عنوان مقاله [English]
نویسندگان [English]چکیده [English]
Earthquake ground-motions show significant variability in both spectral and temporal characteristics. Procedures to generalize and predict strong ground motions may be generally divided into three main disciplines: numerical techniques based on a kinematic source description, empirical attenuation models, and semi-empirical stochastic models. Numerous studies have shown that, in the near-fault regions, i.e. at distances comparable with few fault lengths, the ground motion from moderate to large earthquakes is strongly affected by the evolution of the rupture along the fault plane, causing more complex spatial distribution of the observed values. Also, since the number of near-field records of large earthquakes is usually inadequate, empirical models even those developed specifically for the near-field show a large uncertainty which cannot be improved until sufficient data become available. Although after recent events e.g. the 2003 L’Aquila, Italy, and 2011 Christchurch, New Zealand earthquakes, we have been provided with some additional records in the near-fault region but fifty years of strong-motion records worldwide is not sufficient to cover the whole range of site and propagation path conditions, rupture processes and geometric relationships between source and site that are possible from earthquakes in the near source regions. As an alternative to the use of records from past earthquakes computational geophysical techniques, based on a kinematic source description, can be used to simulate physically based synthetic seismograms for engineering applications. Near-fault ground motions show high spatial heterogeneity due to rupture complexity, fault to site orientation, and also seismic wave propagation and local site effects. Since the number of near-field records of large earthquakes is usually inadequate, empirical ground motion prediction models even those developed specifically for the near-field region show a large uncertainty which cannot be improved until sufficient data become available. For the time being, the use of physically based synthetic ground motions obtained by kinematic simulation approaches may partially overcome the scarcity of near-source data. Here, a discrete wave number/finite element technique is used to compute velocity time series in the low-frequency band (up to 1.5 Hz) and to investigate the variability of the ground motion as a function of different source characteristics and source-to-site geometry. The approach is well suited to study the propagation of seismic waves in a horizontal layered medium. Ground motions from 4 earthquakes with moment magnitudes from 6.0 to 7.5 were simulated, in 0.5 magnitude unit increments, at 3 values of fault distances i.e., 5, 10 and 15 km. For studying the effects of these velocity pulses on dynamic response of structures, some simple models such as rectangular, triangle and sinus have been presented and used in the literature. But, recent studies show that using these kinds of methods in study of dynamic behavior of structure may lead to incorrect conclusions. Mavroeidis and Papageorgiou (2003, hereafter MP2003) using a dataset of near fault records, have presented a new mathematical equation for velocity pulse form that because of its simplicity and precision has been widely accepted .The simulation results of the current study have been used in the context of mathematical modeling proposed by MP2003 to perform a robust parameterization of the model.