| The spatial variation of ground motion refers to the difference in amplitude and phase of seismic records at different sites[1].It is a very common method to simulate the spatial variation of ground motions through a coherent function model.This article focuses on the following two issues:(1)Generally,the ground motion coherence function model given by the regression is obtained by recording the available stations.The parameter d is not considered in segments.It is necessary to analyze whether the obtained model can reasonably reflect the influence of the distance d.(2)Different researchers use different models.There are also great differences in the coherence function curves obtained from the same seismic records of the same array.This makes it difficult for engineers to determine which model is reasonable in practice.This article focuses on the above issues and does the following work.1.Comparising the influence of different pairs of station pairs on the fitting parameters of the coherence function model.Taking the NO.45 and 5 earthquake’s horizontal component acceleration as the analysis object,the author calculates the value of lagged coherency when d is 200m,1000m,and 2000m,all the calculated values of the three distances were fitted and the first set of fitting parameters was obtained.Similarly,the calculated values of all six distances were fitted and the second set of fitting parameters was obtained.The results show that the dispersion of the second set of fitting parameters is superior to the first set of fitting parameters and the coherence coefficient curve calculated based on the second set of fitting parameters is more reasonable.Therefore,when statistical methods are used to obtain coherent functions,the strong motion records at different stations should be considered as much as possible.2.Based on the new fitting parameters of Abrahamson、Loh、Hao and Harichandran,which is obtained in above,the coherence of each distance d between stations was calculated.Comparing the coherence which calculated by different way,the Evaluation index,SSE,MSE,RMSE,and R-square of the four models are obtained,and it is found that in different stations for the distance d range,the accuracy of each model is different.At the last,optimal model under different conditions of distance d is obtained,which can provide reference for engineering application. |