Machine learning based aftershock catalogs of the Mw 7.8, February 6th, 2023, Karamanmaras earthquake

The dataset contains three seismicity catalogs covering the first 5 days of the aftershock sequence of the Mw 7.8 Karamanmaraş and Mw 7.6 Elbistan earthquakes that occurred in Türkiye on February 6th, 2023. The catalogs are derived from machine learning (ML) approaches operating on continuous data from 38 permanent seismological stations covering the area of the aftershock sequence and span the time interval 06.02.2023-10.02.2023. The seismological stations are operated by AFAD (Disaster and Emergency Management Presidency of Turkey) and KOERI (Kandilli Observatory and Earthquake Research Institute). Automatic P- and S-phase picks were obtained using the deep learning PhaseNet software (Zhu & Beroza, 2019), and either GaMMA (Zhu et al., 2022) or GENIE (McBrearty & Beroza, 2023) routines were used to associate these phases into seismic events. The probabilitic NLLoc earthquake location software (Lomax et al., 2009) was used to produce single event locations and final relative relocations were obtained after applying the hypoDD software (Waldhauser & Ellsworth, 2000). This resulted in two single event location NLLoc aftershock catalogs based on GaMMA and GENIE event association and containing 17,550 and 14,805 event detections in the time interval 06.02.2023 01:18 UTC - 11.02.2023 00:00 UTC, respectively.

The hypoDD based catalog of better constrained relative relocations contains 5,215 events. The magnitude range is between M-0.1 and M6.9 with time-variable magnitude of completeness. The catalog covers the area 36.00S-39.00S and 35.40E-40.00E. The full description of the data and methods is provided in the data description file.

本数据集包含3个地震目录，涵盖2023年2月6日发生于土耳其的Mw7.8卡拉曼马拉什（Karamanmaraş）地震与Mw7.6埃尔比斯坦（Elbistan）地震后的前5天余震序列。该系列目录基于机器学习（machine learning, ML）方法，对覆盖余震序列区域的38个固定地震台站的连续观测数据处理得到，时间跨度为2023年2月6日至2023年2月10日。上述地震台站由土耳其灾害与应急管理总局（AFAD，Disaster and Emergency Management Presidency of Turkey）与坎迪利天文台与地震研究所（KOERI，Kandilli Observatory and Earthquake Research Institute）运营。研究人员首先使用深度学习软件PhaseNet（Zhu & Beroza, 2019）自动拾取P波与S波初至，随后采用GaMMA（Zhu et al., 2022）或GENIE（McBrearty & Beroza, 2023）算法将这些初至关联为地震事件，继而使用概率型NLLoc地震定位软件（Lomax et al., 2009）生成单事件定位结果，最终通过hypoDD软件（Waldhauser & Ellsworth, 2000）完成相对重定位。基于GaMMA与GENIE事件关联流程得到的两个NLLoc单事件定位余震目录，分别在2023年2月6日01:18协调世界时（UTC）至2023年2月11日00:00协调世界时（UTC）的时间区间内，包含17550条与14805条地震事件检测记录。
基于hypoDD的定位约束更优的相对重定位目录共收录5215个地震事件，其震级范围为M-0.1至M6.9，完备震级随时间动态变化。该目录覆盖区域为南纬36.00°至南纬39.00°、东经35.40°至东经40.00°。本数据集的详细数据与方法说明，请参阅配套的数据描述文件。