基于强震动数据的龙门山地区地震动预测模型参数反演

基于2008—2015年龙门山地区的强震动记录，采用广义线性反演法计算了龙门山地区的震源参数、品质因子以及83个强震动台站的局部场地放大系数。结果显示:芦山地震发生之前，汶川地震余震的应力降随时间的增加而逐渐增大；芦山地震发生之后，龙门山断裂带上余震的应力降明显减小，并在之后随时间呈缓慢增大的趋势；部分M_S≥4.7余震事件的加速度震源谱中出现明显的高频衰减现象。通过对应力降随时间变化规律的分析认为，这种高频衰减现象可能与震级大小和震源破裂滑动的最大速率有关。此外，反演得到了山区和盆地地区的品质因子分别为Q（ f ）＝264 f 0.75和Q（ f ）＝223 f 1.01；同时获取了龙门山地区83个强震动台站的局部场地放大系数，并计算了基于NGA场地分类标准的A类、 B类和C类场地的平均场地放大系数。结果显示，以上3类场地的平均场地放大系数具有相似的峰值，但是随着场地趋于软弱，峰值平台区向低频部分发育。 

Parameter estimation of ground-motion prediction model in Longmenshan region based on strong motion data

Based on generalized linear inversion scheme, strong motion data recorded from 2008 to 2015 is used to estimate source parameters, quality factor and local site amplification coefficient of 83 strong motion observation stations of Longmenshan region. The result shows that the stress drops of Wenchuan aftershocks tended to increase with time before occurrence of Lushan main shock, after which the stress drops of aftershocks decreased noticeably and then increased slowly with time. Furthermore, a significant high-frequency attenuation phenomenon was observed in source spectra of some aftershocks with M_S≥4.7. According to analyses on the temporal evolution tendency of stress drop, it is inferred that the phenomenon may be related with magnitude and the maximal source rupture slip rate of individual event. As a result, the quality factor of mountain and basins inverted in this study is Q（ f ）＝264 f 0.75 and Q（ f ）＝223 f 1.01, respectively. Finally, the obtained site amplification coefficients of 83 strong motion observation stations are classified into NGA site class A, B and C, and the average site amplification coefficients of the three site classes are calculated. The results indicate that the peak amplification coefficients of the three site classes are similar, and the peak platforms tend to develop to lower frequencies as the site becomes softer.