本文利用台灣中北部地區1991~1996年5~6月梅雨季鋒面過境之地面降水資料,求取各分區於鋒面過境前後各24小時之降雨機率、降雨量及豪(大)雨發生機率逐時氣候値。並利用1997年5~6月間六道梅雨鋒面過境台灣北部的降水資料,分別以預報起始時間前1小時、前3小時、前6小時及前12小時之降雨觀測値,作為不同持續法之預測値,與氣候値之降雨機率、降雨量及豪(大)雨發生機率進行預報校驗,以比較氣候値法與不同持續法的預報誤差。 因梅雨鋒面提供對流有利的綜觀環境與中尺度激發機制,故各分區逐時降雨機率均於鋒面過境達相對最大,而鋒面過境後呈明顯較鋒前增加之勢,各分區降雨機率均在鋒面過境時或過境後4~5小時出現最大値。因此平均降雨機率預報誤差顯示,氣候値法(CL)在各分區中的預報能力,不論鋒前或鋒後幾乎均優於6小時與12小時的持續法(PT06與PT12)。在分區1與2中,CL法在鋒前的預報能力更優於PT01與PT03,顯見CL法在鋒面過境台灣北部之基隆、台北與桃園地區前有預報參考價値。然因1997年鋒面過境前後降雨強度與豪(大)雨發生頻率均顯著較氣候値為小,故對不同時段的降雨量與豪(大)雨發生頻率的預報而言,各分區之CL法誤差均較PT法為大,此乃氣候値法之基本特徵。
Mesoscale Climatology of Meiyu frontal rainfall over northern and central Taiwan was obtained using hourly rainfall data in the period of 1991-1996.Mesoscale climatology and different persistence methods were used to evaluate the forecast skill of rainfall intensity(QPF),probability of precipitation(POP),and heavy rainfall frequency using Meiyu frontal rainfall data in 1997. Results showed that the rainfall intensity, probability of precipitation and heavy rainfall frequency were much higher during and after the frontal passage as compared to those ahead of front for each subregion. For the forecast evaluation, mesoscale climatology demonstrated some skill on 6-12 hours POP in each subregion and better skill on 3-6 hours POP in subregion 1 and 2.Results also showed that mesoscale climatology obviously provide useful information for forecasting POP in northern Taiwan during and after frontal passage. Nevertheless, rainfall intensity and heavy rainfall frequency in 1997 were smaller than climatology, therefore the mesoscale climatology forecast errors in each subregion were higher than different persistence methods.