自然河廊形貌演變數值模擬：堰壩拆除與河岸可動性之影響分析

Abstract

臺灣河川常進行疏濬、截彎取直及堤防護岸等工程，以達到疏洪與土地開發的目的，但也因此易造成洪氾平原功能喪失、棲地結構單一化，進而導致生物多樣性下降。近年河川復育常以堰壩改善、除役或移除，以及鬆綁河岸束縛（如堤防後退、移除護岸）作為手段。既有研究指出，堰壩因攔截泥砂而降低蜿蜒遷移率，限制蜿蜒河道自然演化；然而，針對堰壩拆除如何促進蜿蜒演化及其作用機制之系統性量化研究仍相對不足，尤其是在河岸可動性（河岸侵蝕能力）條件下的平面型態演變。 本研究旨在探討恢復縱向輸砂連續性（堰體拆除）與恢復河岸可動性（可移動河岸）兩項條件下，河道蜿蜒形貌的演化歷程與促進效果。研究以 iRiC Nays2DH 水理輸砂模式建立數值模型，針對臺灣北磺溪主流二號攔河堰上游約 2.7 km 河段（累距 3,179–5,783 m）進行六組情境模擬，組合包含三種拆除程度（不拆除、部分拆除、完全拆除）與兩種河岸條件（固定河岸、可移動河岸）。蜿蜒特徵以 MATLAB 工具 MStaT 量化（蜿蜒度、平均振幅、平均弧波長等），並配合水動力、推移載輸砂量、沖淤分布與代表性斷面分析，釐清其形貌演化機制。 結果顯示，堰體拆除程度越高，河岸侵蝕與形貌活動越明顯；六組情境中以「完全拆除＋可移動河岸」之蜿蜒特徵最顯著，而「不拆除＋固定河岸」最不顯著。相較於後者，前者之平均振幅、蜿蜒度與平均弧波長分別增加 22.13%、1.51% 與 2.59%，顯示同時恢復縱向連續性與河岸可動性最能促進蜿蜒發展。機制分析指出：堰體拆除後，上游輸砂量增加使彎道內岸淤積（河曲沙洲）面積與高度提升，並促使剪應力分布更偏向外岸，強化外岸侵蝕，形成「沙洲推進（Bar Push）」效應；在可移動河岸條件下，外岸侵蝕增強使曲率上升並增加側向來砂量，進一步放大內岸淤積與主流外移，形成「河岸拉動（Bank Pull）」的正回饋，從而加速蜿蜒演化。 基於研究結果，本文對中高坡度且具蜿蜒潛勢河段提出以下管理建議：（1）優先恢復河岸可動性（如移除護岸、堤防後退），其促進蜿蜒之效果高於單純拆堰；（2）採分期或部分拆除策略，以兼顧生態復育與保全對象安全；（3）於特定彎道保留河岸緩衝帶，以降低侵蝕風險；（4）拆堰可能使推移載輸砂量同時出現坡度誘發峰值與流量誘發峰值，建議於中小流量期間執行，以避免峰值疊加造成劇烈衝擊。

River regulation measures such as channelization, dredging, levees, and bank revetments can simplify river habitats and disconnect floodplains, ultimately reducing biodiversity. In river restoration, decommissioning or removing weirs and restoring bank mobility are widely adopted approaches. Previous studies have shown that weirs can reduce meander migration by trapping sediment and disrupting downstream sediment continuity; however, quantitative and process-based evidence on how weir removal promotes planform meandering—especially under erodible bank conditions—remains limited. This study investigates the evolution and enhancement of channel meandering under (i) restored longitudinal sediment continuity through weir removal and (ii) restored bank erodibility through movable banks, representing conditions closer to a natural river corridor. Using the iRiC Nays2DH hydro-sediment transport model, we simulated six scenarios for a 2.7-km reach of the North Sulfur Creek mainstream in northern Taiwan (chainage 3,179–5,783 m), which includes the No. 2 Weir. The scenarios combine three weir conditions (no removal, partial removal, complete removal) with two bank conditions (fixed banks, movable banks). Meandering metrics (e.g., sinuosity, average amplitude, and average arc wavelength) were quantified using the MATLAB-based Meander Statistics Toolbox (MStaT), and the underlying mechanisms were examined through analyses of hydrodynamics, bedload transport, erosion–deposition patterns, and representative cross-sections. Results show that morphological activity and bank erosion generally increased with the degree of weir removal. Among the six scenarios, "complete removal with movable banks" exhibited the most pronounced meandering characteristics, whereas "no removal with fixed banks" exhibited the least. Compared with the latter, the former increased average amplitude, sinuosity, and average arc wavelength by 22.13%, 1.51%, and 2.59%, respectively, indicating that simultaneously restoring longitudinal continuity and bank erodibility is most effective in promoting meandering. Mechanistically, weir removal increased upstream sediment supply and enhanced inner-bank deposition (point-bar growth), which shifted shear stress toward the outer bank and intensified outer-bank erosion—consistent with the "Bar-Push" effect. Under movable-bank conditions, intensified outer-bank erosion increased curvature and lateral sediment supply, further enhancing inner-bank deposition and promoting additional outer-bankward flow shifting, forming a positive feedback consistent with the "Bank-Pull" effect. Based on these findings, management implications for medium-to-high slope rivers with meandering potential include: (1) prioritizing the restoration of bank mobility (e.g., removing revetments or setting back levees), which can be more effective than weir removal alone; (2) adopting phased or partial weir removal to balance restoration benefits and infrastructure safety; (3) reserving bank buffer zones at erosion-prone bends; and (4) scheduling removal during low to moderate discharge to avoid the superposition of slope-induced and discharge-induced peaks in bedload transport that may trigger abrupt geomorphic impacts.