強震預警系統主要目的在於提前發出精準的警報，讓近震央區域有時間做好預防措施，以減少強震波帶來的傷亡與財損。目前交通部氣象局採用的是區域型的強震預警系統，並以縣市為單位發出警報，對於大部分地震可做出良好的預測，但在近震央地區的預警時效性卻不盡理想，因此，本研究的目標是期望能夠增加近震央地區的預警時效以縮小盲區，同時也要減少誤報的發生率。
近年來受惠於電腦硬體的不斷進步，電腦的運算相較於以往更加迅速，這使得深度學習的應用蓬勃發展，其中在強震預警領域也不例外。前人已將卷積神經網路(CNN)應用於現地型強震預警，並用歷史地震驗證得出現地型強震預警能有一定的實用性。因此，本研究想更進一步的將此成果應用於氣象局之Earthworm系統中，希望可以提升近震央區域預警的速度。而為了要減少誤報的發生，本研究也嘗試以多測站預測之PGA皆大於門檻值作為測站發報條件。接著利用近十年發生的有感地震進行模擬警報發布，並將現地型警報離線分析的結果加入到區域型警報變成複合式的強震預警，經過統計後本研究建議的方法在無容許誤差時警報的準確率與召回率分別為69%、85%；有容許誤差時警報的準確率與召回率分別為96%、100%，而複合式的強震預警在近震央地區可以比區域型警報增加約1~2秒的警報時效。

The earthquake early warning system aim to alert before the arrival of strong seismic waves, so that the area near the epicenter has time to take preventive measures. At present, the Central Weather Bureau (CWB) adopts a regional earthquake early warning system (EEW) and issue earthquake alerts in units of counties and cities. The regional EEW performs well for most earthquakes, but the timeliness of earthquake alerts in areas near the epicenter is not ideal. Therefore, the goal of this study is to improve the timeliness in the area near the epicenter to narrow the blind area but not substantially increase the incidence of false alarms at the same time.
The previous research had applied the convolutional neural network (CNN) to the on-site EEW, and the CNN-based on-site EEW had illustrated its ability using recent damaging earthquakes in Taiwan. Therefore, this study intends to further apply the CNN-based on-site EEW to the Earthworm system of the CWB, hoping to improve the speed of early warning in the area near the epicenter. In order to reduce the occurrence of false alerts, this study also proposed to use the conditions based on peak ground accelerations predicted by multiple stations to be greater than the threshold. Then use the earthquakes that occurred in the past ten years to issue simulated alarms, and add the results of on-site EEW to the Earthworm system to form a hybrid earthquake early warning system. After statistics, the accuracy and recall of the method proposed in this study when there is without tolerance are 69% and 85% respectively; when there is with tolerance are 95.93% and 100% respectively. The hybrid earthquake early warning system can increase the warning time of about 1~2 seconds in the area near the epicenter compared with the regional earthquake early warning system.

[1] L. Geiger, "Probability method for determination of earthquake epicenters from arrival time only," 1912.
[2] D.-Y. Chen, T.-L. Lin, H.-C. Hsu, Y.-C. Hsu, and N.-C. Hsiao, "An approach to improve the performance of the earthquake early warning system for the 2018 Hualien earthquake in Taiwan," Terr. Atmos. Ocean. Sci, vol. 30, pp. 423-433, 2019.
[3] R. M. Allen, P. Gasparini, O. Kamigaichi, and M. Bose, "The status of earthquake early warning around the world: An introductory overview," Seismological research letters, vol. 80, no. 5, pp. 682-693, 2009.
[4] M. Böse, T. Heaton, and E. Hauksson, "Rapid estimation of earthquake source and ground‐motion parameters for earthquake early warning using data from a single three‐component broadband or strong‐motion sensor," Bulletin of the Seismological Society of America, vol. 102, no. 2, pp. 738-750, 2012.
[5] Y.-M. Wu and H. Kanamori, "Rapid assessment of damage potential of earthquakes in Taiwan from the beginning of P waves," Bulletin of the Seismological Society of America, vol. 95, no. 3, pp. 1181-1185, 2005.
[6] T.-Y. Hsu et al., "Rapid on-site peak ground acceleration estimation based on support vector regression and P-wave features in Taiwan," Soil Dynamics and Earthquake Engineering, vol. 49, pp. 210-217, 2013.
[7] 黃昭文, "卷積神經網路應用於初達波預估當站最大地表加速度之研究," 碩士, 營建工程系, 國立臺灣科技大學, 台北市, 2020
[8] T. Hsu et al., "Comparing the performance of the NEEWS earthquake early warning system against the CWB system during the 6 February 2018 Mw 6.2 Hualien earthquake," Geophysical Research Letters, vol. 45, no. 12, pp. 6001-6007, 2018.
[9] T. Y. Hsu, H. H. Wang, P. Y. Lin, C. M. Lin, C. H. Kuo, and K. L. Wen, "Performance of the NCREE's on‐site warning system during the 5 February 2016 Mw 6.53 Meinong earthquake," Geophysical Research Letters, vol. 43, no. 17, pp. 8954-8959, 2016.
[10] S. J. Lee, H. W. Chen, Q. Liu, D. Komatitsch, B. S. Huang, and J. Tromp, "Three-dimensional simulations of seismic-wave propagation in the Taipei Basin with realistic topography based upon the spectral-element method," Bulletin of the Seismological Society of America, vol. 98, no. 1, pp. 253-264, 2008.
[11] V. Y. Sokolov, C. H. Loh, M. Eeri, and K. L. Wen, "Empirical models for site- and region dependent ground-motion parameters in the Taipei area: A unified approach," Earthquake Spectra, vol. 17, no. 2, pp. 313-331, 2001.
[12] V. Y. Sokolov, C. H. Loh, and K. L. Wen, "Empirical Study of Sediment-Filled Basin Response: The Case of Taipei City," Earthquake Spectra, vol. 16, no. 3, pp. 681-707, 2000.
[13] J. Wang, X. Yun, H. Kuo-Chen, and Y.-M. Wu, "CAV site-effect assessment: A case study of Taipei Basin," Soil Dynamics and Earthquake Engineering, vol. 108, pp. 142-149, 2018.
[14] 陳達毅, "Earthworm平台應用於臺灣地震速報預警研究," 博士, 地質科學研究所, 國立臺灣大學, 台北市, 2015.
[15] D. Y. Chen, N. C. Hsiao, and Y. M. Wu, "The Earthworm based earthquake alarm reporting system in Taiwan," Bulletin of the Seismological Society of America, vol. 105, no. 2A, pp. 568-579, 2015.
[16] N.-C. Hsiao et al., "A new prototype system for earthquake early warning in Taiwan," Soil Dynamics and Earthquake Engineering, vol. 31, no. 2, pp. 201-208, 2011.
[17] 辛在勤, "台灣地區地震預警之初探," 氣象學報, vol. 42, no. 2, pp. 118-134, 1998.
[18] 梁家瑋, "利用初達波特徵推估最大樓板加速度反應之研究," 碩士, 營建工程系, 國立臺灣科技大學, 台北市, 2018.
[19] P.-S. Lin and C.-T. Lee, "Ground-motion attenuation relationships for subduction-zone earthquakes in northeastern Taiwan," Bulletin of the Seismological Society of America, vol. 98, no. 1, pp. 220-240, 2008.
[20] K.-F. Ma, J. Mori, S.-J. Lee, and S.-B. Yu, "Spatial and temporal distribution of slip for the 1999 Chi-Chi, Taiwan, earthquake," Bulletin of the Seismological Society of America, vol. 91, no. 5, pp. 1069-1087, 2001.
[21] M. A. Meier, "How “good” are real‐time ground motion predictions from earthquake early warning systems?," Journal of Geophysical Research: Solid Earth, vol. 122, no. 7, pp. 5561-5577, 2017.
[22] Y.-M. Wu and H. Kanamori, "Experiment on an onsite early warning method for the Taiwan early warning system," Bulletin of the Seismological Society of America, vol. 95, no. 1, pp. 347-353, 2005.
[23] Y. M. Wu and L. Zhao, "Magnitude estimation using the first three seconds P‐wave amplitude in earthquake early warning," Geophysical research letters, vol. 33, no. 16, 2006.
[24] S. Colombelli, A. Caruso, A. Zollo, G. Festa, and H. Kanamori, "AP wave‐based, on‐site method for earthquake early warning," Geophysical Research Letters, vol. 42, no. 5, pp. 1390-1398, 2015.