應用BASINS/HSPF推估翡翠水庫非點源污染量暨模擬最佳管理作業之研究

Abstract

本研究應用BASINS/HSPF集水區模式，進行翡翠水庫集水區水文與水質模擬，依據地理環境之差異，將翡翠水庫集水區分為三個集水區，分別為北勢溪上游集水區、魚逮魚堀溪集水區以及金瓜寮溪集水區，探討各個集水區自民國95年至99年間集水區水文與水質變化，水質項目包括懸浮固體以及總磷。利用水文與水質模擬出來的結果，推估歷年來三個集水區內非點源污染量。此外，本研究亦考慮陸域甲類水體水質標準，估算出各個集水區之總磷最大日負荷規劃(total maximum daily load, TMDL)，針對超出分配量之污染負荷，進行總磷污染量之削減規劃，探討農業區之不同土地利用百分比(10%、30%、50%)下，增設各種削減措施時，其總磷削減率變化，以期能達到保護集水區水質及生態系統之目的。 本研究針對水文模擬結果使用相關係數(r) ，決定係數(R^2)，有效性係數(NSE)進行率定驗證，水質則是以平均誤差百分比(MPE)進行率定驗證，其計算結果皆符合模式可接受範圍。集水區非點源污染量推估結果，研究發現北勢溪上游集水區懸浮固體量為6,961 ton/yr，總磷為18,575 kg/yr；魚逮魚堀溪集水區懸浮固體量為6,422 ton/yr，總磷為7,597 kg/yr；金瓜寮溪集水區懸浮固體量為1,033 ton/yr，總磷為2,581 kg/yr。上述結果顯示北勢溪上游集水區為懸浮固體與總磷排放量最高之地區，魚逮魚堀溪集水區次之，金瓜寮溪集水區最少。 根據文獻建議MOS (margin of safety)值與Future Growth值，進行總磷最大日負荷規劃(total maximum daily load, TMDL)之推估，估算結果可知北勢溪上游集水區年平均總磷負荷量為16,343 kg/yr，魚逮魚堀溪集水區年平均總磷負荷量為7,597 kg/yr，金瓜寮溪集水區年平均總磷負荷量為2,581 kg/yr。將上述結果集水區與模擬污染量比較，北勢溪上游集水區需進行17.29%之削減需求，金瓜寮溪集水區需進行18%之削減需求，魚逮魚堀溪上有餘裕空間，不必進行削減措施。 此外，將各個子集水區輸出結果，發現潛在高濃度總磷排放之關鍵區域，均位於農業用地較為廣大之地區，推估可能是由農業活動造成此總磷高濃度排放之原因。故針對部分區域進行農地污染削減措施， 結果顯示人工溼地與草溝對總磷削減率較為顯著，過濾帶對總磷削減率較為不佳，但整體設計而言，均未達水體最大日負荷量之削減需求。

This study used the BASINS/HSPF model to simulate the flow and water quality in the Feisui reservoir watershed. By dividity the watershed into three subwatershedbased on geographical features: Pei-shi, Dai-yu-ku, Jin-gua-liao creek watersheds, data of current five years (2006 to 2010) of hydrology and water quality were used to calibrate and validate the model. This study focused on the suspended solids (SS) and total phosphorous (TP) as the target pollutants. From the matching of actual data, the pollutant loads of suspended solids (SS) and total phosphorous (TP) could be estimated by the model. In addition, the development of TMDL (tatal maximum daily load) was based on the existing environmental standard for protection of water body for each watershed. From TMDL and simulation of pollutant loads of TP, the required TP removal rate for each watershed was derived. Four types of potential BMPs (best management practices) were employed to the areas where TP concentrations were high. There BMPs in the model are wet pond, conctructed wetland, grass swale, and filter strips. The calibration model showed that Pei-shi creek watershed’s SS pollutant loads was 6,961 ton/yr, and TP pollutant loads was 18,575 kg/yr. Dai-yu-ku creek watershed was 64,214 ton/yr of SS, and 7,597 kg/yr of TP. Jin-gua-liao creek watershed was 1,033 ton/yr of SS, and 2,581 kg/yr of TP. Consequently, the Pei-shi creek watershed had the most SS and TP pollutant loads in the watershed. The estimation of TMDL, it was assumed that 5% MOS value and 1% future growth. Compared the simulated pollutant loads with WLA+LA (waste load allocation+ load allocation), Pei-shi and Jin-gua-liao creek watersheds need to employ the BMPs. Base on the result of BMPs simulation, the constructed wetland and grass swale have a better removal rate for TP. While the removal rate by filter strips was lowest.However, there BMPs couldn’t reduce the pollutant loads to below the TMDL.