution seismic interferometry,spectral ratio method and autocorrelation function method,and introducing moving window technology.In this paper,the above methods are applied to the site nonlinear analysis of the 2011 Northeast Earthquake and the 2021Fukushima Earthquake in Japan,and the three nonlinear time-varying identification methods are compared and analyzed.In addition,the characteristics of seismic amplitude on the surface and underground are analyzed,and the amplification effect of the site is discussed.The main results are as follows:1.Based on the traditional deconvolution seismic interferometry,the nonlinear time-varying characteristics of different types of sites are discussed by using the records of main shock,foreshock and aftershock recorded during the 2011 Northeast Earthquake in Japan.It can be seen that the shear wave velocity will decrease significantly when a strong earthquake（main shock）occurs.After a strong earthquake,the shear wave velocity of the site can be observed to recover gradually by interferometric aftershock（weak earthquake）.The recovery of shear wave velocity varies with the site type,and the judgment of nonlinear threshold is not intuitive enough.The nonlinear time-varying analysis of 7 Ki K-net stations in the 2011 Northeast Japan Earthquake（Mw9.0）by deconvolution seismic interferometry with moving window technology shows that this method can better identify the threshold of nonlinearity,the change degree of nonlinearity and the recovery degree after strong vibration.Comparing the moving window deconvolution method with the moving window spectral ratio method,the nonlinear threshold identified by the two methods is close,about 40～100 cm/s2,and has no obvious correlation with the site VS30.At IBRH20 station with lower PGA,the wave velocity drop caused by nonlinearity is small（3%）and almost completely recovered after the earthquake.For the other six stations with PGA between 386～822 cm/s2,the site equivalent shear wave velocity decreased by 13～37%,resulting in significant site nonlinearity and incomplete recovery after the earthquake.Using the coda spectrum ratio,the site properties before and after the strong earthquake are analyzed,and the recovery of site nonlinearity is discussed.It can be seen that after the start of S wave of the main earthquake,the peak frequency decreases from the level before the earthquake to a lower level,and the spectrum ratio at the peak frequency also decreases.From 0 to 10 days after the main earthquake,the peak frequency gradually recovered.Three months to one year after the main earthquake,the peak frequency further recovered,but it is still less than the foreshock,indicating that the site property has not fully recovered after a long period of strong earthquake.2.For the observation records of 6 Ki K-net stations in the 2021 Fukushima Earthquake(MJMA 7.3),the research on the site nonlinear time-varying process based on moving window deconvolution method,moving window spectral ratio method and moving window autocorrelation function method shows that the stability of the three methods to identify the site nonlinear time-varying characteristics is different.Moving window deconvolution method is easier to obtain a more stable nonlinear time-varying process of soil,but it reflects the change of average wave velocity from surface to underground,and the degree of nonlinearity is low.For the soil with strong wave impedance ratio in the shallow surface layer,the moving window spectral ratio method and moving window autocorrelation function method can obtain a stronger degree of nonlinearity.For the soil without strong wave impedance ratio in the shallow surface layer,it is not easy to identify the stable nonlinear time-varying process.The strongest nonlinear moments identified by the three methods are basically the same,which are located near the peak acceleration of the whole record.The nonlinear threshold of 6stations identified based on the three methods is about 40～140cm/s2,the decline ratio of shear wave velocity is between 16～53%,and the site nonlinearity is strong.3.Using the strong earthquake data recorded by 106 Ki K-net stations,the ground motion ratio recorded on the surface and underground is calculated,and the amplification of the site is studied.From the results,with the increase of the period,the slope of the ground motion ratio between the surface and underground gradually decreases,and the estimated value of the error variance obtained after fitting gradually decreases with the increase of the period,indicating that the data discreteness decreases,and the ground motion ratio recorded on the surface and underground decreases with the increase of VS30.For sites with small average equivalent shear wave velocity VS,the amplification effect is lower than that of sites with high average equivalent shear wave velocity in a short period,and there is little change in a long period.In a short period,the ground motion ratio recorded on the surface and underground decreases with the increase of logging depth.In the long-period part,the ground motion ratio recorded on the surface and underground is less affected by the change of average equivalent shear wave velocity VS.The period length has obvious influence on the amplification effect of stations with drilling depth less than 500m,while it has little influence on stations with drilling depth more than 500m. |
