坡面型土石流判釋模式與溪流型土石流潛勢評估方法-以中南部第三紀沉積岩為例

現有土石流發生潛勢之評估方法，常應用統計學、類神經網路、fuzzy、基因演算或專家評分法等技術做為評估模式。其主要基於大量數據或經驗作為工具，因此分析資料庫之良莠，影響模式之預測能力及效力。然溪流型土石流之發生，與其集水區內坡面型土石流之活動息息相關，因此若能由集水區內坡面型土石流活動潛能為指標，來評估溪流型土石流之發生潛勢，更能掌握土石流形成之因果關係。本研究應用坡面單元之概念，探討土石流溪流之地形特徵，並建立坡面型土石流判釋模式及潛勢分析方法，再進一步探討溪流型土石流之潛勢指標與分析方法。選以高屏溪及曾文溪流域沉積岩地區中共84處坡面型土石流與崩塌之案例，發展坡面型土石流判釋模式及潛勢評估方法。研究成果顯示，當以集水區坡面平均坡度、集水區平均寬度、發生區深寬比、流動段長度、流動區深寬比、崩塌潛感面積比與形狀係數比七項因子為組合時，正判率最佳，為87.5%。再將此判釋模式應用於台灣中南部9處聚落及10條溪流，共35處坡面型土石流及49處崩塌案例時，可達82.1%捕捉率。在溪流型土石流方面，本研究以高屏溪、曾文溪及濁水溪流域中共30條土石流潛勢溪流為案例，分析溪流型土石流之潛勢指標。成果顯示，以HDF單元河道平均坡度、HDF單元發生區河道平均坡度、HDF單元流動段河道平均坡度及HDF單元發生區面積比為指標的四種潛勢分析方法皆有良好之正判成效，又以集水區谷線坡度法成效最佳。集水區谷線坡度法以集水區谷線坡度在30ﾟ以上者為高潛勢，小於22ﾟ者為低潛勢，而介於兩者之間則為中潛勢。研究結果顯示，以坡面單元之微地形觀點，考慮溪流集水區內坡面型土石流之特性，可有效作為溪流型土石流之潛能評估。且本研究發展之坡面型土石流判釋模式，及溪流型土石流潛勢分析方法皆有良好的成效，可提供土石流事件之潛能評估，提前掌握土石流危險區域，降低災害之衝擊。關鍵字:坡面型土石流、溪流型土石流、判釋模式、潛勢分析方法

關鍵字

坡面型土石流；溪流型土石流；判釋模式；潛勢分析方法

並列摘要

Common occurrence potential assessment methods of debris flow torrent, often using statistics, neutral network, fuzzy, genetic algorithm, or expert scoring techniques. Primarily based on large amounts of data or experience as a tool,Therefore, the quality of the analysis database will affect the prediction results of the model.However, the occurrence of debris-flows torrents is closely related to the hillslope type debris flows in the catchment area.Therefore, if the activity potential of hillslope type debris flows in the catchment area can be used as an indicator to evaluate the potential of debris-flows torrents, it will be more likely to grasp the factors of debris flow formation. This study uses the concept of slope unit to explore the topographic characteristics of debris flow, and establishes a hillslope type debris flow recognition model and potential assessment method. Further explore the potential indicators and analysis methods of debris-flows torrent.In this study, a total of 84 hillslope type debris flows and landslides in the sedimentary rock areas of the Gaopingxi and Zengwenxi basins were used to develop hillslope type debris flow recognition model and potential assessment methods. The research results show that when combining the seven factors of the average gradient of watershed, average width of watershed, depression ratio of the initiation segment, length of the transport segment, depression ratio of the transport segment, ratio of landslide susceptibility area, and form factor ratio, the probability of correct judgment is 87.5%, is the best. When this recognition model is applied to 9 settlements and 10 streams in central and southern Taiwan, a total of 35 hillslope type debris flows and 49 landslides cases can reach a correct judgment probability of 82.1%. In terms of debris-flows torrents, this study takes a total of 30 potential debris flow torrents in the Gaoping, Zengwen and Zhuoshui streams as examples, and analyzes the potential indicators of debris-flows torrents. The results show that the four potential analysis methods using the valley line slope of the catchment area, the valley line slope of the occurrence area, the average slope of the flow section, and the area ratio of the occurrence area as indicators are all positive, and the valley line slope of the catchment area works best. The valley line slope of the catchment area takes the slope above 30ﾟ as the high potential, those below 22ﾟ as the low-potential, and the intermediate between them. The research results show that considering the characteristics of the hillslope type debris flow in the stream catchment area from the perspective of the micro-topography of the slope unit, it can be effectively used as a potential assessment of the debris-flows torrent. And the hillslope type debris flow recognition model developed in this research and the debris-flows torrent potential assessment method have good results, which can provide potential assessment of debris flow events, and grasp the danger area of debris flow in advance to further reduce the impact of disasters. Keyword: Hillslope type debris flow, debris-flow torrent, recognition model, potential assessment model.