摘要: | 本研究首先應用無母數統計檢定方法分析石門水庫集水區1971~2010年的長期變異程度。研究結果顯示石門水庫集水區多數水文氣象觀測的變異點時間都約在1977~1983年之間。M-K趨勢檢定結果顯示石門水庫集水區年降雨量為上升趨勢,和颱風降雨上升有很大的關係。應用中央氣象局發布的短期氣候預報結合氣候資料產生器、水文模式及系統動力模式探討春雨期間(2至4月)雨量變化及其降雨、逕流對石門水庫水資源之影響。結果顯示預報在雨量有高估的現象,而溫度部分有低估的現象。在入流量的預報推估值部分,受到雨量預報推估值的影響產生偏大的現象,因此水庫水位推估值也高於實際觀測值。統計降尺度預報lead 1的三月預報的模擬水位與觀測值吻合度極佳,原因為預報入流量推估值遠高於觀測值,M5規線也使水庫放較多水,而實際上2015年入流量小、有限水措施,故發生兩者水位吻合的巧合。研究再採用歷史資料模擬建立缺水時期DPD的分級級序,再利用不同二、三、四月降雨等級及不同水庫初始水位組合所模擬的春雨時期缺水DPD來代表水資源脆弱度,結果顯示倘若二月份降雨為偏高等級時,水庫於春雨時的水資源較二月份降雨為正常等級與偏低等級時豐沛。而當有兩個月等級為偏低,在二、三月同時為偏低等級時所造成的負面衝擊最大,其次為二、四月,最後為三、四月。;In this study, several statistical tests including Mann-Whitney-Pettitt, Kruskal-Wallis variance analysis and Mann-Kendall trend test were employed to examine whether significant shifts and trends of hydrological and meteorological data can be identified with long-term observational records from 1971 to 2010. This is helpful to understand characteristics of regional hydrological and meteorological variations. Using seasonal outlooks provided by the Central Weather Bureau (CWB), a weather generator was used to generate multimembers of daily weather data to drive the hydrological model to estimate future inflows. Then a system dynamic model was employed to compute projected reservoir storages and available water supplies. The Deficit Percent Day (DPD) Index is further employed to assess variations of carrying capacity of the the Dahan River water supply system. Results of statistical analyses indicate the shift of most hydro-meteorological observations occurred around 1977 to 1983. It is therefore concluded that there are significant variations in hydrometeorology of the Shihmen reservoir watershed for the past 30 years. Regarding the trends, the increasing trends in yearly precipitation, wet season precipitation, typhoon season precipitation, the number of rainy days with 0 to 0.49 mm/day, the number of rainy days with 81 to 200 mm/day, and most station precipitation were found. On the other hand, the decreasing trends in number of rainy days with 0.5 to 10 mm/day, air temperature of 10-year and 20-year time series, pan evaporation were found. The increasing trend of yearly precipitation is strongly correlated with the increase of precipitation during typhoon seasons. Besides, the number of rainy days for both 0 to 0.49 mm/day and 81 to 200 mm/day were found increase indicating changes of rainfall patterns toward unproductive rains and torrential rains. Precipitation projected with the CWB climate forecasts are all greater than observations and the temperature forecasts are underestimated in the study case of 2015. The estimates of inflows and reservoir stages are both higher than observed values as higher projected rainfall. The lead-1 water level simulations with the statistical-downscaling climate forecasts released in February are close to observed values, which is caused by higher inflow will lead to higher discharge and is not the real match of projections. The DPDs of the Dahan River water supply system were given with three grades determined by two thresholds of exceedance probabilities of 0.3 and 0.7 by using historical inflows as indices of drought risks. Combining different reservoir initial stages and all possible projected rainfalls in February, March, and April, drought risks of each combination can be evaluated in advance. When reservoir stage is below 225m, it is suggested to implement restricted management measures no matter what kind of projected rainfalls. When the projected rainfall of February is normal, there is high possibility of the DPD greater than 1000 in conjunction with reservoir stages below 230m. When the projected rainfall of February is below normal, there is high risk of water shortage. Under the conditions of having below normal rainfall projections for two months, the case occurred in February and March will lead to higher risk of drought than the case occurred in February and April, and less risk for the case of March and April. With graded DPD indices, it can help better decision making for water resources agency during drought. |