石門水庫與翡翠水庫為北台灣重要水庫，供應北部地區主要民生、工業與農業用水，近年來經濟社會之快速發展增加水資源之需求，氣候變異則對水庫管理帶來挑戰。水資源的規劃管理需要明瞭水文量長期變化與特徵，並不一定需要使用繁雜之分散式模式來達成目的，透過年月尺度之水文模式即可估算水庫入流量長期變化，對水資源規劃與管理提供基本資料所需。本研究應用月尺度水文模式即年尺度經驗公式，估算石門集水區與翡翠水庫集水區入流量，主要分為：(1)在不考慮集水區地表下水儲蓄變化，根據Budyko(1974)假設，以水平衡方程式推估入流量年變化趨勢。(2)在月尺度部分，利用 Thornthwaite Mather Water Balance(TMWB) 模式模擬水庫集水區月尺度入流量變化。研究中使用石門水庫與翡翠水庫集水區的流量與氣象站資料，檢定模擬結果並分析年月入流量變化趨勢。結果顯示：(1) 應用Hamon潛勢能蒸發散公式與傅抱璞蒸發散估算之組合在年逕流模擬準確度較高。 (2) TMWB模式也可以掌握月逕流趨勢。但不論是年或月尺度之模式對於極端事件之模擬較難掌握，推論主要原因為潛勢能蒸發散高估。因此在潛勢能蒸發散計算中則加入折算係數，並進一步針對TMWB模式加入含水層與基流概念，提高模式對極端事件逕流模擬之掌握能力。結果顯示修正後之TMWB模式可以解決原先連續乾旱或極端降雨月份的模擬明顯低估的問題，符合二集水區實際逕流趨勢。此外考慮使用者在收集大量的水文資料之後，還需要花時間進行複雜的模式檢定工作，因此透過Microsoft Visual Studio 2012 RC，發展圖形化使用者介面(GUI)。透過介面只需輸入水文資料與簡單的操作，即可得到模擬結果與其相關程度。Both the Shihmen and Feitsui Reservoirs are the most important reservoirs in northern Taiwan providing water resources for domestic, industrial and agricultural usages. Booming social and economic developments in recent years requires increasing demands of water resources. On the other hand, climate variability brought challenges on reservoir managements. Predicting long-term variability and characteristics of reservoir inflows are essential to water resources planning and managements, which are not necessary to be estimated by distributed hydrological models. Monthly and Annual hydrological models provide alternative approach with simplicity to understand long-term inflow variations of watershed to support water resources planning and management. In this study, we applied monthly and annual water balance model to estimate inflows of the Shihmen and Feitsui Reservoirs. For annual scales, the water balance equation was given based on the Budyko assumptions by neglecting changes of subsurface storages. For monthly scales, the Thornthwaite - Mather Water Balance (TMWB) model was employed to estimate reservoir inflows.In this study, we collect the Shihmen and Feitsui Reservoir inflow and meteorological data for model calibrations. Result showed that the combination of Hamon equation (for estimating potential evapotranspiration) and Fu equation (for estimating evapotranspiration) provided the best skill on estimating annual reservoir inflows. The TWMB has skills on estimating monthly reservoir inflows. However, both monthly and annual water balance approaches underestimated inflows of extreme events. Overestimations of potential evapotranspiration were suggested to be the cause of less skill on extreme events. A reduction coefficient was proposed to modify the amount of potential evapotranspiration estimated in both monthly and annual water balance models. For the TMWB model, a layer of subsurface storage was added with a base flow coefficient. Results showed modifications proposed slightly improve the skills of both approaches on estimating reservoir inflows.In order to ease complicated calibration processes, a Graphical User Interface of models employed in this study was developed via the Microsoft Visual Basic 2012 RC to provide a user friendly interface with simultaneous and instantaneous graph drawing capability.
