臺灣山區地勢高峻、地質破碎，加以氣候變遷導致極端降雨頻率持續增加，山區土石流災害已成為重大安全課題。2009 年莫拉克颱風期間，高雄市六龜區新開部落遭受土石流衝擊，為台灣近年最具指標性的土石流致災案例之一。數值模擬為重建歷史災害過程、評估潛在危害潛勢之核心工具，如何在僅有靜態災後足跡的條件下，對模型進行科學且客觀的流變參數率定，為防災科學之關鍵課題。
本研究採用開源質量流模型 r.avaflow 單相流模式，以高雄市六龜區編號 DF075 土石流潛勢溪流之莫拉克颱風事件作為研究案例，進行數值模擬與空間統計驗證。研究針對 Voellmy 流變模型之核心摩擦參數——底床摩擦角（ϕB）與紊流摩擦係數（ξ）——以 5 × 5 組合執行交叉參數敏感度分析（ 25 組），並以臨界成功指數（CSI）與 ROC 曲線下面積（AUROC）作為空間量化驗證指標進行參數率定。
交叉敏感度分析揭示 ϕB 與 ξ 之間具有相互影響之特性，並非獨立作用；單因子分析進一步確認，ϕB 對流動距離與最終堆積高度具備高敏感度，為流體停滯行為之主控因子；ξ 則主要調控高速輸送段之動能耗散速率與側向擴散幅度，對垂直堆積高度之控制力相對偏低。經參數率定，本場址最佳參數組合為 ϕB = 6°、ξ = 40 m/s²、內摩擦角 δ = 17°，空間量化指標 CSI 達 0.626，AUROC 達 0.924（優良等級）。率定結果中偏低之 ξ 值（40 m/s²），反映本案例高含水量泥流型土石流在超額孔隙水壓作用下庫倫摩擦大幅衰減之流變特性，需透過強紊流阻力加以補償，為單相流假設下之等效參數調整機制。
整體結果顯示，r.avaflow 單相流模式在合理之流變參數定標下，無需分相計算仍能高度鑑別極端災害事件之流動邊界與最終地貌特徵，具備應用於台灣本土土石流潛勢溪流危害評估之潛力。

Taiwan's steep terrain and increasingly frequent extreme rainfall events make debris flows a persistent hazard. This study applies the open-source mass-flow model r.avaflow in single-phase mode to simulate the catastrophic debris flow triggered by Typhoon Morakot (2009) at the DF075 debris-flow prone stream (Xinkai settlement, Liugui District, Kaohsiung City), where cumulative rainfall reached 1,330.7 mm, claiming 28 lives and burying 30 households. A 5×5 cross-parameter sensitivity analysis was conducted on the two core Voellmy rheological parameters — basal friction angle (ϕB) and turbulent friction coefficient (ξ) — using the Critical Success Index (CSI) and AUROC as spatial validation metrics. The analysis revealed a clear equifinality phenomenon between ϕB and ξ. Parameter calibration yielded an optimal combination of ϕB = 6°, ξ = 40 m/s², and internal friction angle δ = 17°, achieving CSI = 0.626 and AUROC = 0.924. The model successfully reproduced the highly transient disaster characteristics, with the flow front reaching the alluvial fan apex within 56.9 seconds and inundating the downstream settlement within approximately 298 seconds. Results demonstrate that r.avaflow in single-phase mode can reliably delineate debris flow boundaries for extreme events under well-calibrated parameters.

[1]. Chen, J.-C., & Chuang, M.-R. (2014). Discharge of landslide-induced debris flows: case studies of Typhoon Morakot in southern Taiwan. Natural Hazards and Earth System Sciences, 14, 1719–1730.
[2]. Chen, J.-C., Wang, J.-S., Chuang, M.-R., & Jeng, C.-J. (2014). Numerical simulation of the inundation area for landslide-induced debris flow: a case study of the Sha-Xinkai gully in southern Taiwan. WIT Transactions on Ecology and The Environment, 184, 35-45.
[3]. Formetta, G., Capparelli, G., & Versace, P. (2016). Evaluating performance of simplified physically based models for shallow landslide susceptibility. Hydrology and Earth System Sciences, 20, 4585–4603.
[4]. Hungr, O. (1995). A model for the runout analysis of rapid flow slides, debris flows, and avalanches. Canadian Geotechnical Journal, 32, 610-623.
[5]. Iverson, R. M. (1997). The physics of debris flows. Reviews of Geophysics, 35(3), 245-296.
[6]. Mergili, M., Fischer, J.-T., Krenn, J., & Pudasaini, S. P. (2017). r.avaflow v1, an advanced open-source computational framework for the propagation and interaction of two-phase mass flows. Geoscientific Model Development, 10, 553–569.
[7]. Meyrat, G., McArdell, B., Müller, C. R., Munch, J., & Bartelt, P. (2023). Voellmy-type mixture rheologies for dilatant, two-layer debris flow models. Landslides, 20(11), 2415–2429. https://doi.org/10.1007/s10346-023-02092-w
[8]. Mikoš, M., & Bezak, N. (2021). Debris Flow Modelling Using RAMMS Model in the Alpine Environment With Focus on the Model Parameters and Main Characteristics. Frontiers in Earth Science, 8, 605061.
[9]. Pandey, N. K., Satyam, N., & Gupta, K. (2024). Landslide-induced debris flows and its investigation using r.avaflow: A case study from Kotrupi, India. Journal of Earth System Science, 133(97).
[10]. Schraml, K., Thomschitz, B., McArdell, B. W., Graf, C., & Kaitna, R. (2015). Modeling debris-flow runout patterns on two alpine fans with different dynamic simulation models. Natural Hazards and Earth System Sciences, 15, 1483–1492.
[11]. Sosio, R., Crosta, G.B., Chen, J.H., & Hungr, O. (2013). Runout Prediction of Rock Avalanches in Volcanic and Glacial Terrains. In: C. Margottini et al. (eds.), Landslide Science and Practice, Vol. 3. Springer-Verlag Berlin Heidelberg, 289-295.
[12]. Vicari, H. (2018). Physical and numerical modelling of debris flows [Master's thesis, Politecnico di Torino]
[13]. Koo, R.C.H., Kwan, J.S.H., Lam, C., Goodwin, G.R., Choi, C.E., Ng, C.W.W., Yiu, J., Ho, K.K.S., Pun, W.K. (2018) Back-analysis of geophysical flows using three-dimensional runout model. Canadian Geotechnical Journal, 55 (8). 1081-1094.
[14]. 行政院農業委員會水土保持局（2009）。98年莫拉克颱風重大土石災例速報：98年莫拉克颱風-高雄六龜-003
[15]. 專案計畫團隊（2009）。高屏溪流域上游坡地莫拉克風災整體復建規劃 成果報告（參見 6.2.7 高雄縣六龜鄉新發村 23 鄰(新開部落)。
[16]. Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters, 27(8), 861–874.
[17]. Mousavi Tayebi, S. A., Moussavi Tayyebi, S., & Pastor, M. (2021). Depth-integrated two-phase modeling of two real cases: A comparison between r.avaflow and GeoFlow-SPH codes. Applied Sciences, 11(12), 5751.
[18]. Pianosi, F., Beven, K., Freer, J., Hall, J. W., Rougier, J., Stephenson, D. B., & Wagener, T. (2016). Sensitivity analysis of environmental models: A systematic review with practical workflow. Environmental Modelling & Software, 79, 214–232.
[19]. Quan Luna,B.,et al.(2011). The application of numerical debris flow modelling. Natural Hazards and Earth System Sciences, 11, 1–14.
[20]. Saito, T., & Rehmsmeier, M. (2015). The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS ONE, 10(3), e0118432.
[21]. Zhuang, Y., Bartelt, P., & McArdell, B. W. (2024). [Manuscript concerning the geophysical mass flows and rheology]. Natural Hazards and Earth System Sciences.
[22]. 彭帥英、李霄琳、牛岑岑 等. (2023). 泥石流模擬實驗裝置及虛擬仿真平台設計與應用. 實驗技術與管理, 40(10), 153–158.
[23]. Beven,K.,& Freer,J.(2001).Equifinality, data assimilation, and uncertainty estimation in mechanistic modelling of complex environmental systems using the GLUE methodology. Journal of Hydrology, 249, 11–29.
[24]. Mergili, M., Frank, B., Fischer, J.-T., Huggel, C., & Pudasaini, S. P. (2018). Computational experiments on the 1962 and 1970 landslide events at Huascarán (Peru) with r.avaflow: Lessons learned for predictive mass flow simulations. Geomorphology, 322, 15–28.
[25]. Lin, G. F., Chen, L. H., & Lai, J. N. (2006). Assessment of risk due to debris flow events: A case study in central Taiwan. Natural Hazards, 39, 1–14.
[26]. Baggio, T., Mergili, M., & D'Agostino, V. (2021). Advances in the simulation of debris flow erosion: The case study of the Rio Gere (Italy) event of the 4th August 2017. Geomorphology, 381, 107664.
[27]. Christen, M., Kowalski, J., & Bartelt, P. (2010). RAMMS: Numerical simulation of dense snow avalanches in three-dimensional terrain. Cold Regions Science and Technology, 63, 1–14.
[28]. Hasegawa, S., Dahal, R. K., Yamanaka, M., Bhandary, N. P., Yatabe, R., & Inagaki, H. (2008). Causes of large-scale landslides in the Lesser Himalaya of central Nepal. Environmental Geology.
[29]. 彭繼賢 (2006)。應用FLO-2D 於臺灣中部地區土石流流況分析之研究 [碩士論文，國立臺灣大學]
[30]. Naef, D., Rickenmann, D., Rutschmann, P., & McArdell, B. W. (2006). Comparison of flow resistance relations for debris flows using a one-dimensional finite element simulation model. Natural Hazards and Earth System Sciences, 6(1), 155–165.