A random synthesis procedure based on finite fault model is adopted for near field strong ground motion simulation in this paper. The fault plane of the source is divided into a number of sub-sources, the whole moment...A random synthesis procedure based on finite fault model is adopted for near field strong ground motion simulation in this paper. The fault plane of the source is divided into a number of sub-sources, the whole moment magnitude is also divided into more sub-events. The Fourier spectrum of ground motion caused by a sub-event in given sub-source, then can be derived by means of taking the point source spectrum, attenuation with distance, energy dissipation, and near surface effect, into account. A time history is synthesized from this amplitude spectrum and a random phase spectrum, and being combined with an envelope function. The ground motion is worked out by superposition of all time histories from each sub-event in each sub-source, with time lags determining by the differences between the triggering times of sub-events and distances of the sub-sources. From the example of simulations at 21 near field points in a scenario earthquake with 4 dip angles of the fault plane, it is illustrated that the procedure can describe the rupture directivity and hanging wall effect very well. To validate the procedure, the response spectra and time histories recorded at three near fault stations MCN, LV3 and PCD during the Northridge earthquake in 1994, are compared with the simulated ones.展开更多
Suppose that the time series Xt satisfieswhere α0≥δ>0,αi≥0 for i=1,2,…,q;βi,i=1,…,p, are real numbers; p and q are the order of the model. The sequence {ξt};(0,1) and is independent of {hs,s≤t} for fixed ...Suppose that the time series Xt satisfieswhere α0≥δ>0,αi≥0 for i=1,2,…,q;βi,i=1,…,p, are real numbers; p and q are the order of the model. The sequence {ξt};(0,1) and is independent of {hs,s≤t} for fixed t. The above model is usually written as AR(p)-ARCH(q).We consider stationary series AR(p)-ARCH(q) model and assume the stationary field is θ0. We express this statement asH1:α1≥α2…≥αq,β1≥β2≥…≥βp and we consider an order restricted testing problem, which is to testH0:α1=α2=…=αq,β1=β2=…=βpagainst H1-H0. We derive the likelihood ratio (LR) test statistic and its asymptotic distri-展开更多
基金Earthquake Science Foundation under Contract No.201009
文摘A random synthesis procedure based on finite fault model is adopted for near field strong ground motion simulation in this paper. The fault plane of the source is divided into a number of sub-sources, the whole moment magnitude is also divided into more sub-events. The Fourier spectrum of ground motion caused by a sub-event in given sub-source, then can be derived by means of taking the point source spectrum, attenuation with distance, energy dissipation, and near surface effect, into account. A time history is synthesized from this amplitude spectrum and a random phase spectrum, and being combined with an envelope function. The ground motion is worked out by superposition of all time histories from each sub-event in each sub-source, with time lags determining by the differences between the triggering times of sub-events and distances of the sub-sources. From the example of simulations at 21 near field points in a scenario earthquake with 4 dip angles of the fault plane, it is illustrated that the procedure can describe the rupture directivity and hanging wall effect very well. To validate the procedure, the response spectra and time histories recorded at three near fault stations MCN, LV3 and PCD during the Northridge earthquake in 1994, are compared with the simulated ones.
文摘Suppose that the time series Xt satisfieswhere α0≥δ>0,αi≥0 for i=1,2,…,q;βi,i=1,…,p, are real numbers; p and q are the order of the model. The sequence {ξt};(0,1) and is independent of {hs,s≤t} for fixed t. The above model is usually written as AR(p)-ARCH(q).We consider stationary series AR(p)-ARCH(q) model and assume the stationary field is θ0. We express this statement asH1:α1≥α2…≥αq,β1≥β2≥…≥βp and we consider an order restricted testing problem, which is to testH0:α1=α2=…=αq,β1=β2=…=βpagainst H1-H0. We derive the likelihood ratio (LR) test statistic and its asymptotic distri-