雲解析模式對臺灣梅雨季豪大雨定量降水預報技術之評估研究

dc.contributor王重傑zh_TW
dc.contributorWang, Chung-Chiehen_US
dc.contributor.author莊璧瑜zh_TW
dc.contributor.authorChuang, Pi-Yuen_US
dc.date.accessioned2019-09-05T00:47:40Z
dc.date.available2015-09-15
dc.date.available2019-09-05T00:47:40Z
dc.date.issued2015
dc.description.abstract本論文為探討雲解析模式對臺灣梅雨季豪大雨的定量降水預報技術。在2005年之前的梅雨季12-h定量降水預報技術評估中,其結果表示,模式對小雨雨量門檻的預報技術較有掌握,而對中雨及大雨雨量門檻的預報技術不佳(ETS介於0.05~0.15)。 本論文依降雨事件的雨勢規模之不同,將其設計成七個組別,並使用五種技術得分(TS、BS、POD、FAR及OR)和設定13個雨量門檻(0.05、2.5、10、25、50、75、100、130、160、200、250、350及500 mm),評估2012~2014年臺灣梅雨季各組的三天24-h定量降水預報技術,而2.5 km和5 km網格間距的模式預報技術之評估結果較過去十年為佳。 針對不同雨量門檻累計的所有項目(hit、false alarm、miss及correct negative)之總雨量站數後,再計算設計組別的技術得分,由評估結果知,在各雨量門檻,模式對雨勢規模較大的降雨事件之預報技術較雨勢規模較小者為佳,即「當降雨事件的雨勢規模越大時,模式的定量降水預報技術越佳。」且在中高雨量門檻(50~500 mm),模式對規模大的降雨事件之預報技術也較佳,即對可能成災的豪大雨事件預報技術較佳。 本論文另比較2.5 km和5 km網格間距的模式預報技術,其結果表示,當模式解析度提高時,對雨勢規模較大的降雨事件有較佳的預報技術,尤其是在中高雨量門檻(50~500 mm)。同時,在規模大的豪大雨預報不足方面,2.5 km較5 km網格間距的模式預報有所改善,且隨模式預報時間拉長,預報技術降低的速率也較慢。因此,「當模式的解析度提高時,模式對規模大的豪大雨事件之定量降水預報技術較佳。」zh_TW
dc.description.abstractThis study is focused on the quantitative precipitation forecasts (QPFs) for heavy-rainfall events in Taiwan mei-yu season by a Cloud-Resolving Storm Simulator (CReSS). According to the result of the 12-h QPFs evaluation before 2005, the ensemble model didn't catch the skill at the medium and high thresholds but the low thresholds. However, there is a better result of the 24-h QPFs by CReSS than before in this study. After the category of rainfall events by the rainfall among and magnitude, we use 5 skill scores (TS, BS, POD, FAR, and OR) to evalue the QPFs for each groups at 13 thresholds (0.05, 2.5, 10, 25, 50, 75, 100, 130, 160, 200, 250, 350, and 500 mm) in Taiwan mei-yu season form 2012 to 2014. The result in this study is that CReSS is more skillful when the rainfall among and magnitude of events is higher and bigger, especially at 50-500 mm threshold. What's more, we compare the different horizontal resolution of the QPFs for heavy-rainfall events by CReSS. And the interval of grids in CReSS is 2.5-km and 5-km. The result of the comparison is that CReSS is more skillful when the horizontal resolution is higher, especially at 50-500 mm threshold. Besides, the higher resolution CReSS is, the less underforecast will be at 50-500 mm threshold.en_US
dc.description.sponsorship地球科學系zh_TW
dc.identifierG060144026S
dc.identifier.urihttp://etds.lib.ntnu.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=id=%22G060144026S%22.&%22.id.&
dc.identifier.urihttp://rportal.lib.ntnu.edu.tw:80/handle/20.500.12235/101155
dc.language中文
dc.subject臺灣梅雨季zh_TW
dc.subject定量降水預報zh_TW
dc.subject豪大雨事件zh_TW
dc.subject技術得分zh_TW
dc.subjectTaiwan mei-yu seasonen_US
dc.subjectquantitative precipitation forecasten_US
dc.subjectheavy-rainfall eventen_US
dc.subjectskill scoreen_US
dc.title雲解析模式對臺灣梅雨季豪大雨定量降水預報技術之評估研究zh_TW
dc.titleAn Evaluation of Quantitative Precipitation Forecast Skill for Mei-Yu Heavy-Rainfall Events in Taiwan by a Cloud-Resolving Modelen_US

Files

Original bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
060144026s01.pdf
Size:
7.41 MB
Format:
Adobe Portable Document Format

Collections