AI 中文总结
研究随机和自然地震噪声对波形互相关性能的影响，通过用随机噪声计算WCC公告及对旁向敏感性建模等方法，解决相关问题，证实基于WCC事件的统计显著性，增强WCC检测并使其接近最佳性能。
AI 中文摘要
波形互相关（WCC）应用于地震阵列数据可将检测阈值降低约一个数量级。应用于国际监测系统（IMS）数据时，检测到的最小WCC事件比国际数据中心（IDC）报告的至少低一个震级单位，揭示了2025年7月29日地震前低震级活动模式与2013年5月24日鄂霍次克海地震前相似。震级接近堪察加半岛复发曲线拐角值的事件数量持续增加可能用作大震准备的前兆指标。基于WCC的事件信噪比低于IDC检测阈值，分析师常看不见。由于缺乏随机噪声参考和IMS阵列潜在的旁向敏感性，WCC事件假设的统计显著性不确定。本文解决了这两个问题。用随机噪声计算WCC公告，结果显示WCC事件数量极少，证实了基于WCC事件的统计显著性。用2011年3月11日东北地震对旁向敏感性进行建模。在WCC处理中使用堪察加半岛区域内的主事件（MEs）时会产生大量假事件，添加来自东北区域的MEs后所有假事件消失。将随机噪声缩放到处理区间内最大振幅的一小部分时，WCC检测增强，产生更多检测结果和基于WCC的事件，这类似于噪声白化效应，使匹配滤波器检测器在随机噪声条件下更敏感且接近其最佳性能。
英文摘要
Waveform cross-correlation (WCC) applied to data from seismic arrays allows for the reduction of the detection threshold by approximately an order of magnitude. When applied to the data of the IMS, the smallest WCC detected events were by at least one magnitude unit lower than those reported by the IDC and revealed that the pattern of low-magnitude activity prior to the July 29, 2025, earthquake was similar to that prior to the May 24, 2013, Sea of Okhotsk earthquake. The consistent increase in the number of events with magnitudes approaching the corner value of the Kamchatka recurrence curve can likely be used as a precursory indicator of a mega-earthquake preparation. The WCC-based events are characterized by detections with a signal-to-noise ratio below the IDC detection threshold and are often not visible to analysts. This makes the statistical significance of the WCC event hypotheses uncertain due to the absence of a random noise reference and the potential side-sensitivity of the IMS arrays. Both problems are addressed. Random noise is used to calculate a WCC bulletin. This exercise shows a negligibly low number of WCC events and confirms the statistical significance of the WCC-based events. The side-sensitivity is modelled using the March 11, 2011, Tohoku earthquake. When master events (MEs) within the Kamchatka region are used in the WCC processing, a large number of false events are generated. When MEs from the Tohoku zone are added, all false events disappear since the Tohoku MEs win the conflict resolution process. When the random noise is scaled to a small fraction of the maximum amplitude within the processed interval, the WCC detection is enhanced with more detections and WCC-based events created. This is an effect similar to noise whitening, which makes the matched filter detector more sensitive and closer to its optimal performance under stochastic noise conditions.
Comments27 pages, 5 figures, 6 tables
