暴雨事件下飲用水水源管理課題分析-以新店溪水源為例

Abstract

新店溪為大臺北地區重要水源，其上游臺北水源特定區風化作用旺盛，地表土層鬆軟，且地形多為陡峭之坡地，強降雨發生時易沖刷地表土石並引發淺層崩塌。2015年蘇迪勒颱風所帶來的強降雨，除再次重創原先崩塌地，也在桶後、烏來等中上游地段造成多處崩塌。大量崩塌泥沙因強降雨被帶入河川，新店溪原水濁度因而超過下游直潭淨水廠原水濁度最大處理能力，最後導致大臺北地區停水。為提升大臺北地區供水系統面對高原水濁度之應變能力，本研究從上游原水至下游淨水廠探討水源管理課題。並利用2015年至2019年上游福山測站、攬勝橋測站與下游下龜山橋測站蒐集之雨量資料與濁度資料進行相關性分析，並考慮北勢溪匯流之影響，嘗試建立下游濁度推估模式。 濁度延時分析結果發現，福山測站與下龜山橋測站之濁度延時約為2.6小時。而攬勝橋測站與下龜山橋測站之濁度延時約為1小時。 雨量與濁度的相關性分析，依照中央氣象局頒布之雨量分級，選定大雨等級以上之強降雨事件作為分析主軸。其結果顯示，福山測站與攬勝橋測站各起事件之各別分析上，自身測站濁度資訊或與下龜山橋測站濁度資訊回歸，其線性回歸與乘冪回歸相關性R2落在0.6~0.7。將各起事件綜合分析結果，發現福山測站及攬勝橋測站利用日降雨資訊與日濁度資訊相關性回歸結果較差，線性R2約落在0.3~0.6之間，乘冪R2約落在0.2~0.4之間。然而若是利用「累積至最大濁度發生時之降雨量」與事件濁度資訊，福山測站因調整過後資料數不足故無綜合事件分析，而攬勝橋測站線性R2在最大濁度資訊結果為0.78，與下龜山橋測站線性R2在最大濁度資訊結果為0.77。由此可知「累積至最大濁度發生時之降雨量」為推估濁度之重要氣象因子。Xindian river is an important water source for the greater Taipei area. The upper reaches of the river have strong weathering, soft topsoil, and steep slopes. When heavy rainfall occurs, it is easy to scour the surface and cause shallow collapse. In 2015, the heavy rainfall brought by typhoon Soudelor not only once again hit the collapse site hard, but also caused many collapses in the middle and upper reaches such as Tunghou and Wulai. A large amount of collapsed sediment was carried into the river by heavy rainfall, and the turbidity of raw water in Xindian river exceeded the maximum treatment capacity of raw water at the downstream Zhitan water purification plant, resulting in the closure of water in the greater Taipei area.In order to improve the strain capacity of the water supply system in the greater Taipei area to the turbidity of plateau water, this study explored the topic of water source management from the upstream raw water to the downstream water purification plants. The correlation analysis of rainfall data and turbidity data collected from the upstream Fushan observation station, Lansheng bridge observation station and downstream Xiaqiushan bridge observation station from 2015 to 2019 was made, and the influence of the confluence of Beishi river was considered to establish an early warning model of turbidity in the downstream. The turbidity delay analysis results show that the turbidity delay of Fushan station and Xiaqiushan bridge station is about 2.6 hours. The turbidity delay of Lansheng bridge station and Xiaqiushan bridge station is about 1 hour. The correlation analysis of rainfall and turbidity is based on the rainfall classification issued by the Central weather bureau. The results show that the turbidity information of Fushan station and Lansheng bridge station is regressive to the turbidity information of Xiaqiushan bridge station, and the correlation R2 of linear regression and power regression is between 0.6 and 0.7.Based on the comprehensive analysis of the events, it was found that the regression results of the correlation between daily rainfall information and daily turbidity information used by the Fushan and Lansheng bridge stations were poor, with the linear R2 falling between 0.3 and 0.6, and the power R2 falling between 0.2 and 0.4. However, if the cumulative rainfall and event turbidity information of each event were used, there was no comprehensive event analysis at Fushan station due to insufficient data after the adjustment, while the maximum turbidity information at Lansheng bridge station was 0.78 in linear R2 and 0.77 in linear R2 at Xiaqiushan bridge station. In addition, "precipitation at the time of accumulation to maximum turbidity" is an important factor in estimating turbidity information.