都市不透水面積增加與極端降雨頻繁易造成地表逕流排水不及之淹水災害，而綠屋頂具降雨滯留潛力，為近年洪水管理的有效方案之一；本研究於中央大學建立包含實驗組(綠屋頂)與對照組(沒有綠屋頂)之觀測系統，分析綠屋頂減洪效用，採用指標包含滯留體積、尖峰削減比和尖峰延遲時間，同時利用半年觀測資料結合綠屋頂水文模式(NTU-GR)做排出量之檢定與驗證，並應用NTU-GR於歷史三年降雨資料以進行更全面之分析。本研究將降雨事件以總降雨量20公厘為界限分成大型降雨和小型降雨，NTU-GR驗證之效率係數分別為0.89和0.84，表示NTU-GR在不同的降雨大小都能合理預測綠屋頂的水文表現。觀測資料的減洪效用分析結果顯示滯留體積和尖峰削減比的平均都有50%以上的表現，尖峰延遲時間約0到1小時。模式應用則以不同臨前含水量(40%、30%、20%、10%)為條件輸入過去所有降雨事件模擬，結果顯示若綠屋頂的臨前含水量在30%以下，有75%機率的事件滯留體積與尖峰削減比都能有約50%以上的表現。再以大小型降雨做分類，綠屋頂的臨前含水量若在30%以下，小型降雨事件下綠屋頂的滯留體積和尖峰削減比都有99.6%以上的表現；大型降雨事件且綠屋頂臨前含水量在20%以下的話，有50%的機率事件之滯留體積有50%以上且尖峰削減比有60%以上的表現，而尖峰延遲時間也隨臨前含水量越低平均時間越長。說明臨前含水量和降雨大小都是影響整個綠屋頂減洪效用的關鍵，綠屋頂在臨前含水量低時對於治理洪水有一定程度的貢獻。另外，本研究也發展出一個簡單的公式可以快速且合理推估出特定綠屋頂臨前含水量下每一場降雨事件之滯留體積比。;Due to the increase of urban impervious area and frequent extreme rainfall, the drainage system may not be able to drain stormwater and thus cause flooding. Green roof has the rainfall retention potential. It is one of the effective solutions for flood management in recent years . We build a green roof observation system which including experimental group (green roof) and control group (no green roof) in National Central University to analyze the stormwater reduction of the green roof . The stormwater reduction indicators include retained volume, peak reduction and peak-delay time. Then observation data for half year was input into a green roof hydrological model (NTU-GR) for calibration and validation. After confirming the feasibility of the model, NTU-GR model was applied to simulate the historical three-year rainfall data for a more comprehensive analysis. In this study, the rainfall events are divided into large rainfall and small rainfall with a total rainfall of 20 mm as the threshold. The Nash coefficient of two kinds of events are 0.89 and 0.84, respectively, indicating NTU-GR model performs well in simulating the green roof hydrology under both large rainfall and small rainfall events. The results of stormwater reduction of the observation data shows that both the retained volume and peak reduction have an average performance of more than 50%, and the peak delay time is about 0 to 1 hour. Moreover, the model is applied to simulates all of the rainfall events in the past 3 years with different antecedent water content (AWC=40%, 30%, 20%, 10%). The results of stormwater reduction show that if the AWC of the green roof is below 30%, the retained volume and peak reduction ratio are more than 50% for 75% of the events. When the events of rainfall is less than 20 mm and the AWC of the green roof is below 30%, the retained volume and the peak reduction of the green roof will be more than 99.6%. When the events of rainfall is larger than 20 mm and the AWC of the green roof is below 20%, 50% of the events have a performance of more than 50% in retained volume and more than 60% in peak reduction. When the AWC is lower, the average of delay time is longer. It shows that the AWC and rainfall are the key factors affecting the stormwater reduction of the entire green roof. Green roofs have a certain degree of contribution to stormwater reduction when the AWC is low. Furthermore, this research also develops a simple formula that can quickly and reasonably estimate the retained volume of each rainfall events under a specific green roof′s AWC.
