III: Medium: Collaborative Research: Scaling Time Series Analytics to Massive Seismology Datasets

III：媒介：协作研究：将时间序列分析扩展到海量地震数据集

• 批准号: 2104537
• 负责人: Abdullah Mueen
• 金额: $ 20万
• 依托单位: University of New Mexico
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-10-01 至 2025-09-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2104537&HistoricalAwards=false
• 关键词: III; Medium; Collaborative; Research; Scaling

This project will enable a team of computer scientists and earth scientists at the University of California-Riverside, the University of California-San Diego and the University of New Mexico to develop novel tools to search existing seismographic databases for subtle earthquakes that may have evaded detection. These more complete earthquake catalogues will allow more accurate hazard analysis and risk reduction. The intellectual merit of the proposed work is in creating novel data representations, definitions, algorithms (and ultimately, highly usable open-source code) that will allow the seismological community greatly to expand both the type and the scale of the analytics that they can perform, both offline and in real time. The broader impacts of this project results from the more comprehensive and complete earthquake catalogs created. These have the potential to affect multiple branches of earthquake science. For example, the more accurate hazard and risk models derived from the catalogs can be used by governments and private industry to plan for and mitigate economic and human losses, e.g., by mandating resilient construction and infrastructure, and by accurately assessing insured risk. In addition, the projects comprehensive educational and outreach activities have already been piloted on a small scale and include detailed plans to reach out to underserved communities at the K-12 and college levels, and to create grade-level appropriate teaching resources that exploit the natural interest most K-12 students have in the drama of earthquakes.Humans notice large earthquakes, but the frequently occurring smaller quakes caused by the constant slipping of fault lines typically go unnoticed, even by skilled seismologists with access to telemetry. However, these imperceptible quakes could help us understand the physical processes that trigger hazardous earthquakes, assisting in hazard-reduction efforts. Recently, a novel data structure called the Matrix Profile has emerged as a very promising technique for pattern discovery in large datasets. The PIs, an interdisciplinary team of computer scientists and seismologists, propose to investigate techniques to use Matrix Profile the scale up the size of datasets that can be investigated by 100X magnitude, to find 20X more earthquakes. The intellectual merit of this project will result in novel data representations, definitions, algorithms (and ultimately, highly usable open-source code) that will allow the seismological community to expand both the type and the scale of the analytics that they can perform. The broader impacts of this project are difficult to overstate. Comprehensive and complete earthquake catalogs are foundational to multiple branches of earthquake science, notably the physics of earthquake nucleation, hazard analysis and risk reduction. The hazard and risk models derived from them can be used by governments and private industry to plan for and mitigate economic and human losses, e.g. by mandating resilient construction and infrastructure, and by accurately assessing insured risk. The project’s comprehensive educational and outreach activities have already been piloted on a small scale and include detailed plans to reach out to under-served communities at the K-12 and college levels.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该项目将使加州大学河滨分校、加州大学圣地亚哥分校和新墨西哥州大学的一个计算机科学家和地球科学家小组能够开发新的工具，在现有的地震数据库中搜索可能逃避检测的轻微地震。这些更完整的地震目录将有助于更准确地进行灾害分析和减少风险。拟议工作的智力价值在于创建新的数据表示，定义，算法（最终，高度可用的开源代码），这将使地震学社区大大扩展他们可以执行的分析的类型和规模，无论是离线还是真实的时间。该项目的更广泛的影响来自于创建的更全面和完整的地震目录。这些都有可能影响地震科学的多个分支。例如，从目录中导出的更准确的危害和风险模型可以被政府和私营企业用于规划和减轻经济和人类损失，例如，通过强制执行弹性建筑和基础设施，以及准确评估保险风险。此外，该项目的综合教育和外展活动已经在小规模上进行了试点，包括详细的计划，以接触K-12和大学水平的服务不足的社区，并创建年级适当的教学资源，利用大多数K-12学生对地震戏剧的自然兴趣。但断层线不断滑动所造成的频繁发生的较小地震通常不会被注意到，即使是有遥测技术的熟练地震学家也不会注意到。然而，这些难以察觉的地震可以帮助我们了解引发危险地震的物理过程，有助于减少灾害的努力。最近，一种新的数据结构，称为矩阵配置文件已成为一个非常有前途的技术，在大型数据集的模式发现。PI是一个由计算机科学家和地震学家组成的跨学科团队，他们建议研究使用Matrix Profile的技术，将可以调查的数据集规模扩大100倍，以发现20倍以上的地震。该项目的智力价值将导致新的数据表示，定义，算法（最终，高度可用的开源代码），这将允许地震学社区扩展他们可以执行的分析的类型和规模。该项目的广泛影响难以夸大。全面和完整的地震目录是地震科学多个分支的基础，特别是地震成核物理学，灾害分析和风险降低。政府和私营企业可以使用从中得出的灾害和风险模型来规划和减轻经济和人员损失，例如通过强制执行弹性建筑和基础设施以及准确评估保险风险。该项目的综合教育和外展活动已经在小规模上进行了试点，包括在K-12和大学层面向服务不足的社区伸出援手的详细计划。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

FASER: Seismic Phase Identifier for Automated Monitoring

• DOI: 10.1145/3447548.3467064
• 发表时间: 2021-08
• 期刊: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining
• 作者: F. A. Chowdhury;M. A. Siddiquee;G. Baker;A. Mueen

Combining Filtering and Cross-Correlation Efficiently for Streaming Time Series

有效结合滤波和互相关来处理流时间序列

• DOI: 10.1145/3502738
• 发表时间: 2022
• 期刊: ACM Transactions on Knowledge Discovery from Data
• 影响因子: 3.6
• 作者: Zhong, Sheng;Souza, Vinicius M.;Mueen, Abdullah
• 通讯作者: Mueen, Abdullah

Septor: Seismic Depth Estimation Using Hierarchical Neural Networks

Septor：使用分层神经网络进行地震深度估计

• DOI: 10.1145/3534678.3539166
• 发表时间: 2022
• 期刊: KDD '22: Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining
• 作者: Siddiquee, M Ashraf;Souza, Vinicius M.;Baker, Glenn Eli;Mueen, Abdullah
• 通讯作者: Mueen, Abdullah