2018年0206花蓮地震因觸發了米崙斷層滑動，造成該斷層沿線的花蓮地區遭遇嚴重的災損。而該事件也引發國內外學界對其與米崙斷層構造的高度關注。此後的七年間已有多項相關研究陸續發表，顯示科學界對米崙斷層構造特性的重視。為進一步釐清米崙斷層的構造特性及花蓮地區地震相關性、潛在風險，中央研究院與中央大學地震風險管理中心（Earthquake Disaster & Risk Evaluation and Management Center，簡稱E-DREaM）共同執行了米崙斷層鑽井研究計畫（Milun Fault Drilling and All-inclusive Sensing project，簡稱MiDAS計畫）。該計畫分別於斷層上盤及下盤鑽取700公尺（井A）及500公尺（井B）的科學探測井，並成功於地表及井A、井B中佈署光纖地震儀與井下地震儀陣列，藉此監測米崙斷層的地震活動。本研究利用2023年3月16日至同年9月15日，共184天之MiDAS井下地震儀波形紀錄，建立地震目錄，以探討米崙斷層之微地震活動性。地震挑選方面，將採用人工辨識與深度學習模型RED PAN（Real-time Earthquake Detection and Phase-picking with multi-task Attention Network）兩種方法分別建立地震目錄，並根據S-P到時差及波傳方向對地震事件進行分類。
於此時段內，人工方法共識別出2,857筆地震事件，平均每日可偵測約16筆，其中21.8%為鄰近測站的地震（S-P到時差小於2秒）。進一步與氣象署地震目錄比對發現，鄰近測站的地震中約有95%之事件未記錄於氣象署地震目錄，顯示MiDAS井下地震儀對近場微地震具有優異的偵測能力。此外於鄰近測站的地震中，有306筆事件為鄰近測站的極淺層地震（震源深度至少小於500公尺，分類於S類）。本研究針對其中76筆發生於2023年5月2日之地震群進行深度分析，結果顯示該群地震之震源深度介於4至34公尺，推測為發生於排水溝下的自然極淺層地震，可能為液體流動所誘發的微地震事件。
而於深度學習模型－RED-PAN部分，本研究分別比較偵測時間視窗為60秒及30秒之模型的偵測表現。結果顯示，60秒視窗之模型（RED-PAN(60s)）具最佳表現，其精確率達76%、召回率78%。其中，對於區域型及遠場地震之召回率高達97%（S-P到時差大於2秒），但對鄰近測站之S類地震僅有3%，顯示RED-PAN模型對發生於感測器附近之微地震仍缺乏偵測能力。因此，解決RED-PAN模型對於鄰近事件之偵測能力為未來研究之首要課題，期望能發展出可長期穩定運行於MiDAS觀測系統的自動化處理系統，這將對於未來之二氧化碳封存及地熱計畫提供良好的技術支援。
本研究結果顯示花蓮北部地區地震活動度高，且未被地表地震網偵測到之微地震活動頻繁，未來搭配地震定位及規模估算，則可更完整地解析北花蓮地區微地震活動構造。而由人工地震偵測及深度學習地震偵測結果的差異顯示，若深度學習之模型沒有因區域而校正，則可能會忽略許多地震事件，這點值得給未來計畫使用深度學習模型進行地震偵測的研究借鏡。

The 2018 Hualien earthquake on Feb. 6th triggered coseismic slip along the Milun Fault, causing severe damage in Hualien City. To better understand the Milun Fault’s structure and associated seismic hazards, Institute of Earth Sciences, Academia Sinica and the Earthquake Disaster & Risk Evaluation and Management Center, National Central University initiated the Milun Fault Drilling and All-inclusive Sensing project (MiDAS). This project drilled two scientific boreholes: one was on the hanging wall (700 m, Hole A) and another was on the footwall (500 m, Hole B) of the northern Milun Fault. After drilling, an optical-fiber cable and borehole seismometer arrays were deployed in both boreholes and on the surface to monitor seismicity near the fault. In this study, we analyze seismic waveforms recorded by the borehole seismometers from March 16 to September 15, 2023 (184 days in total) to establish an earthquake catalog near the northern Hualien. We apply two methods to detect seismic events: the manual picking method and a deep learning phase picking technique, called RED PAN (Real-time Earthquake Detection and Phase-picking with multi-task Attention Network). Subsequently, we classify the seismic events based on the arrival time differences of S and P waves (Δts-p) and the directions of waveform propagation in the boreholes.
In the results of manual picking analysis, 2,857 seismic events have been detected during the study period. There are 16 events per day in average. Among these, 21.8% are near-station events (Δts-p < 2 s). About 95% of these near-station events are not detected by the Central Weather Administration (CWA). It indicates that the MiDAS borehole arrays have higher capability in earthquake detection compared to the regional surface network (CWASN). Furthermore, there are 306 events occurring at ultra-shallow depths (< 500 m) during the study period, called Class S. We obtain that a cluster of 76 Class S events on May 2, 2023 occurred along the storm drain nearby with focal depths ranging from 4 to 34 m. These events may be induced by fluid migration.
For the deep learning analysis, we compare RED-PAN models with two different time window lengths (60 seconds and 30 seconds). The model with a 60-second prediction window shows the best performance, achieving a precision of 76% and a recall of 78%. For regional events (Δts-p > 2 s), the recall reaches 97%, indicating the model's strong ability to detect regional earthquakes. However, the recall for Class S is only 3%, suggesting that RED-PAN currently struggles to identify microevents occurring very close to the sensors. Enhancing the model's capability to detect near-field events will be the priority in the future. The results would be useful for the Carbon Capture and Storage as well as geothermal projects.
In conclusion, our results indicate that the seismicity in the northern Hualien is high. Many microearthquakes have not been detected by the regional seismic network (CWASN). We may identify potential structures for these microevents by further estimating their locations and magnitudes in the future. Furthermore, the discrepancy of earthquake detections between the manual and deep learning methods suggests that the deep learning method would be missing amount of microevents if we do not retrain the model by local seismic data. This serves as a valuable reference for future research involving deep learning.

Allen, R. V. (1978). Automatic earthquake recognition and timing from single traces. Bulletin of the Seismological Society of America, 68(5), 1521-1532. https://doi.org/10.1785/bssa0680051521
Bohnhoff, M., Baisch, S., & Harjes, H. P. (2004). Fault mechanisms of induced seismicity at the superdeep German Continental Deep Drilling Program (KTB) borehole and their relation to fault structure and stress field. Journal of Geophysical Research: Solid Earth, 109(B2). https://doi.org/10.1029/2003jb002528
Chan, C. H., Ma, K. F., Shyu, J. B. H., Lee, Y. T., Wang, Y. J., Gao, J. C., Yen, Y. T., & Rau, R. J. (2020). Probabilistic seismic hazard assessment for Taiwan: TEM PSHA2020. Earthquake Spectra, 36(1_suppl), 137-159. https://doi.org/10.1177/8755293020951587
Goertz-Allmann, B. P., Kühn, D., Oye, V., Bohloli, B., & Aker, E. (2014). Combining microseismic and geomechanical observations to interpret storage integrity at the In Salah CCS site. Geophysical Journal International, 198(1), 447-461. https://doi.org/10.1093/gji/ggu010
Huang, H. H., & Wang, Y. (2022). Seismogenic structure beneath the northern Longitudinal Valley revealed by the 2018–2021 Hualien earthquake sequences and 3-D velocity model. Terrestrial, Atmospheric and Oceanic Sciences, 33(1). https://doi.org/10.1007/s44195-022-00017-z
Huang, H. H., Ma, K. F., Wu, E. S., Cheng, Y. Z., Lin, C. J., Ku, C. S., & Su, P. L. (2024). Spatiotemporal Monitoring of a Frequent‐Slip Fault Zone Using Downhole Distributed Acoustic Sensing at the MiDAS Project. In Distributed Acoustic Sensing in Borehole Geophysics (pp. 459-475). https://doi.org/10.1002/9781394179275.ch26
Huang, M. H., & Huang, H. H. (2018). The Complexity of the 2018 Mw 6.4 Hualien Earthquake in East Taiwan. Geophysical Research Letters, 45(24). https://doi.org/10.1029/2018gl080821
Imanishi, K., Ellsworth, W. L., & Prejean, S. G. (2004). Earthquake source parameters determined by the SAFOD Pilot Hole seismic array. Geophysical Research Letters, 31(12). https://doi.org/10.1029/2004gl019420
Kubo, H., Naoi, M., & Kano, M. (2024). Recent advances in earthquake seismology using machine learning. Earth, Planets and Space, 76(1). https://doi.org/10.1186/s40623-024-01982-0
Kuo, C. H., Huang, J. Y., Lin, C. M., Hsu, T. Y., Chao, S. H., & Wen, K. L. (2018). Strong Ground Motion and Pulse‐Like Velocity Observations in the Near‐Fault Region of the 2018 Mw 6.4 Hualien, Taiwan, Earthquake. Seismological Research Letters, 90(1), 40-50. https://doi.org/10.1785/0220180195
Kuochen, H., Wu, Y. M., Chang, C. H., Hu, J. C., & Chen, W. S. (2004). Relocation of Eastern Taiwan Earthquakes and Tectonic Implications. Terrestrial Atmospheric and Oceanic Sciences, 15, 647-666. https://doi.org/10.3319/TAO.2004.15.4.647(T)
Nishigami, K. Y., Ando, M., & Tadokoro, K. (2008). Seismic observations in the DPRI 1800 m borehole drilled into the Nojima Fault zone, south‐west Japan. Island Arc, 10(3-4), 288-295. https://doi.org/10.1111/j.1440-1738.2001.00327.x
Lee, S. J., Lin, T. C., Liu, T. Y., & Wong, T. P. (2019). Fault‐to‐Fault Jumping Rupture of the 2018 Mw 6.4 Hualien Earthquake in Eastern Taiwan. Seismological Research Letters, 90(1), 30-39. https://doi.org/10.1785/0220180182
Liao, W. Y., Lee, E. J., Mu, D., Chen, P., & Rau, R. J. (2021). ARRU Phase Picker: Attention Recurrent-Residual U-Net for Picking SeismicP- andS-Phase Arrivals. Seismological Research Letters, 92(4), 2410-2428. https://doi.org/10.1785/0220200382
Liao, W. Y., Lee, E. J., Mu, D., & Chen, P. (2022a). Toward Fully Autonomous Seismic Networks: Backprojecting Deep Learning-Based Phase Time Functions for Earthquake Monitoring on Continuous Recordings. Seismological Research Letters, 93(3), 1880-1894. https://doi.org/10.1785/0220210274
Liao, W. Y., Lee, E. J., Chen, D. Y., Chen, P., Mu, D., & Wu, Y. M. (2022b). RED-PAN: Real-Time Earthquake Detection and Phase-Picking With Multitask Attention Network. IEEE Transactions on Geoscience and Remote Sensing, 60, 1-11. https://doi.org/10.1109/tgrs.2022.3205558
Lin, Y. Y., Kanamori, H., Zhan, Z., Ma, K. F., & Yeh, T. Y. (2020). Modelling of pulse-like velocity ground motion during the 2018 Mw 6.3 Hualien earthquake, Taiwan. Geophysical Journal International, 223(1), 348-365. https://doi.org/10.1093/gji/ggaa306
Lin, Y. Y., Chan, J. B., Chen, W. Y., & the MiDAS group (2024, October 7–10). 10-m-deep earthquake swarms (Mw -2) near the Milun fault in Hualien, Taiwan, detected by the MiDAS seismic monitoring system [Conference presentation]. 2024 Photonic Seismology: Lighting the Way Forward, Vancouver, Canada. Seismological Society of America.
Lindsey, N. J., Rademacher, H., & Ajo‐Franklin, J. B. (2020). On the Broadband Instrument Response of Fiber‐Optic DAS Arrays. Journal of Geophysical Research: Solid Earth, 125(2). https://doi.org/10.1029/2019jb018145
Ma, K. F., von Specht, S., Kuo, L. W., Huang, H. H., Lin, C. R., Lin, C. J., Ku, C. S., Wu, E. S., Wang, C. Y., Chang, W. Y., & Jousset, P. (2024). Broad-band strain amplification in an asymmetric fault zone observed from borehole optical fiber and core. Communications Earth & Environment, 5(1), 13, Article 402. https://doi.org/10.1038/s43247-024-01558-6
Mousavi, S. M., Sheng, Y., Zhu, W., & Beroza, G. C. (2019). STanford EArthquake Dataset (STEAD): A Global Data Set of Seismic Signals for AI. IEEE Access, 7, 179464-179476. https://doi.org/10.1109/access.2019.2947848
Mousavi, S. M., Ellsworth, W. L., Zhu, W., Chuang, L. Y., & Beroza, G. C. (2020). Earthquake transformer-an attentive deep-learning model for simultaneous earthquake detection and phase picking. Nat Commun, 11(1), 3952. https://doi.org/10.1038/s41467-020-17591-w
Wen, Y. Y., Wen, S., Lee, Y. H., & Ching, K. E. (2019). The kinematic source analysis for 2018 Mw 6.4 Hualien, Taiwan earthquake. Terrestrial, Atmospheric and Oceanic Sciences, 30(3), 377-387. https://doi.org/10.3319/tao.2018.11.15.03
Yu, C., Vavryčuk, V., Adamová, P., & Bohnhoff, M. (2018). Moment Tensors of Induced Microearthquakes in The Geysers Geothermal Reservoir From Broadband Seismic Recordings: Implications for Faulting Regime, Stress Tensor, and Fluid Pressure. Journal of Geophysical Research: Solid Earth, 123(10), 8748-8766. https://doi.org/10.1029/2018jb016251
Yu, S. B., Chen, H. Y., & Kuo, L. C. (1997). Velocity field of GPS stations in the Taiwan area. Tectonophysics, 274(1), 41-59. https://doi.org/https://doi.org/10.1016/S0040-1951(96)00297-1
Zhu, W., & Beroza, G. C. (2019). PhaseNet: A Deep-Neural-Network-Based Seismic Arrival Time Picking Method. Geophysical Journal International. https://doi.org/10.1093/gji/ggy423
林彥宇、林欽仁、陳文瑜：〈MiDAS井下地震儀陣列深度及方位校正〉，中華民國地球物理學會與中華民國地質學會112年年會暨學術研討會，桃園，臺灣，2023。
林啟文、陳文山，劉彥求、陳柏村：〈米崙斷層〉，《經濟部中央地質調查所特刊》，第23號，2009，11-20頁。
林又青、何瑞益、王俞婷、傅鏸漩、梁庭語、施虹如、李威霖、陳珮琦、李士強、呂喬茵、林聖琪、陳偉柏、劉哲欣、張志新：〈112 年度豪雨及颱風事件災情彙整報告〉，國家災害防救科技中心，2023。
花蓮縣政府主計處：〈人口〉，《中華民國一一二年花蓮縣統計年報》，花蓮縣政府，2024，28-89頁。
曾雅筑：〈台灣東部花蓮地區米崙活動斷層之古地震研究〉，碩士論文，國立中央大學，2019。
經濟部中央地質調查所：〈20180206花蓮地震地質調查報告〉，《經濟部中央地質調查所報告》，經濟部中央地質調查所，2018，共131頁。
劉平妹、謝孟龍：〈臺灣東部晚第四紀地質調查及地形演育研究(2/2)〉，《經濟部中央地質調查所報告》，第96-01號，經濟部中央地質調查所，2007，共85頁。
鄭世楠、余騰鐸、葉永田、張建興：〈1951年花蓮-台東地震系列之重定位〉，紀念台灣地區氣象測報一百年天氣分析與預報研討會海象與地震論文彙編，1997，690-699頁。
ASIR官網：〈光學加速度儀照片〉，2025年3月31日，取自https://asirseismic.com/product/broadband-borehole-accelerometer/。
ASIR官網：〈4.5 Hz短週期地震儀照片〉，2025年3月31日，取自https://asirseismic.com/product/4-5-hz-borehole-seismometer-geophone/。
ASIR官網：〈全頻段地震儀照片〉，2025年3月31日，取自https://asirseismic.com/product/fullband-borehole-seismometer/。
Liao, W. Y. ：〈RED-PAN模型程式碼下載平台〉，2023年7月16日，取自https://github.com/tso1257771/RED-PAN。
Liao, W. Y. ：〈RED-PAN版本更新歷史紀錄〉，2025年4月15日，取自https://github.com/tso1257771/RED-PAN/commits/main/。
Nanometrics官網：〈寬頻地震儀照片〉，2025年3月31日，取自https://nanometrics.ca/instrumentation/products/seismometers/meridian-compact-posthole。
Nanometrics官網：〈強震儀照片〉，2025年3月31日，取自https://nanometrics.ca/instrumentation/products/accelerometers/titan。
Seisman：〈SAC參考手冊〉，2025年6月5日，取自https://seisman.github.io/SAC_Docs_zh/。
Silixa官網：〈iDAS儀器及儀器參數〉，2025年1月26日，取自https://silixa.com/technology/idas-intelligent-distributed-acoustic-sensor/。
中央研究院地球科學所電子室：〈the MiDAS project〉，2025年4月17日，取自https://www.elab.earth.sinica.edu.tw/midas-events。
中央氣象署：〈CWA 地面氣象站－花蓮站〉，2024年11月01日，取自https://codis.cwa.gov.tw/StationData。
中央氣象署：〈CWA 2023年潮汐預報〉，2024年11月01日，取自https://www.cwa.gov.tw/Data/service/notice/download/Publish_20220803143837.pdf。
中視新聞：〈海葵重擊花蓮! 七星潭掀大浪 路樹塌路燈吹斷│中視新聞 20230903〉，2025年7月29日，取自https://reurl.cc/VWRO2N。