頁籤選單縮合
題名 | 中央氣象局颱風路徑預報模式之預報檢討及改進=A Study of the Evaluation and Improvement of the Central Weather Bureau Typhoon Forecast Model |
---|---|
作者 | 葉天降; 陳得松; 黃康寧; 謝信良; Yeh, Tien-chiang; Chen, Der-song; Huang, Kang-ning; Shieh, Shinn-liang; |
期刊 | 氣象學報 |
出版日期 | 20020900 |
卷期 | 44:3 2002.09[民91.09] |
頁次 | 頁31-75 |
分類號 | 328.888 |
語文 | chi |
關鍵詞 | 颱風路徑預報模式; 高解析度颱風路徑預報模式; 輻射觀測資料; Typhoon track forecast model; High resolution model; Radiance data; |
中文摘要 | 中央氣象局颱風路徑預報模式(TFS)自1997年改進虛擬颱風渦旋結構初始化處理後,預報路徑之準確度有相當明顯改善、本文主要包括整理與校驗颱風路徑預報模式2000興200l年颱風季之預測結果、分析高解析度颱風路徑預報模式之預測表現、並進行降水預測模擬、同時測試輻射值(radiance data)觀測資料應用於模式之情況等四大項工作。 校驗分析結果,中央氣象局颱風路徑預報模式在2000與2001年颱風季之24/48/72小時距離預測誤差各為148/306/520、155/312/518公里,相對於CLIPFR而言有預測「技術」,唯第三天的預測誤差增幅有比第二天的誤差增幅增大的現象。 在高解析度颱風預報模式測試方面,以2000年之象神(Xangsane)颱風、貝碧佳(Bebinca)颱風,2001年納莉(Nari)颱風等三颱風共21個案做平行測試,結果顯示,巢狀網格颱風路徑預報模式細網格模式之24/48/72小時平均預測誤差為106/219/311公里,較粗網格及當時作業颱風路徑預報模式之預測誤差130/234/353及149/285/475公里小。顯示提高模式解析度及改進虛擬渦旋結構初始化處理程序可有效減低颱風路徑預報模式之預測誤差。 在降雨預測模擬測試方面,以更新地形資料後,15公里解析度颱風預報模式對納莉颱風降水之模擬雖降水量偏少,但降水分型態頗類似,顯示颱風預報模式在使用較高解析度與正確地形資料後,已有預測累積降水分之能力。 在增加輻射觀測資料對預報影響之模擬測試方面,結果顯示,TFS之預測路徑對初始場相當敏感,使用NCEP觀測資料組及SSI分析法可降低TFS路徑預測誤差,模式引進輻射值資料對路徑預測有正面效應。 |
英文摘要 | This study explores and tries to improve the performance of the Typhoon track Forecast Model (TFS), which is one of the key elements in typhoon forecast operation at the Central Weather Bureau. In this year, four primary tasks were accomplished: the evaluation of the performance of TFS for typhoon track forecast in 2000 and 2001, the evaluation of the performance for typhoon track forecast by high resolution models, the study of models' typhoon rainfall simulations, and the study of the impact of radiance data on TFS forecast. From the evaluation of the TFS performance for typhoon track forecasts in 2000 and 2001, we found that the average track errors of the 24/48/78 hours forecast are 148/306/:520 and 155/312/518 kin, respectively. The results showed TFS is skillful as the errors are all smaller than those of CI.IPER forecast for homogeneous cases. The results of the evaluation of the performance of track forecast by the high resolution models showed that increasing the model resolutions and improving the initial vortex structure may reduce the track error. For the 21 cases from three selected typhoons, the average 24/48/72 hours track errors of the 15 km fine grid simulation were 106/219/311 km, which were 14/15/42 kin, respectively smaller than those of the 45 km coarse grid simulation and were 43/66/164 km smaller than that of the operational forecast. Our result of the rainfall simulation study showed that the high resolution model with fine grid over topography was capable of displaying the spatial distribution of precipitation for typhoon Nari, although the amount of the rainfall was smaller than that of observations. In addition, the result of the impact study of the usage of the radiance data showed the TFS forecast was sensitive to its initial conditions. The track forecast error was smaller when the NCEP observations and the SSI analysis method were applied. To include the radiance data in the analysis had positive impact on the model forecast. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。