自1940年代起，隨著全球氣溫逐年升高，災害發生頻率也顯著上升。重大災害事件數從每年約20起增加至每年超過400起，顯示全球均溫上升驅動了更多致災因子，直接反映在災害事件數量上及次生災害之比率。劇烈天氣事件能改變近地表地質特性、觸發地質災害事件如地層下陷、土壤液化、坡地災害等，由於這些地質災害事件可反應在地表變形資料上，如何利用環境資料（包含大氣類、地下水、潮汐等資料）預測地表變形資料（GNSS和地震資料）及坡地災害事件？這幾個問題驅動了我們對「天然災害鏈預測」可行性評估的動機。本研究的主要工作目標為利用長短期記憶模型（LSTM模型）及支持向量迴歸（SVR模型）之演算法，建立地下水位、地表變形及坡地災害之預測模型。首先我們利用2004年至2020年潮州地區兩個自動氣象站所提供的氣溫、雨量、風向、風速，以及目標預測測站之歷史地下水位，預測十二個地下水測站的地下水位，比較LSTM模型和SVR模型的表現，結果顯示，大部分地下水測站的LSTM模型在測試集上的平均決定係數高達0.90。其次，我們應用相同方法，探討2013年至2020年垂直地表變形預測，新加入了氣壓、相對濕度以及潮位資料，發現GNSS測站的LSTM模型在測試集上的平均決定係數達0.94，而更高採樣點率寬頻地震站位移場的平均決定係數達0.89，不同測試皆證實LSTM較SVR模型預測表現更佳。最後，我們以農業部農村發展及水土保持署的坡地災害事件目錄為基礎，評估了不同年份組成的四個資料集。由2010年至2012年的訓練集以及2013年至2014年的測試集組成的資料集表現最佳，顯示了環境資料在坡地災害預測中的潛力，準確率達0.83，精準率為0.95，召回率為0.67。綜合以上， LSTM模型展示了對於時序資料強大的預測能力，並強調了環境資料在地表變形及坡地災害預測中的關鍵作用。本研究並測試六個不同的氣象參數在預測模型之貢獻度，依照重要程度依序如下：氣溫、氣壓、風速、風向、相對濕度、雨量，然而同時考慮此六個氣象參數的模型，其預測效能仍優於單一氣象參數。本研究具體提供了氣象與地質災害間的預測方法論，期能在未來用於近即時警報/預報、並為未來政策制定提供即時參考依據。

Since the 1940s, global temperatures have been gradually rising, correlating with a significant increase in the frequency of nature disasters. The number of major disasters has increased from ~20 per year to over 400 annually, indicating the impact of global warming. Extreme weather events have been discovered to significantly alter near-surface geological properties, leading to geological hazards including land subsidence, soil liquefaction, landslide, slopeland, and so on. The main objective of this study is to develop predictive models of groundwater levels, surface deformation, and slopeland disasters in southwestern Taiwan near Chaozhou. The study area is chosen due to the high subsidence rate and low seismicity. We attempt to establish the possibility of predicting surface deformation data (GNSS and seismic data) and geological hazards using environmental data (i.e., atmospheric, groundwater, and tidal data) and machine learning approaches (Long Short-Term Memory, LSTM and Support Vector Regression, SVR). In the study period of 2004 to 2020, this study initially utilized temperature, rainfall, wind direction, and wind speed data from two automatic weather stations in the study area to predict groundwater levels at twelve various groundwater stations. The resulting prediction performance (averaged coefficient of determination) reached 0.90 at most of groundwater stations using LSTM model. We next applied the same method to explore possibility of vertical surface deformation prediction incorporating additional data such as air pressure, relative humidity, and tide levels. We found that at the targeted GNSS stations, the averaged coefficient of determination up to 0.94. At the broadband seismometer displacement field characterized by much finer time resolution, the coefficient of determination reached 0.89. Various tests confirmed that the LSTM model outperformed the SVR model in prediction accuracy. Finally, we used four sets of environmental data from a variety of data period to predict slopeland disasters. We found that as long as the particular years experienced extreme landslide events were excluded in the training data, the high prediction performance can be reached. The best model reveals the accuracy of 0.83, precision of 0.95, and recall of 0.67. In conclusion, LSTM models showed robust predictive capabilities for time-series data that highlights the pivotal role of environmental data in forecasting surface deformation and landslide events. In the future, developing predictive models for various geological hazard types can be expected with the hope of offering timely warnings and predictions for natural disaster prevention.

Afzaal, H., Farooque, A. A., Abbas, F., Acharya, B., & Esau, T. (2019). Groundwater estimation from major physical hydrology components using artificial neural networks and deep learning. *Water, 12*(1), 5.
Biq, C. C. (1972). Dual-trench structure in Taiwan–Luzon region. *Proc. Geol. Soc. China, 15*, 65–75.
Bonilla, M. G. (1977). Summary of Quaternary faulting and elevation changes in Taiwan. *Mem. Geol. Soc. China, 2*, 43–56.
Chan, Y. C., Hu, J. C., Shen, L. C., Chen, R. F., Rau, R. J., Chen, K. H., ... & Nien, P. F. (2007). Fault activity and lateral extrusion inferred from velocity field revealed by GPS measurements in the Pingtung area of southwestern Taiwan. *Journal of Asian Earth Sciences*. doi:10.1016/j.jseaes.2006.07.020.
Chiang, S. C. (1971). Seismic study of the Chaochou structure, Taiwan. *Pet. Geol. Taiwan, 8*, 281–294.
Ching, K. E., Rau, R. J., Lee, J. C., & Hu, J. C. (2007). Contemporary deformation of tectonic escape in SW Taiwan from GPS observations, 1995–2005. Earth and Planetary Science Letters, 262(3-4), 601-619.
Cutter, S. L., Ismail-Zadeh, A., Alcántara-Ayala, I., Altan, O., Baker, D. N., Briceño, S., ... & Wu, G. (2015). Global risks: Pool knowledge to stem losses from disasters. Nature, 522(7556), 277-279.
Dilley, M. (2005). *Natural disaster hotspots: a global risk analysis* (Vol. 5). World Bank Publications.
Gao, W., Li, Z., Chen, Q., Jiang, W., & Feng, Y. (2022). Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches. *Journal of Geodesy, 96*(10), 71.
Heflin, M., Donnellan, A., Parker, J., Lyzenga, G., Moore, A., Ludwig, L. G., Rundle, J., Wang, J., & Pierce, M. (2020). Automated estimation and tools to extract positions, velocities, breaks, and seasonal terms from daily GNSS measurements: illuminating nonlinear salton trough deformation. *Earth and Space Science, 7*(7), e2019EA000644.
Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735
Hooker, S., Erhan, D., Kindermans, P. J., & Kim, B. (2018). Evaluating feature importance estimates. arXiv preprint arXiv:1806.10758, 2.
Hu, J. C., Angelier, J., & Yu, S. B. (1997). An interpretation of the active deformation of southern Taiwan based on numerical simulation and GPS studies. Tectonophysics, 274(1-3), 145-169.
Hu, J. C., Yu, S. B., Angelier, J., & Chu, H. T. (2001). Active deformation of Taiwan from GPS measurements and numerical simulations. Journal of Geophysical Research: Solid Earth, 106(B2), 2265-2280.
Hu, J.C., Hou, C.S., Shen, L.C., Chan, Y.C., Chen, R.F., Huang, C., Rau, R.J., Chen, K.H., Lin, C.W., Huang, M.H., Nien, P.F., 2007. Fault activity and lateral extrusion inferred from velocity field revealed by GPS measurements in the Pingtung area of southwest- ern Taiwan. J. Asian Earth Sci. doi:10.1016/j.jseaes.2006.07.020.
Jan, C. D., Lee, M. H., & Huang, T. H. (2002). Rainfall threshold criterion for debris flow initiation.
Lewis, C., Chen, S. W., & Yen, P. C. (2004). Magnetic surveying of the Chaochou fault of southern Taiwan: culmination of basement-involved surface thrusting in arc–continent collision. *International Geology Review, 46*, 399–408.
Liu, Y. H., Yeh, T. C., Chen, K. H., Chen, Y., Yen, Y. Y., & Yen, H. Y. (2019). Investigation of single‐station classification for short tectonic tremor in Taiwan. Journal of Geophysical Research: Solid Earth, 124(8), 8803-8822.
Malet, J. P., Maquaire, O., & Calais, E. (2002). The use of Global Positioning System techniques for the continuous monitoring of landslides: application to the Super-Sauze earthflow (Alpes-de-Haute-Provence, France). Geomorphology, 43(1-2), 33-54.
Miao, Y., Shi, Y., & Wang, S. Y. (2018). Temporal change of near-surface shear wave velocity associated with rainfall in Northeast Honshu, Japan. *Earth, Planets and Space, 70*, 1-11.
Moss, J., McGuire, B., & Gilman, J. (1997). UNDER TTHE VOLCANO MEASURING DEFORMATION ON ITALY'S MOUNT ETNA. GPS World, 8(4), 22-33.
Petit, G. (2010). *IERS conventions* (2010).
Petit, G., & Luzum, B. (2010). *IERS conventions*. Technical report, Bureau International des Poids et Mesures, Sevres (France).
Puskas CM, Meertens CM, Phillips D (2017) Hydrologic loading model displacements from the national and global data assimilation systems (NLDAS and GLDAS). UNAVCO Geodetic Data Service Group
Shirzaei, M., Freymueller, J., Törnqvist, T. E., Galloway, D. L., Dura, T., & Minderhoud, P. S. (2021). Measuring, modelling and projecting coastal land subsidence. Nature Reviews Earth & Environment, 2(1), 40-58.
Singh, V. V., Biskupek, L., Müller, J., & Zhang, M. (2021). Impact of non-tidal station loading in LLR. *Advances in Space Research, 67*(12), 3925–3941.
Steer, P., Jeandet, L., Cubas, N., Marc, O., Meunier, P., Simoes, M., ... & Hovius, N. (2020). Earthquake statistics changed by typhoon-driven erosion. *Scientific Reports, 10*(1), 10899.
Sun, J., Hu, L., Li, D., Sun, K., & Yang, Z. (2022). Data-driven models for accurate groundwater level prediction and their practical significance in groundwater management. Journal of Hydrology, 608, 127630.
Take, W. A., Beddoe, R. A., Davoodi-Bilesavar, R., & Phillips, R. (2015). Effect of antecedent groundwater conditions on the triggering of static liquefaction landslides. Landslides, 12, 469-479.
Tóth, J. (1970). A conceptual model of the groundwater regime and the hydrogeologic environment. *Journal of Hydrology, 10*(2), 164-176.
Vapnik, V., & Chervonenkis, A. (1963). A note on class of perceptrons. Automation and Remote Control, 25, 103-104.
Vapnik, V. N., Golowich, S. E., & Smola, A. (1997). Support vector method for function approximation, regression estimation, and signal processing. In Advances in Neural Information Processing Systems (NIPS), 281-287.
Wright, S. (1921). Correlation and causation. Journal of agricultural research, 20(7), 557.
Yan, H. M., Chen, W., Zhu, Y. Z., Zhang, W. M., Zhong, M., & Liu, G. Y. (2010). Thermal effects on vertical displacement of GPS stations in China. *Chinese Journal of Geophysics, 53*(2), 252–260.
Zhai, Q., Peng, Z., Chuang, L. Y., Wu, Y.-M., Hsu, Y.-J., & Wdowinski, S. (2021). Investigating the impacts of a wet typhoon on microseismicity: A case study of the 2009 typhoon Morakot in Taiwan based on a template matching catalog. *Journal of Geophysical Research: Solid Earth, 126*, e2021JB023026. https://doi.org/10.1029/2021JB023026
行政院. (2022). 災害防救白皮書. 行政院.
經濟部水資源局. (1999). 臺灣地區地下水觀測網整體計畫第一期(81∼87年度)成果彙編.
經濟部水利署.(2011). 應用資料同化法推估地下水抽水量. 水利署電子報第 0139 期. 經濟部.
經濟部水利署.(2015). 中華民國 100 年 水利統計. 經濟部.
經濟部水利署. (2023). 地層下陷狀況. 經濟部水利署. 取自https://www.wra.gov.tw/cp.aspx?n=3679
洪偉嘉, & 黃金維. (2008). 應用多重感應器監測雲林地區三維變形 (Doctoral dissertation).
洪如江. (2009). 坡地災害防治. 水利土木科技資訊季刊, 46, 7-15.
簡俊彥. (1987). 台灣沿海地區地層下陷問題. 地工技術雜誌, 20, 50-56.
柯佳宏. (2019). 屏東大潮州地下水人工湖之補注池水文地質特性調查之研究 (Master's thesis, 屏東科技大學土木工程系所), 1-74.
邱正鈞, 萬昱廷, & 王士榮. (2020). 以倒傳遞類神經網路進行地下水位與抽水量推估. Journal of Taiwan Agricultural Engineering, 66(2).
蔡存孝, & 謝勝信. (2020). LSTM於屏東平原地層下陷區地下水位預測模型研究. Journal of Taiwan Agricultural Engineering, 66(2).
陳宏宇. (2000). 台灣山崩之工程地質特性. 地工技術, (79), 59-70. https://doi.org/10.30140/SG.200006.0005
陳振宇, 陳均維, 陳國威, & 林詠喬. (2019). 坡地降雨致災熱區警戒模式. Journal of Chinese Soil and Water Conservation, 50(1), 1-10.
陳樹群, 蔡喬文, 陳振宇, & 陳美珍. (2013). 筒狀模式之土壤雨量指數應用於土石流防災警戒. Journal of Chinese Soil and Water Conservation, 44(2), 131-143.
徐享崑, 劉豐壽, & 鄭昌奇. (1995). 臺灣地區地層下陷之現況, 成因與對策. 台灣水利, 43(3), 19-29.
吳怡瑩, 劉哲欣, & 張志新. (2013). 降雨量與表層土壤含水量關係之研究. 社團法人中華水土保持學會102年度年會.