| 摘要: | 台灣地區由於豐枯水期降雨量懸殊、新水源開發困難、水庫泥沙淤積嚴重、各標的用水逐漸增加等因素，乾旱時期的水資源調度已成為一重要議題。因此，若能建立有效的乾旱預警機制與資訊，提供相關管理單位決策參考外，亦可減緩缺水乾旱對經濟、社會與農業之衝擊。本研究運用不同的乾旱指標BMDI(Bhalme and Mooley Drought Index)、PDSI (Palmer Drought Severity Index)及SWSI (Surface Water Supply Index)分別作為氣象、農業及水資源乾旱的評估，藉由乾旱指標所提供的資訊結合中央氣象局氣候統計降尺度預報產品，以提供作為乾旱預警、應變及決策的參考依據。
研究區域為石門水庫供水區，以1976~2010年的氣象及水文資料做為計算乾旱指標所需的資料基礎，並分析上述不同乾旱指標之特性。選定歷史乾旱年2002年和2003年作為乾旱預警的研究事件，運用中央氣象局氣候統計降尺度預報透過氣象資料合成模式，繁衍出符合預報趨勢之未來三個月日溫度及日雨量資料，再利用水文模式及系統動力模式進行水庫蓄水量推估，進而計算未來三個月之各乾旱指標，配合不同的乾旱嚴重程度擬定供水策略。
研究結果顯示，在分析石門供水區的歷史乾旱事件都有一定的掌握度，於指標特性的部分氣象乾旱指標BMDI易受到雨量的影響而提早結束乾旱期；在極端降雨事件後，PDSI判斷農業乾旱發生的時間點相較於其他的乾旱發生時間點來的晚
SWSI受限於石門水庫庫容較小的關係以至於部分乾旱事件發生的時間點與氣象、農業乾旱發生的時間點極為接近。於乾旱指標結合氣候統計降尺度預報資訊之應用，依據本研究研擬的供水參考依據，可有效的減輕旱情，且付出的休耕補償金相對於歷史上較少，因此可提供決策者預警與管理之參考依據，提早作出較合理的水資源調配應變措施。
Because of significant rainfall discrepancy between wet and dry seasons, difficulties in developing new water resources, serious sedimentation in reservoirs, increases in water demand and so on, effective management of water resource become an important issue in drought periods. Therefore, effective drought early warning is critical to provide drought information to help government agencies for decision making and to reduce impacts to economics, society, and agricultures. In this study, BMDI (Bhalme and Mooley Drought Index), PDSI (Palmer Drought Severity Index) and SWSI (Surface Water Supply Index) were used to access the meteorological, agriculture and water resources droughts, respectively, in association with statistical downscaling climate forecasts from the Central Weather Bureau to provide drought early warning as references for responses and decision makings.
The study area is the water supply district of the Shihmen reservoir. Both meteorological and hydrological data from 1976 to 2010 were used to calculate the three drought indices for the study area to examine whether past drought events can be captured by these indices. Severe droughts in year 2002 and 2003 were used to investigate our approach of combining drought indices with climate forecasts. A weather generator was used to provide daily rainfall and temperature to drive a hydrological model to obtain daily inflow of the Shihmen Reservoir. Then a system dynamic routing of reservoir operation was used to calculate the daily reservoir storage. Based on drought indices calculated with a lead-time of three months, strategies for reducing water supply in association with different drought severities were proposed.
Based on results in this study, the BMDI is easy to be affected by rainfall and drought duration predicted by the BMDI is shorter than the real cases. After extreme rainfall events, the starting time of agricultural drought predicted by the PDSI is lagged. Due to small reservoir capacity in the Shihmen Reservoir, onsets of droughts predicted by the SWSI were similar to those by the other two indices. By combining statistical downscaling climate forecasts and drought indices, the water reduction strategies proposed in this study can effectively mitigate drought severities and reduce amount of fallow compensation. With advantages of climate forecast, the drought early warning is ready to support decision making for water resources allocations and responses measures. |
