Abstract. Seismic arrays provide tools for the localization of events without clear phases or events outside of the network, where the station coverage prohibits classical localization techniques. Beamforming allows the determination of the direction (backazimuth) and the horizontal (apparent) velocity of an incoming wavefront. Here we combine multiple arrays to retrieve event epicenters from the area of intersecting beams without the need to specify a velocity model. The analysis is performed in the time-domain, which allows to select a relatively narrow time window around the phase of interest while preserving frequency bandwidth. This technique is applied to earthquakes and hybrid events in the region of Fogo and Brava, two islands of the southern chain of the Cape Verde archipelago. The results show that the earthquakes mainly originate near Brava whereas the hybrid events are located on Fogo. By multiple-event beam-stacking we are able to further constrain the locations of the hybrid events in the north-western part of the collapse scar of Fogo. In previous studies, these events were attributed to shallow hydrothermal processes. However, we obtain relatively high apparent velocities at the arrays, pointing to either deeper sources or to complex ray paths. For a better understanding of possible errors of the multi-array analysis, we also compare slowness values obtained from the array analysis with those derived from earthquake locations from classical (local network) localizations. In general, the results agree well, however, the arrays also show some aberrations that can be quantified for certain event locations.

中文翻译：

---

佛得角福戈和布拉瓦火山地震信号的多阵列分析

摘要。地震阵列提供了用于事件定位的工具，而没有明确的阶段或网络外部的事件，而站外覆盖禁止经典的定位技术。波束成形可以确定入射波阵面的方向（后方位角）和水平（视在）速度。在这里，我们组合了多个阵列，以从相交的光束区域检索事件震中，而无需指定速度模型。分析是在时域中进行的，这样可以在保持频率带宽的同时，在感兴趣的相位周围选择相对较窄的时间窗口。这项技术适用于佛得角群岛南链的两个岛屿福戈和布拉瓦地区的地震和混合事件。结果表明，地震主要起源于布拉瓦附近，而混合事件位于福戈。通过多事件波束堆叠，我们可以进一步限制混合事件在Fogo塌陷疤痕西北部的位置。在以前的研究中，这些事件均归因于浅层热液过程。但是，我们在阵列处获得相对较高的视在速度，指向更深的光源或复杂的射线路径。为了更好地理解多阵列分析的可能误差，我们还比较了从阵列分析获得的慢度值与从经典（本地网络）定位的地震位置得出的慢度值。通常，结果吻合得很好，但是，阵列还显示了可以针对某些事件位置进行量化的一些像差。