Real-Time and Retrospective Analyses of Volcanic Earthquakes

火山地震的实时和回顾性分析。Thurber，University of Wisconsin-Madisonville- 040929该项目涉及威斯康星大学麦迪逊分校的研究人员与美国地质调查局阿拉斯加火山观测站（AVO）和火山灾害援助计划（VDAP）的科学家之间的合作，对阿拉斯加和世界各地一些火山的地震数据进行回顾性和实时分析。主要的科学目标是更好地了解这些活火山的背景和与火山爆发有关的地震活动的性质，开发新的或改进的内部结构模型，并开发改进的火山爆发警报和监测方法。在该项目中应用了四种主要类型的分析工具：（1）波形对齐方法，（2）事件聚类和自动拾取，（3）双差定位和层析成像，以及（4）实时事件定位处理。工具（1）和（3）是成熟的方法，而工具（2）和（4）的开发将需要作为本项目的一部分进行额外的工作。AVO目前负责对25座火山进行地震监测。平均每1至1.5年爆发一次，每年要分析数千次地震，进行有价值的地震研究的机会是巨大的。该项目确定了四个分析目标：Katmai岩群、Shishaldin、Mount Spurr和Makushin。这些目标中的每一个都是地震活跃的，并有足够数量的地震台站，以便进行详细的重新定位工作，在某些情况下还进行断层扫描。在过去8年中，VDAP深入参与了几次重大火山危机，包括瓜瓜皮钦查、通古拉瓦和科托帕希（厄瓜多尔）、波波卡特佩特和科利马（墨西哥）、拉包尔和帕戈（新几内亚）、苏弗里埃山（蒙特塞拉特）、塞罗内格罗（尼加拉瓜）和阿纳塔汉（北方马里亚纳群岛联邦）。由于VDAP在第三世界国家广泛运作，主要是在危机期间，地震网络必须迅速安装或扩大，通常由不超过四到七个1 Hz垂直传感器和一个三分量台站组成。由于以前没有关于火山结构的工作，在最初一周左右的时间里，通常会浪费大量的时间来试图为火山设计合理的地震速度模型。使用由人类分析员挑选的P和S相位优先运动来定位事件。由于网络小，初至拾取简单，可能存在速度模型误差，所有轴的定位误差可能为几公里。自动化定位方案并集成简单的（1D）建模以快速生成速度模型将在危机的关键时刻节省大量宝贵的时间，在此过程中生成精度高出一个数量级的位置。大多数VDAP和阿拉斯加火山由约6个短周期地震站监测。地震是由分析师挑选的，定位误差通常为1-2公里。因此，VDAP和AVO在定位地震时有共同的困难，特别是在危机时期，分析人员无法跟上地震活动率。我们计划在AVO测试我们的系统，该系统使用与VDAP相同的蚯蚓实时处理系统，以确定其在大规模监测环境中的有效性。拟议的工作将建立由威斯康星大学麦迪逊分校的研究人员对地震活动和活火山结构的调查多年的经验。准确的地震位置对于描述岩浆路径、识别岩浆迁移以及在依赖于位置的分析（如空间b值研究）中获得有意义的结果至关重要。一个增强的实时定位系统将有助于提高火山爆发预警能力，直接提供高质量的定位，间接减少网络运营商的工作量，从而有更多的时间来评估地震活动模式及其随时间的变化。监测地震族的时空发展将有助于我们了解火山的岩浆管道系统，并有助于确定火山何时可能从相对稳定的状态系统转变为更危险的状态。

Real-Time and Retrospective Analyses of Volcanic EarthquakesClifford H. Thurber, University of Wisconsin-MadisonEAR- 040929 This project involves a collaboration between University of Wisconsin-Madison researchers and U.S. Geological Survey scientists at the Alaska Volcano Observatory (AVO) and in the Volcano Disaster Assistance Program (VDAP) on retrospective and real-time analysis of seismic data from a number of volcanoes in Alaska and worldwide. The main scientific goals are improvement in understanding the nature of the background and eruption-related seismicity at these active volcanoes, the development of new or improved models of their internal structure, and the development of an improved eruption warning and monitoring methods. There are four main types of analysis tools that being applied in the project: (1) waveform alignment methods, (2) event clustering and auto-picking, (3) double-difference location and tomography, and (4) real-time event location processing. Tools (1) and (3) are well-developed methods, whereas the development of (2) and (4) will require additional work as part of this project. AVO is currently responsible for the seismic monitoring of 25 volcanoes. With an eruption on average every 1 to 1.5 years and thousands of earthquakes to analyze annually, the opportunities for valuable seismic research are enormous. Four targets have been identified for analysis in this project: the Katmai Group, Shishaldin, Mount Spurr, and Makushin. Each of these targets is seismically active and has an adequate number of seismic stations to allow detailed relocation work, and in some cases tomography as well. During the last 8 years, VDAP has been deeply involved in several major volcanic crises, including Guagua Pichincha, Tungurahua, and Cotopaxi (Ecuador), Popocatepetl and Colima (Mexico), Rabaul and Pago (New Guinea), Soufriere Hills (Montserrat), Cerro Negro (Nicaragua), and Anatahan (Commonwealth of the Northern Marianas Islands). Because VDAP operates extensively in 3rd world countries and predominantly during crises, seismic networks must be installed or expanded hurriedly, and generally consist of no more than four to seven 1-Hz vertical sensors and often one 3-component station. Considerable time is generally lost during the initial week or so trying to devise reasonable seismic velocity models for the volcanoes owing to absence of previous work on the volcano's structure. Events are located using P and S phase first motions picked by human analysts. Owing to the small networks, simple first arrival picking, and probable velocity model errors, location errors are probably several kilometers for all axes. Automating a location scheme and integrating simple (1D) modeling to produce a velocity model quickly would save considerable valuable time during the critical onset of a crisis, producing locations with about an order of magnitude greater precision in the process. Most VDAP and Alaskan volcanoes are monitored by ~6 short period seismic stations. Earthquakes are located by analysts' picks, and location errors are often 1-2 km. Thus VDAP and AVO share difficulties in locating earthquakes, especially during crisis times when the analysts cannot keep up with seismicity rates. We plan to test our system at AVO, which uses the same Earthworm real-time processing system as VDAP, to determine its effectiveness in a larger-scale monitoring environment. The proposed work will build many years of experience by UW-Madison researchers on the investigation of the seismicity and structure of active volcanoes. Accurate earthquake locations are essential for characterizing magmatic pathways, identifying magma migration, and for obtaining meaningful results in analyses (such as spatial b-value studies) that rely on locations. An enhanced real-time location system would contribute to improved eruption warning capability directly by providing high-quality locations rapidly and indirectly by reducing the network operator's work load, allowing for more time to evaluate the seismicity patterns and their changes with time. Monitoring the spatio-temporal development of earthquake families will help us understand a volcano's magma conduit system and help identify when a volcano may shifting from a relatively steady-state system to a more dangerous condition.