本研究主要目的在於整合中央氣象局針對ECMWF展期雨量預報之降尺度產品(未來1~30日),以及氣象局自行發展之全球海氣耦合模式長期雨量預報產品(未來31~180日),輸入修正型HBV水文模式進行全臺重點水庫入流量預報,以輔助抗旱期間水庫未來蓄水量推估與水情研判。面對短期迅速變化之天氣型態(如梅雨與颱風)而言,在相同預報時段內展期雨量預報產品之準確度會優於長期雨量預報產品。經本研究採用ECMWF歷史雨量預報降尺度資料(2000~2019)進行歷史流量預報,亦證實預報第1~3旬流量結果優於採用長期雨量預報進行第1~3旬流量預報結果,因此整合展期與長期雨量預報產品進行流量預報,確實可顯著增加未來第1~3旬之流量預報效能。基此,對於水庫入流量預報未來1~30日可採用ECMWF展期雨量預報降尺度產品,而未來31~180日則採用長期雨量預報降尺度產品,整合前述兩種雨量預報產品輸入水文模式進行未來1~180日水庫入流量預報,除能提升未來3旬面對天氣型態快速變化之能力外,亦能兼顧長期未來第2~6個月之水庫入流量發展情勢。
This work integrated the calibrated ECMWF's extended-range (1 ~ 30 days ahead) and the CWB1T1's seasonal (1 ~ 6 months ahead) rainfall forecasts as the input of the HBV-based hydrological model for performing the inflow forecasting at the key reservoirs in Taiwan. The reservoir inflow forecasting can facilitate to predict the future reservoir storage and support decision-making during the drought period. In the face of short-term rapidly changing weather conditions (such as the mei-yu front and typhoon), the extended rainfall forecasts generally perform more precisely than the seasonal rainfall forecasts during the same forecasting period. This work used the calibrated (downscaled and bias-corrected) ECMWF's extended-range rainfall re-forecasts (2000 ~ 2019) as the input of the HBV-based hydrological model for forecasting the reservoir inflows. The results reveal that the reservoir inflow re-forecasts by calibrated ECMWF's extended-range rainfall re-forecasts have a better performance than those by the CWB1T1's seasonal rainfall re-forecasts for the 1 ~ 30 days lead time. Therefore, the integration of extended-range (1 ~ 30 days ahead) and seasonal (2 ~ 6 months ahead) rainfall forecasts for reservoir inflow forecasting can simultaneously reflect the short-term weather conditions as well as the long-term trend of the weather outlook.