圖書典藏及影音數位平台

近年來氣候變化異常，導致年降雨時間及空間分佈更加不平均且降雨強度有逐漸增強趨勢，氣候變遷造成氣候條件改變，極端事件發生頻率增加，旱災、水災等自然災害頻傳，加深了水資源使用的困難度，然而社會經濟發展必須仰賴可靠之水資源供給，水資源利用受氣候變遷影響甚劇，不管河川可利用流量、地下水、或農業需求量均會受到衝擊，進而影響區域供水系統之穩定性。 苗栗地區之主要水源為中港溪及後龍溪，主要水庫有永和山水庫、田美堰攔河堰及明德水庫，供給標的為苗栗地區的公共給水、工業用水及農業用水，鑑於近年來因氣候變化異常，導致枯水期間不同標的用水分配不均之問題日趨嚴重，苗栗調度區因永和山水庫為供應民生用水且需向北支援新竹地區，可承受風險低；明德水庫蓄水容量小除供應民生用水外，主要供應溉灌用水，極易遇到停灌休耕停灌問題。永和山水庫於95年、98年辦理停灌休耕、明德水庫於93年、95年及98年辦理停灌休耕。因未來水文情況不確定性甚高，較難研判是否辦理停灌休耕。每年12月至1月需決定各水庫是否辦理停灌休耕，目前均以現況水情搭配歷史流量超越機率進行水庫蓄水模擬演算，準確性低、風險高，故建立長期預模式可提高決策可靠性。因此為更精確掌握枯水期河川可用水量及預測未來枯水期發生超越機率流量之發生機率。 本計畫目為將以往氣候異常、取水系統變異因素及近期降雨資訊特性納入模式中，以有別傳統方式來分析乾旱流量，模擬不同乾旱情況下永和水庫及明德水庫之可用水量分析，做為水源調配參考，爰提「永和山與明德水庫長期流量預測模式之研究與建置」

The uneven seasonal rainfall and steep topography have resulted in serious water shortage in Taiwan during low-rainfall periods. Since the river flow in the low-rainfall period is the flow releasing from aquifer due to rainfall from previous months, a physically-based runoff model may be able to provide good simulation for the daily and monthly flow without use of the historical flow record. Moreover, the severe 921 Earthquake might have resulted in significant changes on topography and water conservation in watersheds. Stochastic flow analysis based on previous historical flow record may not adequately reflect current flow system. Consequently, previous and current rainfall data were adopted to predict the incoming rainfall, and a deterministic-based daily-flow model was used to simulate the low-flow for future available water analysis. In this project, historical rainfall was adopted to establish Markov transitional-probability matrixes which were used to predict future ten-day and monthly rainfall. The predicted rainfall was then input to the deterministic TOPMOEDL to generate daily-flow series for further ten-day and monthly flow analysis. Extreme drought conditions were also included in the analysis based on the exceed probability of continue-dry days to provide information for possible drought in the future. In considering the complex procedure in operating the Markov rainfall model and TOPMODEL for daily flow prediction, an integrated long-term flow prediction system for Yung-Ho-Shan and Ming-De reservoirs was developed by using Visual Basic. This Windows-based information system can provide user in a convenient way to predict reservoir upstream inflow for water resources management.