利用紫坪铺水库2004~2007年记录的实际地震数据，通过比较多种滤波方法，发展出一种既能有效滤除地震数据中本底噪音、又对有效信号损伤较小的比值滤波方法。对去噪后的水库诱发地震和构造地震数据，利用小波变换方法，分别进行时频分析获得两类地震的时频谱，并提取可以反映地震发生前后能量分布和聚集情况的时频属性，最后根据各个属性的响应效果，组合得到新的属性。将时频谱与时频属性相结合，总结归纳紫坪铺水库诱发地震与构造地震波谱时频特征的差异性，为地震监测分析提供基础依据。

On the basis of the real seismic data of Zipingpu reservoir from 2004 to 2007,we compared several filtering methods and choosed the ratio filtering method which can effectively filter the background noise in the seismic data without damaging the effective signal in the data. Then,wavelet transform,which has a high time-frequency resolution,was used to the denoising data of the reservoir induced earthquake and tectonic earthquake to get two kinds of seismic spectrum. After that,we extracted the time-frequency attributes from the seismic data and combined the attributes to get a new time-frequency attribute. By analyzing the time-frequency spectrum and attributes,we found the differences in time-frequency characteristics between Zipingpu reservoir induced earthquake and tectonic earthquake,which is capable of providing a basis for earthquake monitoring and analyzing.

P315

中国地震局“三结合”课题（3JH-20203703）资助