本研究分析1991∼2006年5∼6月之134個梅雨鋒面影響台灣期間,台灣地區降雨特徵隨梅雨鋒面移動之變化情形,並以1997∼2006年中央氣象局作業使用的綜觀預報檢查表資料,分析台灣西部6個主要流域之強降雨與綜觀環境之關係。結果顯示,鋒面過境本島時,較大雨勢常出現於盛行西南風之迎風面,其中境內地形與西南風有交角的高屏溪強降雨機率最高,顯示鋒面雖提供舉升機制,但台灣地形之強迫作用更為重要。當鋒面位於台灣北部時,淡水河流域之強降雨發生機率高,顯示鋒面舉升作用與該流域強降雨關係密切。 以統計方法篩選與各流域強降雨相關之重要綜觀因子顯示,梅雨鋒面接近、潮濕環境、低層噴流,以及700/500 hPa有短波槽等項目,與各流域強降雨有較高相關。北部流域12小時強降雨主要需借助高層噴流條提供輻散,以維持對流發展;南部流域強降雨則與台灣位於500-1000 hPa厚度場分流區的關聯性佳。為評估氣候概念模式應用於梅雨定量降雨預報之可用性,本研究應用前述結果建立梅雨季強降雨氣候概念模式,並以2007∼2008年17個獨立個案校驗之。結果顯示,該模式有助於改進氣候模式在較強降雨的預報,並在中南部流域有較佳的預報表現。對中南部流域而言,該模式6月之預兆得分遠高於5月,北部則無明顯趨勢。前估與後符的分析顯示,此模式有過度預報的情況,然而即使整體預報表現尚有改進空間,此模式仍可於數值模式預報結果尚未穩定或各數值模式預報結果有所分歧時,提供額外之參考資訊。
This study focused on the characteristics of rainfall distribution during Mei-yu season in 1991-2006. Operational heavy rainfall checklist data from Central Weather Bureau were also used to study the relationship between heavy rainfall and synoptic environment conditions within 6 selected river basins. The result showed that the heavier rainfall occurred frequently over windward side of southwesterlies. Though the Mei-yu front provided lifting mechanism, the forcing from the Taiwan topography also played an important role. Synoptic environment conditions highly related to the heavy rainfall events filtered by statistic methods included: approaching of Mei-yu fronts, moist environment, low-level jets and 700/500 hPa short wave troughs. The result also indicated that 12-hourly heavy rainfall over northern Taiwan relied on the divergence provided by upper-level jet streaks to help convection developing while heavy rainfall over sourthern Taiwan was highly related to 500-1000 hPa diffluent thickness pattern. Based on the above analyses, this study established a conceptual climatology model and tried to estimate the model usability to the quantitative precipitation forecast (QPF). The validation was done by 17 independent cases during 2007-2008. Result showed that the conceptual climatology model performed better on predicting heavier rainfall than climatology model especially for cenrtral and southern river basins. Analysis of prefigurance and postagreement indicated that the conceptual climatology model was over prediction. However, when the results of numerical weather prediction models are still unstable or disagree with each other, the model can still provide an additional infomation.