請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/52702
標題: | 探討流量、水質與魚類群聚之關係 Explore the impacts of river flow and water quality on fish communities |
作者: | Jia-Hao Hu 胡家豪 |
指導教授: | 張斐章 |
關鍵字: | 生態水文,淡水河流域,調適性網路模糊推論系統,Gamma檢定,魚類群聚,流態, Eco-hydrology,Tamsui River,Adapted Network-Based Fuzzy Inference System (ANFIS),Gamma Test (GT),Fish Community,Flow Regime, |
出版年 : | 2015 |
學位: | 碩士 |
摘要: | 台灣受到先天地理環境及氣候條件之限制,河川坡陡流急,降雨在時空上分布不均,加上颱風豪雨等季節性氣候情況,急驟之降雨易導致河川生態水文環境產生劇烈變化及衝擊。在過去河川流域的研究當中,流量、水質往往有著密切的交互關係,並同時影響著河川生態系統,為了達到永續發展的目的,水資源管理需具備合理與整體之規劃,在開發環境資源時亦能考量與執行生態永續經營之理念為重要的課題之一,因此,本研究藉由探討河川流量、水質及魚類群聚特性之複雜關係,了解彼此間互相影響之機制以尋找改善河川生態系統之方法,進而提升台灣對於河川之水資源經營管理及永續發展之效能。
本研究選定淡水河流域具有較長期(2005~2012)監測之日流量、水質月資料及生態採樣調查資料之新店溪支流為研究區域。本研究根據台灣生態水文指標系統(Taiwan Ecohydrologic Indicator System, TEIS)先將河川流量資料轉換為流態資料(月尺度),再藉由統計方法分析探討流態、水質及魚類群聚在河川上、下游之差異性,進而透過Gamma Test (GT)篩選流態及水質因子,辨識推估模式輸入因子較佳之組合,以降低模式結構的複雜度,並避免容易產生資料過度描述(over-fitting)的情形,接著利用調適性網路模糊推論系統(Adapted Network-Based Fuzzy Inference System, ANFIS)針對魚類之多樣性指標建置推估模式,結果顯示利用GT所篩選之輸入因子能有效地推估魚類多樣性指標,此外,模式輸入項加入流態因子能獲致更好的推估效果。最後透過規則庫中各輸入項之隸屬度函數探討輸入因子對於河川魚類生態之影響,進一步掌握該流域之生態系統變異,期能作為日後相關水文生態實務管理之參考,以維持河川整體的生態穩定性及永續發展。 Subject to the geographic environment and climatic conditions of Taiwan, rivers in Taiwan are of steep slopes and flow into oceans very quickly. Due to the uneven temporal and spatial distribution of rainfall and the severe intensity and short duration of typhoons and storms, sudden rainfall would easily cause huge variations and significant impacts on riverine eco-hydrological environments. The previous research on river basins indicated that river flow and water quality is closely related to each other, which influences river ecosystems simultaneously. To achieve the goal of sustainable development of water resources, rationality and integrity is essential for water resources management while planning to exploit environmental resources in consideration of ecosystem sustainability. Therefore, this study explores the complex relationship among river flow, water quality and fish communities in order to understand the interactive mechanism among each other and develop a methodology suitable for improving river ecosystems; and consequently promotes the management and sustainable development of water resources. In this study, the Xindian River, one of the three major tributaries of the Tamsui River, is chosen as the as study area, which has long-term (2005-2012) daily (river flow) and monthly (water quality) observational data as well as fishery data collected from field surveys. Based on the Taiwan Ecohydrologic Indicator System (TEIS), river flow data is converted into flow regime in a monthly scale. Statistical analyses are then conducted to explore the differences of flow regime, water quality and fish communities between the upstream and downstream areas of the Xindian River. This study next estimates fish diversity indexes by using the Adapted Network-Based Fuzzy Inference System (ANFIS) based on key input factors determined by the Gamma Test (GT), a powerful tool for reducing model complexity of artificial neural networks (ANNs). The results reveal that the constructed model with key input factors selected by the GT can effectively estimate fish diversity indexes, and the estimation performance becomes even better if flow regime can be incorporated as model inputs. Finally, the investigation on the membership functions of the ANFIS can explore the impacts of each input on fish diversity to keep abreast of the variation of the river ecosystem. The results of this study can provide valuable findings as a guiding reference for the practices of eco-hydro system management and the planning of sustainable water resources management of the whole river basin in the future. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/52702 |
全文授權: | 有償授權 |
顯示於系所單位: | 生物環境系統工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-104-1.pdf 目前未授權公開取用 | 4.26 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。