摘要：The site amplification factor was usually considered to bescalar values, such as amplification of peak ground acceleration or peakground velocity, or increments of seismic intensity in the earthquake earlywarning (EEW) system or seismic-intensity repaid report system. This paper focuseson evaluating an infinite impulse recursive filter method that could producefrequency-dependent site amplification and compare the performance of thescalar-value method with the infinite impulse recursive filter method. Alarge number of strong motion data of IBRH10 and IBRH19 of the Kiban Kyoshinnetwork (KiK-net) triggered in more than 1000 earthquakes from 2004to 2012 were selected carefully and used to obtain the relative siteamplification ratio; we model the relative site amplification factor with acasual filter. Then we make a simulation from the borehole to the surface and alsosimulate from the front-detection station to the far-field station. Compared todifferent simulation cases, it can easily be found that this method couldproduce different amplification factors for different earthquakes and couldreflect the frequency-dependent nature of site amplification. Through thesesimulations between two stations, we can find that the frequency-dependentcorrection for site amplification shows better performance than the amplification factor relative to velocity (ARV)method and station correction method. It also shows better performance thanthe average level and the highest level of the Japan Meteorological Agency (JMA)earthquake early warning system in ground motion prediction. Some cases in which simulationdid not work very well were also found; possible reasons andproblems were analyzed and addressed. This method pays attention to theamplitude and ignores the phase characteristics; this problem may be improvedby the seismic-interferometry method. Frequency-dependent correction forsite amplification in the time domain highly improves the accuracy ofpredicting ground motion in real time.
