土石流潛勢溪流評估模式應用於不同時期不同區域之研究─以濁水溪、烏溪和高屏溪(荖濃溪、濁口溪、隘寮溪)流域為例

Study for Potential Debris Flow Torrent Assessment Model Applied to Different Time And Different Regions─ Chuo-shui River, Wu River And Gaoping River (Lao-nong River, Zhuo-kou River, Ai-liao River) Basin As Examples

蘇莘(Hsin Su)

指導教授 ： 劉家男 ； 馮智偉

國立暨南國際大學/科技學院/土木工程學系/碩士(2016年)

摘要

台灣土石流災害由來已久，然而土石流之好發性主要受地形與土石堆積物所影響，且土石堆積物又受到集水區所在地質區位及岩層種類而有不同，進而造成土石流啟動之動能條件之差異。本研究首先將烏溪、濁水溪、高屏溪流域小集水區岩石種類進行分類，並彙整三個地文因子包含：地形因子(相對高度、溪床坡度、邊坡坡度比)、流量因子(集水面積、主流長度、形狀係數、水系密度)、材料因子(殘土率、颱風事件前後崩塌地面積及崩塌面積比)，以及雨量誘發因子(RTI值)等因子，由桃芝颱風事件資料以邏輯式迴歸建立全因子土石流判別模式，再由敏督利颱風驗證模式正確性。第二步進行因子相關性分析，找出獨立因子，並探討不同岩石分區之獨立因子特性，再依不同岩石分區之三類地文因子中，選出獨立因子，並加入水文因子進行邏輯式迴歸，建立特徵因子土石流判別模式。最後，針對全因子與特徵因子兩項土石流判別模式進行分析。其中，沉積岩區以溪床坡度因子與集水面積相關係數最低，其餘皆較高，板岩區則以邊坡坡度比與其他地文因子參數相關性最低。因此本研究沉積岩區特徵因子採用溪床坡度因子、集水面積、事件前崩塌面積比與RTI值等4項，板岩區則採用相對高度、溪床坡度之外8項因子。成果顯示沉積岩區全因子與特徵因子土石流判別模式為中等一致性，正確性亦在87.4%以上，板岩區全因子與特徵因子土石流判別模式為幾乎完全一致性，正確性亦在96%以上。且兩種不同岩石種類之地文因子特性不同，影響土石流判別模式精度，建議需考量不同岩石種類進行研究。
最後本研究採用三個流域資料以邏輯式迴歸建立通用土石流判別模式，結果顯示通用土石流判別模式仍具高準度預測能力，後續設定發生門檻值及假設最大降雨強度，得到累積雨量值以建立小集水區警戒累積雨量。

關鍵字

土石流判別模式 ； 邏輯式迴歸 ； 地文因子參數相關性

並列摘要

It has been a long time that debris flows occurs in Taiwan. But the predilection of debris flows is mainly affected by the terrain and the rock materials. What’s more, the rock materials are also different which affected by the geological location of catchments and the rock types, and then bring out the differneces of the kinetic energy conditions of the debris flows’ occurences. First, to classified the rock types of the small catchment of Chuo-shui river, Wu river and Gaoping river basins, and aggregated the three physiographic factors, including: topographic factors (relative height, channel-bed slope and slope ratio), the flow factors (watershed area, the mainstream length, form coefficient and drainage density), material factosrs (hypsometric curve, landslide areas before and after typhoon events and the ratio of landslide areas before and after typhoon events), as well as rain-induced factor (RTI value) and other factors. Establishing to form the full factorial determination modes of debis flows from the datas of Typhoon Toraji event by logic regression formula, then verifing the justification of the modes by Typhoon Mindulle event. The second step of factor correlation analysis is to identify independent factors, and to explore the characteristics of the independent factors in the different rock f partition. To select independent factors in three physiographic factors in different rock partitions, and to add the hydrological factors into logical regression formula, then model the debris flow determination mode of characteristic factors. At the last, to analyze the debris flow determination modes of the full factorial factors and features factors. According the analysis, the correlation coefficient between the ratio of channel-bed slope factor and watershed area factor is the lowest in the sedimentary district, and the others are all higher. And the correlation coefficient between the slope ratio and other physiographic factors is the lowest in the slate district. Therefore, 4 factors considering the channel-bed slope, watershed area, landslide areas before typhoon events and RTI value as the features factors in the sedimentary district. 8 factors considering except the relative height and channel-bed slope in the slate district. According the final result, the full factors and characteristic factors of the debris flow determination mode are medium consistent in the sedimentary district, the accuracy is more than 87.4%. And the full factors and characteristic factors of the debris flow determination mode are almost exclusively consistent in the slate district, the accuracy is also more than 96%. The differences of physiographic factors characteristics between the two different rock types affect the accuracy of the debris flow determination mode. So the author recommended to consider the necessary to study the different rock types.
Finally, this research used the datas of the three basins to establish a common debris flow determination mode by logical regression formula, and it showed that the common debris flow determination mode is provied with high-accuracy prediction. Then setting the threshold of the subsequent occurrence and assuming the maximum intensity of rainfall, to establish the alert accumulated rainfall of the small catchment by the accumulated rainfall values.

並列關鍵字

debris flow determination mode ； logical regression formula ； the correlation coefficient of the physiographic factors parameter

參考文獻

1. Agresti, (2002),”A. Categorical Data Analysis (2nd ed.)”, New York: John Wiley, p.p.710

2. Alberto Carrara, Giovanni Crosta, Paolo Frattini, (2008),"Comparing models of debris-flow susceptibility", Geomorphology,Vol.94,p.p.353-378

3. Aleotti, P.; Chowdhury, (1999),” Landslide hazard assessment: summary review and new perspectives”, Bulletin of Engineering Geology and the Environment,Vol.58,p.p.21-44

4. Alfredo H.A. and Wilson H.T., (1993),"Probability Concepts in Engineering Planning and Design",Engineering Geology,Vol.33,pp.305-321

5. Aykut,Akgun., Serhat,Dag and Bulut,Fikri,(2008) ”Landslide susceptibility mapping for a landslide-prone area (Findikli, NE of Turkey) by likelihood-frequency ratio and weighted linear combination models”