Earthquake Detection and P-Wave Arrival Time Picking Using Capsule Neural Network
Earthquake Detection and P-Wave Arrival Time Picking Using Capsule Neural Network
复制标题
DOI：
10.1109/tgrs.2020.3019520
复制
发表时间：
2021-07-01
影响因子：
8.2
通讯作者：
Chen, Yangkang
中科院分区：
文献类型：
--
作者：
Saad, Omar M.;Chen, Yangkang
地震探测是观测地震学的重要环节。我们建议利用胶囊神经网络（CapsNet）来自动识别和检测地震。CapsNet是新一代的深度学习架构。它具有从小数据集学习的能力，具有很好的泛化性能。我们使用50 x 0025;的南加州地震数据（225万个4-s-三分量地震窗口）训练CapsNet，并使用来自不同地震区的222 395个波形来评估CpasNet性能，例如，美国西部、欧洲和日本。结果，CapsNet遗漏了367个事件，检测到217 305个事件，准确率为97.71x0025;。在这些拾取的事件中，210 498个事件的到达时间误差低于0.2 s（96.86x0025;），197968个波形的到达时间误差低于0.1 s（91.11x0025;）。CapsNet的精确度、召回率和F1得分分别为97.78x0025;、99.83x0025;和98.79x0025;。此外，CapsNet使用100 000个60 s三分量地震噪声波形进行测试。CapsNet显示出1384的低误报率，这使得CapsNet的准确度为98.61x0025;。此外，CapsNet使用与阿肯色州地区发生的24小时微震群相关的连续地震数据进行测试。因此，CapsNet检测到221次地震，并发布了37次假警报，检测精度为85.65x0025;。CapsNet可以检测到许多震级很小的微震，低至x2212;1.3 Ml，并检测到低信噪比（SNR）的地震，例如，低至x2212;8.07 dB。将CapsNet的结果与基准方法进行比较，例如，短时平均/长时平均（STA/LTA）和GPD方法。CapsNet显示出最高的拣选精度，并优于基准方法。
Earthquake detection is an essential step in observational earthquake seismology. We propose to utilize a capsule neural network (CapsNet) to automatically identify and detect earthquakes. CapsNet is the new generation of deep learning architecture. It has the capability of learning with a great generalization performance from a small dataset. We train the CapsNet using 50x0025; of the Southern California seismic data (2.25 million 4-s-three-component seismic windows) and use 222 395 waveforms from different seismic areas to evaluate the CpasNet performance, e.g., western United States, Europe, and Japan. As a result, the CapsNet misses 367 events and detects 217 305 events with an accuracy of 97.71x0025;. Among these picked events, 210 498 events have an arrival time error below 0.2 s (96.86x0025;) and 197968 waveforms with an arrival time error below 0.1 s (91.11x0025;). The CapsNet precision, recall, and F1-score are 97.78x0025;, 99.83x0025;, and 98.79x0025;, respectively. In addition, the CapsNet is tested using 100 000 60-s-three-component seismic noise waveforms. CapsNet shows a low false alarms rate of 1384, which gives the CapsNet an accuracy of 98.61x0025;. In addition, CapsNet is tested using continuous seismic data associated with the 24-hours microearthquakes swarm that occurred in the Arkansas area. Accordingly, the CapsNet detects 221 earthquakes and releases 37 false alarms with a detection accuracy of 85.65x0025;. CapsNet detects many microearthquakes with a small magnitude, as low as x2212;1.3 Ml, and detects earthquakes that have a low signal-to-noise ratio (SNR), e.g., as low as x2212;8.07 dB. The results of the CapsNet are compared to the benchmark methods, e.g., short-time average/long-time average (STA/LTA) and GPD methods. The CapsNet shows the highest picking accuracy and outperforms the benchmark methods.