泥沙運輸及河道變遷於礫石河道之研究-以士文溪為例

Sediment Transport and Morphodynamics in a Gravel Bed Channel, Shi-Wen, Taiwan

席克懷(Samkele Sikhulile Tfwala)

指導教授 ： 王裕民

國立屏東科技大學/工學院/土木工程系所/博士(2016年)

摘要

本研究主要探討颱洪期間洪水對礫石河床之泥砂運移及河床沖淤演變。研究對象為南台灣之士文溪，士文溪河床由礫石組成，全長22.3公里。研究區域平均每年遭受3至4個颱風的侵襲，因此洪水在此區域是普遍現象。頻繁的洪水事件以及集水區內土地利用的改變，造成河流形態的多樣性變化，可應用河流形態動力學或發展具有的準確預測能力之河床演變模式進行河流形態變遷之探討。河道輸砂量常用流量及泥砂輸運率之率定曲線來評估，本研究應用類神經網路模式進行產砂量估計結果發現可提高30％之準確性。自80年代即有士文溪土地利用對泥砂產量衝擊之相關研究且本研究證實自1984年至2015年地表覆蓋的改變，泥砂產量已增加為102.68%。為了更適當管理集水區，應繪製集水區內易沖蝕之區域。除了河流泥砂組成外，以三種方法進行河流形態改變之研究：河床變化、利用無人載具和數值方法。研究發現洪水對礫石河床之沖淤改變具有巨大衝擊，就單一洪水事件可增加下游泥砂之比例高達35%。河床形態的改變亦造成重大之衝擊例如河道的曼寧粗糙係數，而其可能對集水區之相關工程設施或設計產生重大的影響。本研究所選定之河段內藉由無人載具辨識沖蝕與淤積，並計算橫向之變化，進一步由ArcGIS分析顯示單一颱洪水事件橫向變化可達約1公尺。增加數值高程地形之空間解析度是研究河川形態演變之關鍵，本研究洪水前後皆達到6公分的解析度。本研究針對造成河流形態變遷與評估方法間之關係提供更好的工具，以促進河流及其流域的管理。河流形態變化需要長時間進行調查，然而在平均寬度70公尺之相對狹窄的河流中得到60公尺解析度的衛星影像是不可能的，因此很難確定關鍵變化。無人載具可望成為這一限制下的解決方案。最後， CCHE2D被應用作為模擬不同重現期距年(25、50、100、200年)洪水事件下河床沖淤演變之數值模式，並藉由洪水事件後河道橫斷面高程變化加以驗證。河床高程變化在重現期距25年洪水下沖蝕深度為5公尺，重現期距200年高達8公尺，結果顯示二維模式可合理評估在不同的暴雨下士文溪河床之沖淤演變。

關鍵字

類神經網路 ； CCHE2D ； 土地使用 ； 河床 ； 產砂量 ； 無人載具

並列摘要

This study investigates the influence of storm events, termed typhoons in the local region, on sediment dynamics and fluvial morphological changes on a 23 km gravel bed river of Shi‑Wen, southern Taiwan. Floods are prevalent in this region, with an average of 3 to 4 typhoons annually. These combined with land use changes within this watershed have resulted in several modifications of the fluvial morphology. Selected drivers of morphology were explored either to identify their contribution to morphodynamics or to develop models capable of predicting them accurately. Currently sediment rating curves are used to estimate sediment transport and discharge. A neural network model is proposed and developed, and it improved sediment yield estimations by 30 %. Impacts of land use/cover dynamics on Shi‑Wen sediment yield since the 80’s is studied, and land cover changes was found to have increased the yield by 102.68 % from 1984 to 2015. Areas more prone to erosion within the watershed were mapped for better management. Besides the sediment components, morphological changes were studied in three forms; riverbed variation, use of unmanned aerial vehicles and numerical means. The impacts of floods on gravel bed changes was found to have huge impacts, with just single flood increasing the proportions of sand downstream by up to 35 %. Variation in bed changes will have major impacts for example on Manning’s roughness coefficients, which may consequently impact engineering works that are on the rise in this watershed. The capability of UAV’s was realised as erosion and deposition, lateral changes within a selected segment of the river was computed. ArcGIS analysis revealed lateral changes of up to about 1 m following a single typhoon event. High resolution DEMs of pre and post flood events were achieved, having a resolution of 6 cm. Enhanced spatial resolutions are of paramount importance in the success of morphological studies. The relations between drivers of morphology, and the methods developed in this research work should provide better tools to facilitate better management of this river and its watershed. Moreover, additional surveys are still needed to evaluate the linkages between the sediment yield variations and the resulting changes in morphology. Data used in this study was gathered for only 5 years and morphodynamics takes a longer period. Long term morphological changes need to be investigated; however, currently this is not possible since this stream is relatively narrow, on average 70 m in width and satellite images available had a resolution of 60 m, making it difficult to identify key changes. UAVs are promising to be the viable solutions to this limitation. Finally, a numerical model, CCHE2D is then applied to model likely bed changes under storms of different return periods; 25 , 50 , 100 and 200 year. Prior to the application of the model, it was validated by cross section elevation changes following a storm event. Bed elevation changes entailed erosion depths of 5 m under 25 year return period up to 8 m under 200 year period. The 2D model has demonstrated the capability to estimate likely changes under different storms at Shi‑Wen river.

並列關鍵字

artificial neural network ； CCHE2D ； land use ； riverbed ； sediment yield ； unmanned aerial vehicle

參考文獻

Environmental Protection Administration. 2009. Typhoon Morakot. Tertiary Taiwan: Retrieved from http://www.epa.gov.tw.

Einstein, H. A. 1950. The Bed Load Function for Sediment Transportation in Open Channel Flows, Technical Bulletin No. 1026. Washington DC.

Hu, S. C., J. Y. Lu, I. Y. Wu, and T. F. Chen. 1995. "Interrill Erosion of Rates of High Rates of Clay Soils in Lien-Hua-Chi Area," Bulletin of the Taiwan Forest Research Institute, 10(1): 33-40.

Chen, Z. S., Z. Y. Hseu, and C. C. Tsai. 2015. The Soils of Taiwan. Edited by A. E. Hartemink, World Soils Book Series. New York: Springer Science.

Tsai, C. T. and C. H. Tsai. 1997. "A Study on the Applicability of Discharge Formulas for Trapezoidal Broad-Crested Weirs," Journal of Taiwan Water Conservation, 45(2): 29-45.