納莉(2001)颱風之數值模擬研究

文化大學機構典藏 CCUR > 理工學院 > 理學院 > 地理學系 > 博碩士論文 > Item 987654321/27402

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: https://irlib.pccu.edu.tw/handle/987654321/27402

题名:	納莉(2001)颱風之數值模擬研究 A Modeling Study of Typhoon Nari (2001)
作者:	黃小玲
贡献者:	地學研究所
关键词:	納莉颱風降水 Typhoon Nari precipitarion
日期:	2006
上传时间:	2014-06-04 15:27:51 (UTC+8)
摘要:	本研究藉使用PSU-NCAR MM5模式來模擬納莉(2001)颱風之路徑、強度及降水結構。控制組(CTL)實驗模擬結果發現，納莉颱風的路徑模擬與氣象局的觀測路徑比較相當接近。模擬颱風於9月16日1000 UTC在基隆登陸，而實際觀測颱風於9月16日1300 UTC在東北角登陸，登陸時間誤差為3小時，登陸地點誤差約為30公里。控制組(CTL)實驗模擬納莉颱風於9月17日的24小時累積降水比較，得知最細網域(2公里)的模擬全台灣之24小時累積降水面積平均達觀測資料的80.7 %。 CTL實驗顯示，納莉颱風結構於登陸前後有明顯的差異。颱風登陸前（於海上）呈現垂直直立的結構，其凝結(condensation)及凝固(deposition)加熱率分布在颱風眼牆的中至高層大氣中；蒸發(evaporation)冷卻率主要分布在近地面1公里高度以下，其水平範圍超過100公里；雨水、雪花和軟雹混合比之垂直剖面分布，集中在颱風眼牆及螺旋雨帶結構。颱風登陸後由於受到台灣複雜地形的影響，使得颱風呈現非軸對稱且隨地形傾斜的結構，其凝結及凝固加熱率分布在颱風眼牆的低至中層大氣；蒸發冷卻率主要分布在山區，水平範圍約為40公里，且強度較強。登陸後之雨水、雪花和軟雹混合比之垂直剖面分布，集中在颱風眼牆，另於中南部山區的雨水混合比強度亦較強。本研究之模擬結果與雷達觀測資料的比較，發現控制組(CTL)實驗計算出來的雷達回波強度與徑向風風場分布，皆與五分山雷達觀測資料有不錯的符合度；距離颱風中心附近45公里距離的水平氣壓梯度（溫度梯度）約有9 ~ 10 hPa (5~ 6 K)，而經由雷達資料透過熱動力反演方法求得的氣壓梯度（溫度梯度）約有5 ~ 6 hPa (3 ~ 4 K)，兩者的數量級大致接近，且都有一明顯的暖心颱風結構。為瞭解大氣場之水氣和降水粒子的雲微物理轉換過程，對降水效率的影響，本研究參考Sui et al. (2005)之降水效率方法，分析三維MM5模式模擬納莉颱風於海上時期的降水效率資料。模擬結果發現水氣透過水氣凝結成液態水，或凝固成冰相水的轉換過程形成降水粒子，使得大尺度降水效率(large-scale precipitation efficiencies; LSPE )幾乎與雲微物理降水效率(cloud-microphysics precipitation efficiencies; CMPE)相當。於強降水區域的降水效率皆可達50 %以上，而弱降水區域之降水粒子輻合效應提供了正向貢獻，使得弱降水區域之降水效率可達到100 %以上，甚至可超過300 % 。由地形敏感度實驗測試結果，發現75 % Ter、50 % Ter、 25 % Ter及NoTer四組地形敏感度實驗之全台灣於9月16日之24小時累積降水量的面積平均，分別為控制組(CTL)實驗的75.88 %、81.88 %、64.13 %及45.40 %，由此可知台灣地形效應對於颱風降水研究的重要性。另於觀測、CTL、地形及雲微理敏感度實驗的降水頻率百分比分布，發現模式對於弱降水高估（24小時降水累積 < 400公厘）及強降水（24小時降水累積 > 400公厘）低估的結果，與Colle et al. (1999), Chien et al. (2002), 及Yang et al. (2004)的研究結果相近。 In this study, the PSU-NCAR MM5 model is used to investigate the track, intensity structure and precipitation processes associated with Typhoon Nari. The simulated track of the Control (CTL) experiment is much close to the observed track, and the landfall error is about 30 km with 3 hour timing error. The simulated rainfall averaged over Taiwan on the 2-km domain can reach 80.7 % of the observed amount on September 17. The CTL run shows that the typhoon structure has significant differences before and after landfall. Before landfall, the condensation latent heating is distributed at mid-to-high levels with typhoon’s eyewall, and the evaporation cooling is distributed at the low levels with wider areas. The rain water, snow and graupel are distributed near the eyewall and spiral rainband. After landfall, the typhoon structure is more asymmetric and has evident titling by the topography. The condensation heating rate is distributed at low-to-mid levels, and the evaporation cooling rate is distributed along the mountain range with 40-km width. We have compared the simulated radar reflectivity versus the observed one from the NEXRAD Doppler radar located over the WuFeng Mountain (RCWF). The observed radar reflectivity, tangential wind, and radial wind velocity are well simulated by the MM5 CTL run. A horizontal pressure (temperature) gradient of 9 ~ 10 hPa (5 ~ 6 K) within 45 km across the eyewall is simulated, in good agreement with the radar-derived pressure (temperature) gradient of 5 ~ 6 hPa (3 ~ 4 K) using a thermodynamic retrieval method. We follow Sui et al. (2005) to perform a similar examination of cloud-microphysics and vapor budgets from the 3D simulation of Typhoon Nari (Yang and Huang 2004). When the hydrometeor convergence becomes the dominant term in the cloud budget, the CMPE can be larger than 100% in light-rain conditions (Ps < 5 mm h-1). On the other hand, a loss of clouds due to hydrometeors diverging out to the neighboring columns would make the CMPE smaller (50 ~ 75 %). This occurs mostly in heavy rain conditions (Ps > 5 mm h-1). Four terrain sensitivity experiments are conducted with 75%, 50%, 25%, and 0% of the actual Taiwan orography and four microphysics sensitivity experiments are conducted as well. The terrain effect on the storm track is very nonlinear, and the enhancement of orographic precipitation is increased with higher terrain. We find that the model tends to over-predict the occurrence of light rainfall (24-h rainfall < 400 mm) and under-predict the occurrence of heavy rainfall (24-h rainfall > 400 mm) for all topographic and microphysical sensitivity experiments.
显示于类别:	[地理學系] 博碩士論文

文件中的档案:

没有与此文件相关的档案.

检视Licence

在CCUR中所有的数据项都受到原著作权保护.

数据加载中.....