AI 中文总结
本研究将随机噪声添加技术应用于匹配滤波前，结合波形互相关方法，大幅提升了2025年7月29日堪察加巨型逆冲地震前小震级地震活动的检测与复原能力。
AI 中文摘要
2025年7月29日发生的堪察加巨型逆冲地震是21世纪最大的地震事件之一，也是仪器记录最完善的案例，拥有大量区域及远震台站，包括国际监测系统（IMS）的高性能地震台阵。应用于地震台阵的波形互相关（WCC）方法结合震相关联算法的改进，大幅提升了信号检测能力。由于WCC检测器是匹配滤波方法的核心算法，其效率高度依赖环境地震噪声与模板信号的相干性，完全随机噪声可提供最优检测条件。为压制相干噪声分量并提升WCC检测性能，本研究添加了经振幅调整至自然噪声水平的数值生成随机噪声序列，该方法与诸多声学应用中的噪声白化类似，但本研究将其应用于匹配滤波之前。检测阈值增益可比纯WCC检测器获得的结果高出数倍。通过对IMS台阵及三分量台站应用添加随机噪声后的WCC方法，2025年7月29日堪察加地震前的小震级地震活动复原效果得到显著提升。
英文摘要
The July 29, 2025, Kamchatka megathrust earthquake is one of the largest events in the 21st century. It is likely the best-instrumented case, with numerous stations at regional and teleseismic distances, including high-performance seismic arrays of the International Monitoring System (IMS). The waveform cross-correlation (WCC) method applied to seismic arrays allowed a significant improvement in signal detection accompanied by enhancements in phase association algorithms. Since the WCC detector is the core algorithm of the matched filter method, its efficiency critically depends on the ambient seismic noise coherency with the template signal, with fully stochastic noise providing optimal detection conditions. To suppress the coherent noise component and improve WCC detection, we add a numerically generated random noise sequence adjusted in amplitude to the natural noise. This method is similar to noise whitening used in many acoustic applications, but in this study, it is applied before the matched filter is used. The detection threshold gain can be several times higher than that obtained with a pure WCC detector. The recovery of the low-magnitude seismicity obtained by the WCC applied to the IMS arrays and three-component stations prior to the July 29, 2025, Kamchatka earthquake is significantly improved by the random noise addition.
Comments25 pages, 15 figures
