為了統計降尺度定量降水推估的改進和未來朝向氣候變遷降水研究之目的。因此本研究利用綜觀天氣分型結合邏輯和非線性迴歸建立統計降尺度降水模式並評估其在台北測站之效用與改進。 本研究使用了臺北測站的地面觀測資料和NCEP高空再分析資料來建構降水模式，研究包含下列兩個主要部分: (1)自動天氣分型，及(2)發展降水天氣型之降水推估模式。天氣分型的結果可以歸納出7類和降水有關的天氣型態，分別為冷季的冷鋒、滯留鋒Ⅰ、滯留鋒Ⅱ、暖季滯留鋒、熱帶低壓、局地熱對流及颱風。結果顯示，有進行天氣分型且搭配邏輯迴歸的機率預報和非線性之降水量推估與不進行天氣分型的降水推估結果相比是有改進的，尤其是小雨(<7.5mm)推估為優、良的比例有5~10%提升，大雨(>32.5mm)因變異度大改善有限甚至不理想。天氣類型中暖季滯留鋒為間歇性的中尺度對流降水，統計模式對於大部分的移動性對流降水較難以掌握，因測站還未能反應移動的鋒面對流系統天氣變量時，對流系統已帶來降水，導致迴歸效果不佳。檢驗後發現，熱帶低壓容易與輕颱混淆，且颱風不適用此天氣分類方法，導致檢驗時的降水推估表現不佳，除此之外其他天氣類型都有良好的改善。為改善颱風類型的降水推估，獨立挑出颱風個案改以路徑分類做測試，結果有很好的改善。 本研究利用綜觀天氣分型搭配颱風路徑分類所建立的統計降尺度日降水估計模式，其在台北測站之應用成效良好，未來經進一步改善，應可以應用於氣候變遷降水推估研究上。 ;The purpose of this study is to improve statistical downscaling of quantitative precipitation forecast and use these daily rainfall simulation models to project changes in frequency and magnitude of future daily rainfall. Therefore, an automated synoptic weather typing and stepwise cumulative logit/nonlinear regression analyses were employed to estimate the occurrence and quantity of daily rainfall events in Taipei station. Taipei station hourly and daily observed and NCEP reanalysis weather data for each year of 1992-2012 without 1997 are used in this study. The analyses are divided into four steps: (i) automatic synoptic weather typing, (ii) identification of weather types associated with rainfall events, (iii) development of within-weather-type rainfall simulation models, and (iv) validation of the rainfall simulation models using an independent dataset. The 7 rainfall-related weather types are cold front, quasi-stationary front I, quasi-stationary front II in cold season (November- April), quasi-stationary front III, tropical low, local convection and typhoon in warm season (May-October). The results show that within-weather-type rainfall simulation models demonstrated significant skill in the discrimination and prediction of the occurrence and quantity of daily rainfall events with exceptions of localized convective storms. The percentage of excellent and good simulations for the light rainfall events improved by 5~10%, but the heavy rainfall simulations have no significant improvement because of greater variability. At the time of raining, the weather variables at the station didn’t reflect the characteristics of the moving convective system, so statistical downscaling models cannot capture moving and mesoscale convective rainfall. Typhoon is not suitable for automatic synoptic weather typing and it is easily mix up with tropical low. Then we use typhoon path classification methods and the simulation results will be better. The results from this study show that a combination of synoptic weather typing and cumulative logit/nonlinear regression procedures can be useful to simulate historical daily rainfall occurrence and quantity in Taipei station. If there are more improvements in these models, the statistical downscaling method can be applied to study occurrence and quantity of daily rainfall events under climate change in the future.
