Research On Key Parameters Of Stochastic Simulation Method Of Near-field And High-frequency Ground Motions

China is a country with frequent earthquakes,large earthquake intensity and severe disasters.With the development of national economy and science and technology,the major engineering projects are also increasing,which is inevitable to conduct prior analysis on the intensity zone where the projects are located and the seismic performance of the structures.Due to the uneven distribution of seismic network,it is difficult to obtain sufficient records in many regions.In addition,the observation records have spatial distribution inhomogeneity,non-repeatability and dependence on the sites.The method of establishing a simulated ground motion model in combination with the actual engineering environment is important to the seismic analysis of the current structure.In engineering practice,the natural period of the structures is small,so the engineering is more interested in the high-frequency range of 1 to 25 Hz.In this paper,two stochastic methods for high-frequency ground motions simulation are introduced,and a stochastic finite-fault method that better reflects the source characteristics is discussed.The expression of the corner frequency is derived,and the high-frequency energy scaling factor based on the corner frequency derivation is improved.Based on the records of the main aftershocks of the Kumamoto earthquake acquired by K-NET and KiK-net,the relationships between the high-frequency delay parameter kappa and the source,path and site are statistically analyzed.The main work and conclusions are listed as follows:(1)The origin and development process of stochastic point-source method and stochastic finite-fault method are systematically summarized,as well as some current research status at home and abroad.The method principle,practical scope and parameter meaning of stochastic point-source method and stochastic finite-fault method are introduced in detail.The advantages,disadvantages and similarities and differences of stochastic point-source method and stochastic finite fault method are expounded.The general expression of the corner frequency is derived.The various assumptions in the derivation process are introduced,and the relationship between the rupture velocity and the corner frequency is given.It is recommended that when the technology and theory are sufficient to support obtaining the function of the rupture velocity over time,the corner frequency of each subfault can be calculated by the rupture velocity,seismic moment and stress drop.Thus,while improving the simulation accuracy can also reflect the non-uniformity of the fault rupture.(2)This paper introduces the high-frequency energy scaling factor of stochastic finite-fault method and another proposed by Sun Xiaodan et al.Based on the corner frequency of Sun Xiaodan et al.,this paper proposes a new expression of high-frequency energy scaling factor.According to the subfault sizes,it is divided into four groups of 1km,2km,5km and 10 km.By assuming a special case earthquake,thefar-field radiant energies of the improved model and the original model are compared.Under the uniform slip,the differences between the energy levels obtained by the original model and the improved model is not significant,and both models show ideal results.Under the random slip,the energy levels by the original model are significantly different in different size subfaults,and increase as the number of subfaults increases in the high-frequency band,while the improved model there are also differences but not obvious.In addition,under the different subfault sizes,the Kumamoto 7.1earthquake was selected as an example to compare the effects of subfault size division on the simulated PSA in the two models.The results show that the four PSA curves obtained by the improved model are less different than original model,and are closer to the actual recorded PSA.(3)Under the same site conditions,statistical analysis is performed on kappa which characterizes the attenuation of the amplitude spectrum of the high-frequency band in the stochastic simulation method.Firstly,the relationships between the kappa ratio of east-west direction,north-south direction to the vertical direction and the moment magnitude,epicentral distance are calculated,and the empirical relationships are given.The results show that the kappa ratio is moderately correlated with and strongly correlated with the epicentral distance.With the increase of and the epicentral distance,the difference between the horizontal two-component kappa and the vertical kappa decreases.Then the relationships between kappa and,epicentral distance are calculated,and the empirical relations are given.The results are as follows: there is a certain correlation between kappa and or epicentral distance;the kappa distribution in the east-west direction and the north-south direction are very close,and both increase with the increase of or the epicentral distance;the kappa distribution is relatively discrete,especially the moment When the magnitude is less than 5,the kappa distributions shows great randomness,but the overall concentration is between 0.06 s and 0.08s;the kappa distribution is relatively discrete,but kappa is almost always concentrated in the range of twice the standard deviation.Appropriate adjustments can be made within this range in practical applications.(4)Using the soil information provided by NIED,the of 81 stations with an epicenter distance of 250 km was calculated.The relationship between kappa and in the from 4.0 to 7.1 is calculated,and the empirical relationships are given.Kappa has a strong correlation with.The harder the near-surface geological structure,the less the attenuation caused by the internal friction of the medium,and the smaller the kappa.Since the more restrictive conditions in the source,path,and sites during statistics,the fewer kappa sample points that can be used for statistics,combined with the presence of random factors,will have a large impact on the statistical results.This paper suggests that nonlinear fitting or multivariate fitting can be considered infuture statistics.The results of this study can be used in the stochastic simulation method in conjunction with the amplifications of the sites to improve the accuracy of the simulation,and to provide kappa estimates for regions with only drill records.