Seismology Based Ground Motion Attenuation Relationship For Tohoku Area

Since ground motion attenuation relationship plays an important role inengineering seismology, it has been a hot topic for decades. Researcherspublished hundreds of research documents in the field. Most of them areempirical. Some key factors are s summarized from a lot of strong motionrecords, which is described as simple mathematical models, and thendifferent statistical methods are used to obtain corresponding regressioncoefficients. It is shown that this kind of widely used methods depends onstrong ground motion records. So, for most countries and regions in theworld, without enough strong motion records, these empirical methods arenot appropriate.Seismic data observed by a broadband digital seismic network is used inthis paper. By seismological methods, ground motion attenuation relationshipis constructed. It is going to be solved that strong motion data is notenough for empirical regression in most countries and regions.To validate the feasibility, northeast area of Japan （E138°-143°, N36°-40°） is selected as the target area, since there are enough data from smallearthquakes and moderate earthquakes. Total of850records from632smallevents （Mw=3.5⁴.5, focal depth≤30km） recorded by12stations of F-net,and4323records from257events （M_w≥4.5, focal depth≤30km） recorded by89bedrock stations of K-NET. The former is used to estimate the regionalsource and crustal media parameters, the latter is used for comparisons andthe demonstration of the feasibility.The envelopes of Fourier spectrum from the F-net data are used as theobjective curves. From seismological studies, five key inverse parameters,Δσ, Q_o, η, R₁and R₂, and the ranges are determined, and the optimalsolutions are obtained by micro-Genetic Algorithm （μGA）.These inverse parameters are taken into the simplified ground motionroot-mean-square calculation model and the non-simplified one, then thesimplified and non-simplified peak ground acceleration attenuationrelationships are constructed respectively. Compared with K-NET recordsand some empirical attenuation relationships, the PGA attenuationrelationships from this paper reveal an agreement with strong ground motiondata than empirical relations. When there is a lot of strong earthquakedata,the results of this paper is slightly worse than empirical relations;Whenthere is not so much data,the results are better than empirical relations.Thefeasibility of the method is demonstrated, which is going to be thefoundation of further application in practice.Finally, some suggestions are given for future work, which areencountered in this work.