- 學位論文
濱水生態系統管理模式參數敏感度分析
Parameter Sensitivity of the Riparian Ecosystem Management Model
摘要
濱水緩衝帶泛指沿河川兩側之植生區域，它具有穩定河道、減緩地表逕流、攔阻泥砂、削減營養鹽、保護生物多樣性等多項功能，可有效地減少非點源污染，保育河川之生態環境，為最佳管理措施(Best Management Practices, BMPs)之ㄧ。目前國內水保法規定緩衝帶寬度應為三十公尺至五十公尺，然而緩衝帶設計宜因地制宜作適當的植生配置與大小形狀規劃，全面採取一定寬度的設計方式顯然不適宜且非最佳方式。實際田野監測與數理模擬為評估緩衝帶污染削減效率的唯二方法，然而長期連續地監測在人物力和時間上皆不太可行，因此，應用驗證後的數理模式來分析不同氣候、地文及管理情況下緩衝帶之效能，為目前規劃緩衝帶最有效且可行之方式。 濱水緩衝帶泛指沿河川兩側之植生區域，它具有穩定河道、減緩地表逕流、攔阻泥砂、削減營養鹽、保護生物多樣性等多項功能，可有效地減少非點源污染，保育河川之生態環境，為最佳管理措施(Best Management Practices, BMPs)之ㄧ。目前國內水保法規定緩衝帶寬度應為三十公尺至五十公尺，然而緩衝帶設計宜因地制宜作適當的植生配置與大小形狀規劃，全面採取一定寬度的設計方式顯然不適宜且非最佳方式。實際田野監測與數理模擬為評估緩衝帶污染削減效率的唯二方法，然而長期連續地監測在人物力和時間上皆不太可行，因此，應用驗證後的數理模式來分析不同氣候、地文及管理情況下緩衝帶之效能，為目前規劃緩衝帶最有效且可行之方式。 由美國農業部(USDA-ARS)所發展之濱水生態系統管理模式(Riparian Ecosystem Management Model, REMM)具有模擬河川濱水區內水文循環、土壤冲蝕與泥砂傳輸、營養鹽循環及植物生長之功能，為緩衝帶規劃與設計之有效工具。此模式模擬複雜的濱水生態系統，輸入參數多且繁雜。因此，分析模式之參數敏感度不但有助於了解模式物理特性，更可提供使用者進行模式檢定時，優先調整參數和調動範圍的參考依據。本研究敏感度分析以狀態數(Condition Number, CN)作為參數敏感度高低之評估，此值為模式輸入及輸出參數的相對誤差之比值，因此，當CN值越大表示敏感度越高。研究選取模式中38個緩衝帶輸入參數將其區分為三大類，分別為水文參數12個、土壤冲蝕5個和營養鹽參數21個，並以翡翠水庫集水區內逮魚崛溪濱水區場址建立模式模擬背景值，再依次變動各參數起始值±5%和±10%進行模擬，以推求各參數對應不同輸出之CN值。研究結果顯示對逕流、滲漏量敏感之參數為土壤水份含量(θ)、田間容水量(FC)和土壤孔隙度(ψ)；而表土裸露百分比(BS)與糙度係數(R)對於泥砂輸出量具有較高之敏感度；本研究選用之營養鹽參數，對營養鹽氮和磷負荷量影響程度較低，然而對逕流量敏感之水文參數卻發現明顯地影響營養鹽承載輸出。歸納三類參數之CN值之結果顯示，土壤水份含量、孔隙度和田間容水量三種參數對所有輸出量皆具有高敏感度。
並列摘要
Riparian buffer zones are essential to control nonpoint source pollution and create healthy ecosystem for stream, inasmuch as they slow nutrients and sediments and filter them out of surface runoff before those nutrients and sediments reach the water bodies. Moreover, the buffer zones stabilize streambanks and floodsplains as well as provide habitat for wildlife and fish. The riparian buffer ecosystem has been recommended as one of the best management practices (BMPs) to mitigate nonpoint source pollution effectively. However, the variation of pollution reduction efficiency depended on the hydrologic condition, sizes of buffer zones, vegetation characteristics, and biomass harvesting managements. Provided that field experiments are expensive and limited to hydrologic conditions, physiographic characteristics and herbaceous buffer scenarios, mathematical models become the best tools available to help quantify the water quality benefits of riparian buffers under varying site conditions. The Riparian Ecosystem Management Model (REMM) has been developed to simulate surface and subsurface riparian buffer hydrology, sediment transport, vegetation growth, and nutrient (C, N, P) dynamics. The REMM model requires numerous inputs to simulate the complex riparian ecosystem. This study aims to explore the mechanism and evaluate the sensitivity of parameters in the REMM model for its application on Feitsui Reservoir watershed to analyze nonpoint source pollution reduction. The Condition Number (CN) method was employed to measure sensitivity of various parameters for associated model outputs. The results indicate that volumetric water content (θ), field capacity (FC), and porosity (ψ) are the most sensitive parameters for deep seepage, surface and subsurface runoff predictions. In addition, particulate surface organic N, organic P, and sediment yield are moderately sensitive to percentage of bare soil (BS) and interrill roughness (R). As for all nutrient loadings, nutrient-associated parameters are found to be less sensitive. However, parameters sensitive for runoffs also have great impacts on nutrient loadings. Overall, volumetric water content (θ), field capacity (FC), and porosity (ψ) are highly sensitive for all model outputs, which suggest the potential uncertainty of model predictions. Additional attentions are advised for those parameters during modeling and calibration.