台灣由於地理環境與氣候條件的關係，土石流發生頻繁，因此發展土石流預警系統，提高預測準確率，落實防災避難疏散，以減少土石流災害的損失，為當前土石流防治工作的重要目標。形成土石流的基本條件包含豐富的鬆散土石、陡峻的坡度和充足的水源。對某一個集水區內的土石流潛勢溪流而言，該集水區內的地文條件在一般情形下變化不會太大，主要的變化是降雨事件，因此掌握降雨事件的雨量變化可以做為土石流的預測基準。目前國內土石流降雨警戒基準值的訂定係採用降雨強度及有效累積雨量做為評估指標，然而有效累積雨量的計算方式可能有數種不同的計算方法，本研究想探討這些不同的計算方式對於有效累積雨量計算結果之差異及其對判別土石流警戒時間之影響。
本研究以高屏地區為研究對象，考慮四種有效前期降雨的計算方式，其中方法A、方法B及方法C均考量七天前期降雨及24小時衰減係數α，只是在前期降雨的起算時間方式略有不同；方法D則以半衰期的概念來計算衰減係數。本研究採用2008年卡玫基、2009年莫拉克及2010年凡那比等三場颱風所帶來的降雨事件進行分析比較，結果顯示方法A、方法B及方法C所得知有效累積雨量在降雨初期24小時內相當接近，沒有顯著之差異，但是當時間超過24小時後，由於方法A固定前期降雨的起算時間，降雨時間超過24小時後仍視為當場降雨，因此所得之有效累積雨量較大，會提早到達土石流降雨警戒基準值。方法B的計算方式是在每24小時內採用固定前期降雨的起算時間，但每滿24小時的時候，前期降雨的起算時間往後順延24小時，因此每滿24小時因為衰減係數α的關係有效累積雨量計算結果會有些微下墜現象，因此所得之有效累積雨量會略低於方法A的計算結果。方法C沒有固定前期降雨的起算時間，而是讓它隨著時間每小時往後順延，所得的有效累積雨量在時間方面較具連續性。。方法D具有的半衰期的概念，半衰期T的選擇決定衰減係數的大小，在T值越大時衰減係數的變化越小。當半衰期T取72小時，方法D的計算結果和方法C (α取0.7)的計算結果相當接近。
在預警時間分析上，對於新發雨量站在卡玫基颱風和莫拉克颱風兩場降雨事件中，按照方法C和方法D (T=72小時) 計算所得的有效累積雨量分析降雨預警時間晚於實際土石流發生時間；此說明單純按照有效累積雨量進行土石流預警時間分析是不夠的。當把降雨強度也加入考量時，本研究以降雨驅動指標(RTI =有效累積雨量降雨強度)重新分析，發現加入降雨強度的考量後比較容易掌握土石流發生降雨警戒時間。

With steep topography, special geological conditions, typhoon seasons, and over developed hillsides, Taiwan frequently suffered debris flow hazards. Debris flows caused by typhoon Morakot in 2009 resulted in many fatalities, missing, and seriously economic loss in Taiwan. Therefore, it will be necessary to enhance public awareness of many debris-flow hazards and educate people how to react to these hazards. A rainfall-based debris-flow warning system is needed for preventing or mitigating debris-flow hazards. The basic conditions for debris-flow occurrence are the abundant loose soils, steep slope, and the large amount of water. For a specified watershed of a potential debris-flow torrential, the changes of the topographical and geological conditions as well as the loose soil conditions in a period of normal time are negligible compared with the change of rainfall. The rainfall variability played a main inducement to result in hazardous debris flow. The rainfall intensity I and the effective rainfall accumulation are the parameters used for establishing the warning debris flow warning index in Taiwan. This research aims to discuss the effect on the warning time for debris flow occurrence due to different methods used to evaluate the effective accumulated rainfall.
Four methods were used to calculate the effective accumulated rainfall in this study. This methods were applied to the rainfall events brought by three severe typhoons, such as Kalmaegi in 2008, Morakot in 2009, and Fanapi in 2010) to assess the critical time for the warning of debris flow in the Gaoping watershed. These four methods were named as A, B, C, and D. All methods A, B, and C consider 7-day antecedent rainfall and daily reduction weighting factor α, but they have different methods to calculate 7-day antecedent rainfall. Method D considers hourly reduction weight factor with an idea of half-life time. The results shows that the effective accumulated rainfall calculated by the Method A has larger amount and this will result in earlier reaching to warning criteria. The initial time for considering the rainfall as a previous rainfall is changed for each 24 hours in Method B. Therefore, the effective accumulated rainfall calculated by the method B has a little drop for each 24 hour. The initial time for considering the rainfall as a previous rainfall is changed for each hour in Method C. Method D has a consideration of half-life time in assessing weighting factor, and the value of weighting factor is changed for each hour. The higher half-life time T the smaller weighting factor in Method D. For T = 72 hours, the effective accumulated rainfall calculated by the Method D is closed to that by Method C.
Based on the rainfall data in the Xinfa rainfall station for the events of Typhoons Kalmaegi and Morakot, the effective accumulated rainfall calculated by Methods C and D (T=72 hours) failed to play the role of debris-flow warning. This implies that a single rainfall parameter such as the effective accumulated rainfall is not enough to play the role of debris-flow warning. The rainfall intensity is taken into considered for debris flow warning, by using the rainfall triggering index (RTI) that is the product of the effective accumulated rainfall and the rainfall intensity. The warning time can be earlier to the real time of debris flow occurrence when the rainfall intensity is taken consideration with effective accumulated rainfall.

1. 曾奕超 (2004)，「土石流發生降雨地文綜合警戒指標之研究」，國立成功大學水利及海洋工程研究所碩士論文
2. 張智瑜 (2005)，「地文條件對土石流發生降雨警戒指標之影響」，國立成功大學水利及海洋工程研究所碩士論文
3. 黃文舜 (2012)，「降雨變遷及地震因素對土石流發生影響之研究」，國立成功大學水利及海洋工程研究所博士論文
4. 連惠邦 (2014)，「降雨變遷及地震因素對土石流發生影響之研究」，私立逢甲大學水利工程與資源保育研究所碩士論文
5. 黃婷卉 (2002)，「土石流發生降雨特性之研究-以陳有蘭溪流域為主」，國立成功大學水利及海洋工程研究所碩士論文
6. 李明熹 (2006)，「土石流發生降雨警戒分析及其應用」，國立成功大學水利及海洋工程研究所博士論文
7. 徐永明 (2012)，「河階地危險性評估-以陳有蘭溪流域為例」，朝陽科技大學營建工程系碩士班碩士論文
8. 江永哲與林啟源（1991），「土石流之發生雨量特性分析」，中華水土保
持學報，第22 卷，第2 期，第21-37 頁
9. 謝正倫、陸源忠、游保杉、陳禮人 (1995)，「土石流發生臨界降雨線設定方法之研究」，中華水土保持學報，第26 卷，第3 期，第167~172 頁
10. 范正成 (1996)，「土石流防災與監測之研究-雨量分析、降雨預報應用於土石流預警(一)」，行政院國家科學委員會專題研究計畫成果報告，計畫編號：NSC85-2621-P-002-052。
11. 范正成 (1997)，「土石流防災與監測之研究-雨量分析、降雨預報應用於土石流預警(二)」，行政院國家科學委員會專題研究計畫成果報告，計畫編號：NSC86-2621-P-002-022。
12. 范正成、姚正松 (1997)，「台灣東部地區土石流發生的水文及地文條件之初步研究」，八十六年度農業工程研討會，台南市，第525-531 頁。
13. 范正成（1998），「土石流防災與監測之研究-雨量分析、降雨預報應用於土石流預警（ 三）」行政院國家科學委員會專題研究計畫成果報告，NSC87-2621-P-002-039。
14. 范正成、吳明峰、彭光宗(1999)，「豐丘土石流發生降雨臨界線之研究」，地工技術，第74 卷，第39-46 頁。
15. 范正成、吳明峰(2001)，「一級溪流土石流危險因子及其與臨界降雨線之關係」，中華水土保持學報，第32 卷，第3 期，第227-234 頁。
16. 范正成、劉哲欣、吳明峰(2002)，「南投地區土石流發生臨界降雨線之設定及其於集集大地震後之修正」，中華水土保持學報，第33 卷，第1 期，第31-38 頁。
17. 李明熹、詹錢登 (2004)，「土石流發生降雨警戒模式」，中華水土保持學報，第35 卷，第3 期，第275~285 頁
18. 姚善文（2000），「土石流發生之水文特性探討」，中央大學土木工程系碩士論文。
19. 羅文俊、沈哲緯、蕭震洋、辜炳寰、曹鼎志、鄭錦桐 (2014)，「運用隨機森林探討莫拉克颱風災區土石流發生因子關連性」，災害防救科技與管理，第 3 卷，第 1 期，第 41~ 67 頁
20. Caine, N, (1980), “The Rainfall Intensity Duration Control of Shallow Landslides and Debris Flows.” Geografiska Annaler
21. Cannon, S.H. and S.D. Ellen (1985), “Rainfall Conditions for Abundant Debris Avalanches in San Francisco Bay Region California,” California Geology
22. Wieczorek , G.F. (1987), “Effect of Rainfall Intensity and Duration on Debris Flows in Central Santa Cruz Moutains,” California, Flows/ Avalanches : Process, Recognition and Mitigation, Geological Society of America, Reviews in Engineering Geology
23. Johnson, A. M. and J. R. Rodine (1984), “Debris Flow”, Slope Instability, Edited by Brunsden and Prior, John Wiley and Sons Ltd. Chapter 8, pp.275-361.
24. Keefer, D. K., R. C. Wilson, R. K. Mark, E. E. Brabb, W. M. Brown III, S. D. Ellen, E. L. Harp, G. F. Wieczorek, C. S. Alger, and R. S. Zatkin(1987), “Real-Time Landslide Warning During Heavy Rainfall”, Science, 238: 921-925.
25. Liu Xilin (2002), ”Empirical Assessment of Debris Flow Risk on a Regional Scale in Yunnan Province, Southwestern China”, Environmental Management, 30 (2): 249-264.
26. Scott A. Anderson, and Nicholas Sitar (1995), “Analysis of rainfall-induced debris flow”, Journal of Geotechnical Engineering, 121 (7).
27. Sitar, N., S.A. Anderson, and K.A. Johnson (1992), “Conditions for Initiation of Rainfall-Induced Debris Flow”, Stability and performance of slopes and embankments: Proceedings of a Special Conference at U. C. Berkeley, ASCE.
28. Takahashi, T. (1974), “Debris Flow on Prismatic Open Channel＂, Journal of the Hydraulics Division, 106 (HY3): 381-396.
29. Takahashi, T. (1978), “Mechanical Characteristics of Debris Flow＂, Journal of the Hydraulics Division, 104 (HY8): 1153-1169.
30. Takahashi, T. (1981a), “ Estimation of Potential Debris Flows and their hazardous zones,” Journal of Natural Disaster science, 3 (1): 57-89.
31. Takahashi, T. (1981b), “ Debris Flow,” Ann. Rev. Fluid Mech., 13: 57-77.
32. Takahashi, T. (1991), Debris Flow, Monograph series of IAHR, Balkema.
33. Terzaghi, K., R. B. Peak, and G. Mesri (1996), Soil Mechanics in Engineering Practice, John Wiley &Sons, Inc., pp.36-37, N. Y.
34. Tsaparas, I., Rahardjo, H., Toll, D., and Leong, E. (2002). Controlling parameters for rainfall-induced landslides. Computer and geotechnics , 29, 1-27.