Spatial Correlation Of Ground Motion In The Kanto Region,Japan

Peak ground acceleration(PGA),peak ground velocity(PGV)and spectral acceleration(Sa)at two locations caused by the same seismic event are related to each other.At present,there are some spatial correlation models of ground motion intensity,such as peak ground acceleration and spectral acceleration.Although the forms of the models are similar,the decay rate of the correlation varies significantly with the increase of the separation distance between the sites.The main reasons for this difference are the selected database,the ground motion model used to derive the spatial correlation model,the estimation method and the regional geological conditions.In this paper,the observation data of strong vibration recorded in Kanto region of Japan were used to carry out the research,and the main research work completed is as follows:1.Based on 5760 sets of strong seismic data collected from 61 moderately strong earthquakes(M_W 5?7)at 87 K-NET stations and 29 KIK-net stations in the Kanto region of Japan,the spatial correlation study was conducted.Based on the kanto region of strong motion database established ground motion prediction equation,the acceleration response spectrum is analyzed under the zero distance between the events,events and total residual correlation,the results showed that between events,events,and the correlation of total residual are similar,but residual between events than in the event and the total residual have stronger correlation.2.Based on the correlation research results of inter-event residuals,the fitting model of inter-event residuals correlation is given.The spatial correlation coefficients of the ground motion parameters of different periods are obtained by using the semi-variable function,and the exponential attenuation model of the ground motion parameters with the separation distance between stations is given.Based on the prediction model of inter-event correlation and the spatial correlation model of response spectra of different cycles,the spatial correlation models of response spectra of different cycles and separation distances between different stations are given.3.By comparing the spatial correlation models within events in the Kanto region with those in different regions,it is found that the decay rate of spatial correlation is significantly different with the increase of the separation distance between stations.The spatial correlation models of the same region are basically similar.The established correlation models are applied to the conditional mean spectrum and the simulated random seismic field.