氣候變遷對集水區水量及水質之衝擊評估-以翡翠水庫集水區為例

Assessing the Effects of Climate Change on Stream Flow and Water Quality in Watersheds-A Case Study of Fei-Tsui Reservoir Watershed

張祉國(Chih-Kuo Chang)

指導教授 ： 闕蓓德

國立臺灣大學/工學院/環境工程學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.03491

摘要

了解集水區在氣候變遷下之水量、水質衝擊為集水區管理的重要一環。本研究使用美國環保署(USEPA)所發佈的BASINS (Better Assessment Science Integrating point & Non-point Sources)平台內之HSPF (Hydrological Simulation Program-Fortran)模式及CAT (climate assessment tools)分析工具，探討水量、水質(懸浮固體、總磷)可能受到之衝擊。
本研究以台灣北部的翡翠水庫集水區作為案例研究，此集水區總面積303平方公里，為全台灣第二大水庫。模式中將翡翠水庫上游根據流域範圍切割為三個區塊：北勢溪(110.3 Km2)、魚逮魚堀溪(78.1 Km2)和金瓜寮溪(24.6 Km2)，每一區塊又根據地形切割成數量不等的小區塊，藉以探討進入水庫的水量及污染物總量。
本研究使用2008年之水量、水質資料進行率定。並根據IPCC (Intergovernmental Panel on Climate Change)之AR4 (the Fourth Assessment Report)報告設定二種情境：A1F1 (2100, CO2=970 ppm, 26.9 ℃)、B1 (2100, CO2=550 ppm, 21.6 ℃)。此外針對不同的雨量變異(0 %、+-10 %、+-20 %)進行衝擊分析，並與現況(2008，CO2=385 ppm, 20.5 ℃)進行比較，以得知集水區在不同溫度及雨量變異下對水量、懸浮固體和總磷之衝擊。
水文率定之結果顯示：北勢溪、魚逮魚堀溪及金瓜寮溪流域之相關係數分別為：0.9、0.96和0.91。而水質率定之結果在觀測資料不足的情況下，對於懸浮固體和總磷仍有良好的預測結果。而參數敏感度分析之結果顯示：LZSN、KVARY、INFLIT、KGER和WSQOP為對於水量、水質影響較大之參數。
由衝擊分析的結果可知：研究區域於2100年A1F1及B1情境下，河川平均流量將分別減少1.51 %及0.17 %；懸浮固體輸出量分別上升41.12 %及1.12 %；總磷輸出量分別減少1.57 %及增加0.06 %。另二個情境在雨量 10 %、 20 %的變異下，將對流量產生-23.67 %~+21.88 % 之變動；懸浮固體之變異為-14.21 %~+84.48 %；總磷為-20.28 %~+16.97 %。
由本研究之結果可知：溫度升高和雨量強度增加，皆會對預測的水量及水質產生嚴重的衝擊。其中雨量造成之衝擊較大且直接；而溫度造成的衝擊較小，但所影響的範圍較廣，且容易受水質參數設定之影響。

關鍵字

氣候變遷 ； HSPF模式 ； 衝擊分析 ； 流量 ； 懸浮固體 ； 總磷

並列摘要

Understanding the potential impact of discharge and water quality of watershed under climate change is a critical assignment for watershed management. In this study, HSPF (Hydrological Simulation Program-Fortran) and CAT (climate assessment tools) within BASINS (Better Assessment Science Integrating point & Non-point Sources) which is developed by U.S. EPA are used for investigating the impact caused by climate change.
Fei-Tsui Reservoir Watershed (303 Km2) which is the second largest reservoir of Taiwan is chosen for a case study. The upstream of reservoir is divided into three regions based on different basins, Pei-shih (110.3 Km2), Tai-yu-chueh (78.1 Km2) and Ching-kua-liao (24.6 Km2). And each region is divided into small sub-regions base on topography by model.
Discharge (daily) and water quality (monthly) data of 2008 are used for calibration. In this study, two climate scenarios are made based on AR4 (the Fourth Assessment Report) of IPCC(Intergovernmental Panel on Climate Change), A1F1 (2100, CO2=970 ppm, 26.9 ℃) and B1(2100, CO2=550 ppm, 21.6 ℃).Different precipitation variances(0 %, ±10 % and ±20 % ) are modeled to execute impact analysis for two scenarios and compared with baseline scenario (2008, CO2=385 ppm, 20.5 ℃).
The correlation coefficient between simulated and observed data are 0.90、0.96 and 0.91 respectively of three regions in hydrological simulation. Though water quality data are not actually sufficient for calibration, output of SS and TP are acceptable.
In A1F1 and B1 scenario, stream flow decreases by 1.51 % and 0.17 % respectively, SS increases by 41.12 % and 1.12 % respectively, and TP decreases by 1.57 % and increases by 0.06 % respectively. Precipitation variance by ±10 % and ±20 % generally changes discharge, SS and TP by -23.67 %~+21.88 %, -14.21 %~+84.48 % and -20.28 %~+16.97 % respectively in three regions.
These results show that rising temperature and precipitation variance cause critical impact on predicted discharge and water quality. The impact caused by precipitation variance is larger than the impact caused by rising temperature and the mechanisms are simpler. In addition, the impact caused by temperature is more sensitive to parameter settings

並列關鍵字

climate change ； HSPF ； impact analysis ； stream flow ； suspended solid ； total phosphorus

參考文獻

張尊國，2006，「青潭堰集水區上污染源調查及水質管理計畫」，行政院環境保護署。

張尊國、張文亮，2005，「翡翠水庫水源保護區污染源調查計畫」，行政院環境保護署。

范正成、張尊國，2006，「翡翠水庫集水區污染源削減計畫」，行政院環境保護署。

黃政龍，2006，「暴雨時期河川污染物磷負荷量推估方法之研究」，碩士論文，國立台灣大學土木工程學研究所。

Bernstein, L., Bosch, P., Canziani, O. and Chen, Z. (2007) Climate Change 2007:Synthesis Report, Intergovernmental Panel of Climate Change.