考量土石流沖積扇風險的路線最佳化:現地調查、實驗與模型研究

Abstract

本論文致力於探討合併多個支流沖積扇形貌的變化、與主流的互動及對道路工程設計上的影響。透過現地調查、實驗、理論計算及數值分析來了解三者之間的關係。主要分為三個部分如下: 第一部分以台灣南部地區的荖濃溪與其支流布唐布納斯溪、清水溪及玉穗溪匯流處的沖積扇作為研究案例，進行無人飛行載具航拍及河川縱坡面的測量，進而了解其形貌的變化。第二部分則根據現地調查的結果，進行實驗的設計，並且藉由一套雷射掃描裝置來紀錄地形的演變，以助於了解在現地中無法記錄的演變過程。接著，根據實驗的結果與前人的理論比較，並以靜力平衡及材料特性的角度分析，考量凝聚力對原先的理論進行修正。第三部分則應用於道路規劃，本研究模型使用歷史數據來校準伽馬隨機過程，搭配蒙地卡羅模擬來評估災難事件對路線成本的影響，並且考慮最大道路曲率和道路坡度的限制，分別在道路路線和道路剖面中進行路線優化。 結果顯示，凝聚力會影響沖積扇堆積的理論與實驗的剖面有相當程度的吻合，同時在現地調查中也可觀察到類似的情況。此外，於道路規劃的初步分析中我們發現沖積扇料源的供給為序率性，而主流的搬移作用則為定常性。最後，若要在陡峭的地形上執行路線優化，有必要共同考慮道路路線和剖面。

This thesis examines the morphological evolution of coalescing tributary fans, their interaction with the trunk river and their impact on route engineering design. To understand the relationship between these three, the research includes field survey, experiment and modeling study. The field survey focused on the debris fans produced by Pu-Tun-Pu-Nas, Qingshui and Yusui tributaries at their confluence with Laonong river, Southern Taiwan. The aerial photography of UAV and the long profile of the river were conducted. The experiment is designed based on the results of the field survey. The terrain is recorded with laser scanning system to help understand the evolution process. The experimental results are compared with the previous theory. Then, the previous theory is modified by analyzing the static balance and material properties. The modeling results are further applied to route planning. The model uses historical data to calibrate a gamma stochastic process and uses Monte Carlo simulation to estimate the impact of disaster events on route cost. Route optimization is then conducted separately in terms of its horizontal profile (road alignment) and vertical profile (road profile), considering limitation on maximum road curvature and road slope. The results show that the theory considering cohesion produces deposition that are made consistent with the experimental profiles, and similar situations could be observed in situ. In addition, the analysis indicates that debris supply is stochastic, but trunk river removal is approximately deterministic. To perform route optimization in a steep terrain, finally, it appears necessary to jointly consider both road alignment and profile.