氣候變遷可能改變侵台颱風的強度與頻率，進而影響山崩風險的時空分布特徵。本研究探討山崩風險根據IPCC的定義分成三項指標，分別是脆弱度、危害度與暴露度。選定居民為保全對象以計算暴露度，加上氣候因素與非氣候因素作為危害度與脆弱度。研究區域為曾文水庫集水區，首先使用隨機森林分類法將四次颱風事件山崩目錄整理成訓練集（賀伯颱風、海棠颱風、卡玫基颱風）和測試集（莫拉克颱風）。主要的山崩潛感因子包括坡度、坡向、高程、NDVI、河距和道路距，合併誘發因子總降雨量和最大時雨量以建立山崩潛感值模型。以莫拉克事件驗證模型，AUC為0.81，顯示模型有能力判別颱風事件所誘發的山崩。
去除誘發因子建模，可計算基礎山崩潛感值作為脆弱度。危害度計算以台灣氣候變遷推估資訊與調適知識平台AR5 GWLs升溫2度與4度動力降尺度颱風事件的降雨量資料進行分析。事件總降雨量與最大時雨量以標準分數法計算後為危害度，其在世紀中的變化率為0.4%，在世紀末為1.9%。統計人口密度與推估資料，預計人口在世紀中將下降約16.4%，在世紀末將下降34.1%，經標準分數法計算後為暴露度指標。整合以上三個指標便可計算出該地區的氣候變遷山崩風險值，風險變化率在世紀中為-0.8%，在世紀末為1.6%。
經由山崩風險評估，可知曾文水庫集水區的山崩風險在不同時期的差異不大，但空間分布具有明顯差異，當得知高風險區域後，便能優先處理該區域，以降低氣候變遷下山崩造成的威脅。

Climate change may cause changes in typhoons’ intensity and frequency, affecting the spatial and temporal characteristics of landslide risks. Following the IPCC framework, landslide risk is assessed by the combination of hazard, vulnerability, and exposure. Vulnerability refers to basic landslide susceptibility. Population density is the exposure, while maximum hourly rainfall and total rainfall in a typhoon event represent the hazard. The study area is the Zengwen Reservoir watershed. Landslide susceptibility was modeled using a Random Forest classifier trained on three typhoon-induced landslide inventories (i.e., Herb, Haitang, Kalmaegi) and validated on Morakot. Key factors include slope, aspect, elevation, NDVI, distances to rivers and roads, and total and maximum hourly rainfall. The model achieved an AUC of 0.81, indicating high predictive accuracy.
Vulnerability was estimated using a model excluding triggering factors. Hazard was derived from downscaled rainfall projections under AR5 GWLs 2°C and 4°C warming scenarios from the TCCIP (Taiwan Climate Change Projection Information and Adaptation Knowledge Platform). Standardized scores of maximum hourly rainfall and total rainfall showed hazard increases of 0.4% by mid-century and 1.9% by end-century. Exposure was quantified using projected population density, indicating a decline of 16.4% and 34.1% for the same periods, respectively. Integrating these factors, landslide risk is projected to decrease slightly by 0.8% in mid-century and increase by 1.6% by the end of the century.
Although overall landslide risk in the Zengwen Reservoir watershed shows minimal variation, spatial distribution shifts significantly, identifying high-risk areas allows for targeted mitigation to reduce landslide impacts under future climate conditions.

Broecker, W. S. (1975). Climatic change: are we on the brink of a pronounced global warming? Science, 189(4201), 460-463.
Cardona, O. D., Van Aalst, M. K., Birkmann, J., Fordham, M., Mc Gregor, G., Rosa, P., Pulwarty, R. S., Schipper, E. L. F., Sinh, B. T., & Décamps, H. (2012). Determinants of risk: exposure and vulnerability. In Managing the risks of extreme events and disasters to advance climate change adaptation: special report of the intergovernmental panel on climate change (pp. 65-108). Cambridge University Press.
ÇEllek, S. (2019). Morphological parameters causing landslides: a case study of elevation. Bulletin Of The Mineral Research and Exploration, 1-10. https://doi.org/10.19111/bulletinofmre.649758
Chung-Yi (2019). ML入門（十七）隨機森林(Random Forest). Medium. Retrieved 3/20 from
https://medium.com/chungyi/ml%E5%85%A5%E9%96%80%E5%8D%81%E4%B8%83%E9%9A%A8%E6%A9%9F%E6%A3%AE%E6%9E%97-random-forest-6afc24871857
Crozier, M. J. (2010). Deciphering the effect of climate change on landslide activity: A review. Geomorphology, 124(3-4), 260-267.
Dai, F., Lee, C. F., & Ngai, Y. Y. (2002). Landslide risk assessment and management: an overview. Engineering Geology, 64(1), 65-87.
de Almeida, L. Q., Welle, T., & Birkmann, J. (2016). Disaster risk indicators in Brazil: A proposal based on the world risk index. International journal of disaster risk reduction, 17, 251-272.
EarthExplore. (2023, 2/14). Landsat-5 image courtesy of the U.S. Geological Survey. https://earthexplorer.usgs.gov/
Esri. (2021a). Curvature function. esri. Retrieved 2/21 from https://desktop.arcgis.com/en/arcmap/latest/manage-data/raster-and-images/curvature-function.htm
Esri. (2021b). Flow Direction. Retrieved 3/20 from https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-flow-direction-works.htm
Esri. (2021c). How Aspect works. esri. Retrieved 2/21 from https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-aspect-works.htm
Esri. (2021d). How Slope works. Retrieved 2/21 from https://desktop.arcgis.com/en/arcmap/latest/tools/3d-analyst-toolbox/how-slope-works.htm
Gariano, S. L., & Guzzetti, F. (2016). Landslides in a changing climate. Earth-science reviews, 162, 227-252.
Groeneveld, R. A., & Meeden, G. (1984). Measuring skewness and kurtosis. Journal of the Royal Statistical Society Series D: The Statistician, 33(4), 391-399.
Huang, P.-K. (2023). 集成學習-Ensemble Learning. Retrieved 1/24 from https://medium.com/data-science-navigator/0%E9%9B%86%E6%88%90%E5%AD%B8%E7%BF%92-ensemble-learning-f575461889bd
IPCC. (2012) Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change [Field, C.B., V. Barros, T.F. Stocker, D. Qin, D.J. Dokken, K.L. Ebi, M.D. Mastrandrea, K.J. Mach, G.-K. Plattner, S.K. Allen, M. Tignor, and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, UK,and New York, NY, USA, 582 pp.
Lavell, A., Oppenheimer, M., Diop, C., Hess, J., Lempert, R., Li, J., & Myeong, S. (2012). Managing the risks of extreme events and disasters to advance climate change adaptation. A special report of working groups I and II of the intergovernmental panel on climate change (IPCC), 3, 25-64.
Lee, C.-T. (2017). Landslide trends under extreme climate events. TAO: Terrestrial, Atmospheric and Oceanic Sciences, 28(1), 3.
Mitchell, T. M., & Mitchell, T. M. (1997). Machine learning (Vol. 1). McGraw-hill New York.
Pradhan, B., & Lee, S. (2010). Landslide susceptibility assessment and factor effect analysis: backpropagation artificial neural networks and their comparison with frequency ratio and bivariate logistic regression modelling. Environmental Modelling & Software, 25(6), 747-759.
Qi, T., Zhao, Y., Meng, X., Chen, G., & Dijkstra, T. (2021). Ai-based susceptibility analysis of shallow landslides induced by heavy rainfall in Tianshui, China. Remote Sensing, 13(9), 1819.
Quesada-Román, A. (2021). Landslide risk index map at the municipal scale for Costa Rica. International journal of disaster risk reduction, 56, 102144.
Reichenbach, P., Rossi, M., Malamud, B. D., Mihir, M., & Guzzetti, F. (2018). A review of statistically-based landslide susceptibility models. Earth-science reviews, 180, 60-91.
Różycka, M., Migoń, P., & Michniewicz, A. (2017). Topographic Wetness Index and Terrain Ruggedness Index in geomorphic characterisation of landslide terrains, on examples from the Sudetes, SW Poland. Zeitschrift für Geomorphologie, Supplementary Issues, 61(2), 61-80. https://doi.org/10.1127/zfg_suppl/2016/0328
Sim, K. B., Lee, M. L., & Wong, S. Y. (2022). A review of landslide acceptable risk and tolerable risk. Geoenvironmental Disasters, 9(1), 3.
Simpson, N. P., Mach, K. J., Constable, A., Hess, J., Hogarth, R., Howden, M., Lawrence, J., Lempert, R. J., Muccione, V., & Mackey, B. (2021). A framework for complex climate change risk assessment. One Earth, 4(4), 489-501.
United Nations Office for Disaster Risk Reduction (UNDRR). 2017. The Sendai Framework Terminology on Disaster Risk Reduction. "Disaster risk". Accessed 11 June 2025. https://www.undrr.org/terminology/disaster-risk.
Viner, D., Ekstrom, M., Hulbert, M., Warner, N. K., Wreford, A., & Zommers, Z. (2020). Understanding the dynamic nature of risk in climate change assessments—A new starting point for discussion. Atmospheric Science Letters, 21(4), e958.
Wang, X., Wang, Y., Lin, Q., & Yang, X. (2023). Assessing Global Landslide Casualty Risk Under Moderate Climate Change Based on Multiple GCM Projections. International Journal of Disaster Risk Science, 14(5), 751-767.
Wei, L.-W., Huang, C.-M., Chen, H., Lee, C.-T., Chi, C.-C., & Chiu, C.-L. (2018). Adopting the I 3–R 24 rainfall index and landslide susceptibility for the establishment of an early warning model for rainfall-induced shallow landslides. Natural Hazards and Earth System Sciences, 18(6), 1717-1733.
Ying, X. (2019). An overview of overfitting and its solutions. Journal of physics: Conference series,
Yu, L., Zhou, C., Wang, Y., Cao, Y., & Peres, D. J. (2022). Coupling data-and knowledge-driven methods for landslide susceptibility mapping in human-modified environments: a case study from Wanzhou County, three Gorges Reservoir Area, China. Remote Sensing, 14(3), 774.
土石流及大規模崩塌防災資訊網重大災害事件統計(2024，12/30)。取自 https://246.ardswc.gov.tw/Service/OpenData
內政部20公尺網格數值地形模型資料 (2022，8/24)。取自https://data.gov.tw/dataset/35430
內政部統計處（2012）。統計區分類系統建置及應用。
內政部統計處社會經濟資料服務平台 (2017)。111年6月南投縣、嘉義縣、嘉義市、高雄市統計區人口指標_最小統計區。
王筱媛、李明旭、董家鈞、李錫堤（2024）。氣候變遷情境颱風事件誘發山崩風險評估：以曾文水庫集水區為例。中華民國地質學會與中華民國地球物理學會113年年會暨學術研討會。
朱芳儀、李欣輯、林丞庭、鄭兆尊、姜欣妤、林銘郎（2022）。未來升溫情境下之坡地衝擊工具研發與圖資建置。臺灣氣候變遷推估資訊與調適知識平台計畫。
朱芳儀、林士堯（2024）。AR5 坡地危害圖資資料生產履歷(2.1版)。
吳淵洵、周南山（2006）。台灣山區道路邊坡災害及搶修處理工法之探討。臺灣公路工程第32卷第12期，2-32。
何謹余（2021）。災害風險減輕與氣候變遷調適融合框架於社區韌性調適能力建構。國立臺灣大學生物環境系統工程學系學位論文，2021，1-95。
何謹余、童慶斌、王涴雅 (2020)。減災與氣候變遷調適在區域治理上之融合研究─以坡地災害為例。農業工程學報，66 (2)。
林彥廷、顏筱穎、張乃軒、林宏明、韓仁毓、楊國鑫、陳俊杉、鄭宏逵、徐若堯（2021）。結合時空因子與 insar 觀測資料之地表崩塌變位預測分析。中國土木水利工程學刊，33（2），93-104。
林昭遠、藍欣茹（2015）。山區道路與小林村崩塌關係之探討。中華水土保持學報，47（1），1287-1296。
林思穎，林士堯（2023）。 AR5動力降尺度颱風說明文件(3.5版)。[2025/1/15]， 取自臺灣氣候變遷推估資訊與調適知識平台https://tccip.ncdr.nat.gov.tw/upload/data_document/20200219113342.pdf
林傑斌、林川雄、劉明德（2004）。Spss 12 統計建模與應用實務。台北: 博碩文化。
柯明淳、黃明偉、林聖琪（2016）。地震山崩快速潛勢評估建立與應用。土木水利，43（3），17-26。
洪明瑞（2007）。Ahp 評選邊坡穩定生態工法之可行性研究。明志學報 第 38 卷第二期。
倢愷Oscar（2021）。不要再做one hot encoding！！。https://axk51013.medium.com/%E4%B8%8D%E8%A6%81%E5%86%8D%E5%81%9Aone-hot-encoding-b5126d3f8a63
國科會臺灣氣候變遷推估資訊與調適知識平台(2023a)。Level 4台灣歷史重建時資料。https://tccip.ncdr.nat.gov.tw/。
國科會臺灣氣候變遷推估資訊與調適知識平台(2023b)。AR5颱風降尺度偏差修正- 全部 - 曾文溪。https://tccip.ncr.nat.gov.tw/。
國家災害防救科技中心（2020）。災害風險概念介紹。(2025/6/9)。https://dra.ncdr.nat.gov.tw/Frontend/Disaster/RiskDetail/BAL0000012。
國家發展委員會（2022）。中華民國人口推估(2022年至2070年)。
許晃雄、王嘉琪、陳正達、李明旭、詹士樑（2024）。國家氣候變遷科學報告2024：現象、衝擊與調適 [許晃雄、李明旭 主編]。國家科學及技術委員會與環境部聯合出版。
陳振宇（2023）。30年見證滄海桑田－消失的新好茶村。https://photo.ardswc.gov.tw/PhotoStory/Home/Story?StoryID=f2f014a1-629f-4ab2-9022-689d506658d2
陳榮河、洪勇善（1999）。崩塌地整治工法之介紹。地工技術，（2），5-12。
陳韻如、林宣汝、陳永明（2017）。氣候變遷下災害風險圖問答集。國家災害防救科技中心。
陳韻如、林宣汝、劉俊志、童裕翔、陳永明（2024）。坡地災害氣候變遷風險圖研發與應用。臺灣氣候變遷推估資訊與調適知識平台計畫。
陸象豫、孫銘源（2021）。植生覆蓋對邊坡穩定的作用。林業研究專訊 Vol. 28 No. 6，60-63。
費立沅（2009）。臺灣坡地災害與地質敏感區的關係。地質專題，16-22。
楊善智、曾義星（2007）。數值高程模型 (dem) 之品質評估。國立成功大學測量及空間資訊學系研究所學位論文。
農業部農村發展及水土保持署（2006）。重大災害事件統計。https://246.ardswc.gov.tw/Achievement/StatictiesDisasters
農業部農村發展及水土保持署（2023）。BigGIS空間資訊系統2006年、2009年SPOT全島影像。
經濟部水利署（2023）。經濟部水利署南區水資源分署環境教育。https://web.wra.gov.tw/swreeis/cp.aspx?n=8135
蔡雨澄（2012）。極端降雨下之山崩潛感分析-以莫拉克颱風誘發山崩為例。中央大學應用地質研究所學位論文。
簡逢助（2015）。山崩潛感值暨降雨量與崩壞比之關係探討。中央大學應用地質研究所學位論文。
譚志豪、嬑陳璇、冀樹勇、蘇泰維、李錦發、費立沅（2009）。降雨誘發山崩災害之預警系統初步研究 海峽兩岸地工技術∕岩土工程交流研討會。