都卜勒雷達在颱風環流結構與平均風反演之應用

Abstract

地基速度軌跡顯示法(GBVTD)在許多登陸台灣的颱風研究中，已被證明能有效的反演颱風運動結構特徵。然而其所使用的反演座標以及忽略平均風垂直雷達波束分量的假設，在平均風垂直雷達波束分量較大的情況下，會造成反演結果有較大的誤差。周等(1996)進一步發展速度距離乘積方位顯示法(VDAD)，在VDAD座標下，平均風呈一組平行線，識別平均風遠比在GBVTD座標下容易。因此，本文以VDAD為基礎，發展一套利用雷達觀測資料反演平均風的疊代方法。 本文將理想阮肯渦旋嵌入在預設之平均風環境裡，對此疊代算法進行測試。結果顯示，反演之平均風大小與方向均與預設值相當接近，風速誤差約1 m s-1，風向誤差則在10°以內。反演結果的誤差則是對於渦旋中心距雷達之距離、資料中的雜訊、以及軸對稱徑向風較為敏感。 本文同時也將此平均風疊代法應用於納坦颱風(2004)。利用五分山和花蓮雷達觀測資料，分別透過GBVTD計算其軸對稱切向風，結果顯示，兩座雷達計算之軸對稱切向風最大值相差達14 m s-1，而這樣的差異可歸因於納坦颱風平均風垂直雷達觀測分量較為顯著所造成。進一步將此平均風疊代法應用於納坦颱風，結果顯示，兩座雷達計算之軸對稱切向風最大值差異顯著降低至1.2 m s-1。配合討論垂直風切對於降雨分布以及納坦颱風接近台灣地形時路徑轉變的可能影響，也論證了此平均風疊代法之重要性。

Ground-based velocity track display (GBVTD) technique has been proven to be effective on retrieving the kinematic structure of landfalling typhoons in Taiwan. However, significant errors are found due to the limitation of GBVTD technique especially when the environmental mean wind component perpendicular to the line connecting the radar and the typhoon center is large. A new iterative method, based on the velocity distance azimuth display (VDAD) technique proposed by Jou et al. (1996), has been developed to solve the environmental mean wind from the observed radial velocity data. In this thesis, an iterative method is designed for solving the environmental mean wind. This method is tested by setting up an experiment that an ideal Rankine vortex is embedded within a pre-assigned uniform mean wind environment. The results show that the error of retrieved mean wind speed is less than 1 m s-1 and the mean wind direction is less than 10°. These errors are sensitive to the distance between the vortex center and the radar, the noise of the data, the, the mean tangential wind speed, and the mean radial wind speed. The mean wind solver has been applied to a real typhoon case, Nock-Ten (2004). Using Hwa-Lien (HL) and Wu-Feng-Shan (WFS) radar data set respectively, the mean tangential wind component of Nock-Ten has been retrieved by original GBVTD method. It is found that the retrieved maximum mean tangential wind from HL is 14 m s-1 less than that solved by WFS data set. This large difference has been attributed to significant mean wind component perpendicular to the line connecting the radar and the typhoon center. After applying the newly developed mean wind solver to the data set, the difference of the retrieved maximum mean tangential wind has been reduced to 1.2 m s-1 which is significantly less than what was calculated originally. The importance of the mean wind solver is also demonstrated by discussing the mean vertical wind shear on precipitation distribution and the possible influence on the track of the storm while it is very close to the Taiwan topography.