A Seismic Investigation of the Magmatic and Hydrothermal Systems of Nabro Volcano and other East African Rift Volcanoes

火山地震学的基本目标之一是发展对活动岩浆系统动力学的理解。通过研究火山的地震活动性，可以绘制火山的地下地图：确定是什么驱动和滋养了岩浆系统，并追踪其随时间的演变。这对评估火山危险性和预测喷发行为具有重要意义。了解火山流体如何相互作用对于评估火山的地热能源生产潜力也至关重要。地震层析成像是一种强大的地球物理工具，用于在各种尺度上对地球内部进行成像。该技术通过反演地震数据来约束二维或三维纵波和横波速度结构以及它们之间的比值（Vp/Vs）。为了更好地了解导致火山动荡的地下活动岩浆过程，它经常在局部尺度上应用于火山变形（例如Greenfield等人，2016；Wilks等人，2020）。火山下的地震速度结构的知识有助于识别熔岩承载区域和深度流体，限制岩浆和热液系统之间的相互作用。横波分裂是另一种地震方法，越来越多地应用于火山过程的研究，特别是地下的应力变化（例如Nowacki et al., 2018）。当地震波在各向异性材料中传播时，会发生横波分裂；通过测量由此产生的地震波速度和传播方向的变化，可以表征地下的各向异性。测量地震各向异性提供了动态过程的指示，例如变形，这是其他技术无法看到的（Wookey, 2012）。Nabro火山是一座离裂谷火山，位于埃塞俄比亚-厄立特里亚边界附近的达纳基尔微板块的中部（Hamlyn et al., 2014）。尽管纳伯罗火山于2011年喷发，但其岩浆管道系统的结构仍然是个谜。总的来说，目前还不清楚岩浆是如何供应给偏离扩张轴的火山的；此外，它们在适应扩展方面的作用尚未得到很好的理解（Maccaferri et al., 2014）。2011年8月至2012年10月在Nabro附近建立的地震台网获得了14个月的地震数据。哈姆林等人（2014）先前对这些数据进行了部分分析；然而，由于手工挑选P波和s波到达的时间很长，大多数地震都没有被编录。最近，一种基于卷积神经网络的自动相位到达检测的新深度学习模型被应用于Nabro的地震数据（Lapins et al., 2020）。由此产生了33950个事件的目录（图1），远远超过了以前手工识别的事件目录，这表明Nabro是地震层析反演和横波分裂各向异性分析的首选。与纳布罗火山一样，东非裂谷系统中还有其他火山，它们将成为地震调查其地下过程的有趣目标。例如，Corbetti火山和Aluto火山都进行了几个月的地震监测，因此有大量未经分析的地震数据可供利用。alutoi目前是一种可行且高产的地热资源（Hochstein等人，2017），并计划在Corbetti建立地热发电厂；对这些火山的热液和岩浆系统的分析将有助于优化地热生产。

One of the fundamental objectives of volcano seismology is to develop an understanding of the dynamics of active magmatic systems. By studying the seismicity of a volcano, it is possible to map the volcanic subsurface:determining what drives and feeds the magmatic system and tracing its evolution through time. This has important implications for assessing volcanic hazard and predicting eruptive behaviour. Understanding how volcanic fluidsinteract is also vital for evaluating a volcano's geothermal energy production potential. Seismic tomography is a powerful geophysical tool used to image the Earth's interior at a variety of scales. Thetechnique inverts seismic data to constrain the 2D or 3D P- and S-wave velocity structures and the ratio between them (Vp/Vs). It has been frequently applied on a local scale to deforming volcanoes, in order to better understandthe subsurface active magmatic processes responsible for volcanic unrest (e.g. Greenfield et al., 2016; Wilks et al., 2020). Knowledge of the seismic velocity structure beneath a volcano helps to identify melt-bearing regions andfluids at depth, constraining interactions between magmatic and hydrothermal systems. Shear wave splitting is another seismic method that is being increasingly applied to the investigation of volcanologicalprocesses, particularly stress variations in the subsurface (e.g. Nowacki et al., 2018). Shear wave splitting occurs when seismic waves propagate through an anisotropic material; by measuring the resulting variation inseismic wave speed and direction of propagation, the anisotropy of the subsurface can be characterised. Measuring seismic anisotropy provides an indication of dynamic processes, such as deformation, which are invisible to othertechniques (Wookey, 2012). Nabro is an off-rift volcano located on the central part of the Danakil microplate near the Ethiopia-Eritrea border (Hamlyn et al., 2014). Despite the fact that Nabro erupted in 2011, the structure of its magma plumbing system remains a mystery. In general, it is still unclear how magma is supplied to such volcanoes offset from the axis of spreading; further, their role in accommodating extension is not well understood (Maccaferri et al., 2014).A seismic network established around Nabro from August 2011 - October 2012 yielded 14 months of seismic data. These data have been partially analysed previously by Hamlyn et al. (2014); however, most of the seismicityhas not been catalogued, due to the time-consuming nature of manually picking P- and S-wave arrivals. A new deep learning model for automating phase arrival detection, based on convolutional neural networks, has recentlybeen applied to the seismic data from Nabro (Lapins et al., 2020). This has produced a catalogue of 33,950 events (Figure 1), far more than the previous catalogues of manually-identified events, indicating that Nabro is a primecandidate for both a seismic tomography inversion and a shear wave splitting anisotropy analysis. As well as Nabro, there are other volcanoes in the East African Rift System that would be interesting targets fora seismic investigation into their subsurface processes. For example, both Corbetti and Aluto volcanoes have been seismically monitored over several months and hence there is a wealth of unanalysed seismic data to exploit. Alutois currently a viable and productive geothermal resource (Hochstein et al., 2017), and there are plans to establish a geothermal power plant at Corbetti; an analysis of the hydrothermal and magmatic systems of these volcanoes willhelp optimise geothermal production.