氣候變遷下台灣颱風豪雨之變化與機制探討

The Mechanism of Taiwan Typhoon Extreme Rainfall Variation under Climate Change

蘇世顥(Shih-Hao Su) ； 陳郁涵(Yu-Han Chen) ； 楊憶婷(Yi-Ting Yang) ； 徐理寰(Li-Huan Hsu) ； 郭鴻基(Hung-Chi Kuo)

《大氣科學》 45卷4期 (2017/12) Pp. 305-331

https://doi.org/10.3966/025400022017124504002

摘要

本研究利用中央氣象局21個長期雨量測站，分析1960-2015年（56年間）共91個西行且穿過台灣陸地之颱風，回顧並整理台灣颱風豪雨事件之研究成果。在記錄中累積降雨量前15名颱風中有10個發生2004年之後，且在颱風接近台灣的過程中，無論在登陸前、登陸期或是離陸後的階段，近30年的颱風平均累積降雨量與平均降雨強度與遠30年相比都有增加的趨勢。其中登陸前與登陸期的降雨平均強度增加15%，但離陸後平均降雨強度更是顯著增加34%。登陸前顯著降雨的颱風其打轉之路徑會增加颱風影響延時，導致颱風極端降雨量較多。登陸期降雨受颱風環流與地形交互作用影響，慢速且登陸位置偏北之颱風，透過被地形鎖定的對流導致的不對稱非絕熱位渦趨勢，將會使颱風移速減慢並使延時增加，導致更多颱風極端降雨，與颱風本身強度之相關性較低。颱風離陸後之降雨強度會正比於西南季風提供之水氣通量，而颱風極端降雨量則受到水氣通量與延時影響，以偏中間之颱風路徑(Type-C)的相關性最顯著，而近30年路徑C的颱風數量增加，也反應在颱風離陸後極端降雨的增加。分析顯示颱風極端降雨事件的頻率與累積極端降雨量除了上升趨勢外，亦具有顯著的週期性變化訊號。兩者除了有很大之年際變化外，並同時具有明顯的年代際變化訊號。另一方面，全球暖化之熱力機制所導致的極端降雨氣候變遷訊號，則受地形效應的影響而被掩蓋。長期氣象測站的空間分佈對於颱風極端降雨事件分析的影響，會低估山區的極端降雨，但長期大範圍的降雨分佈特徵具有一致性。換言之，只要能維持測站空間分布的均勻性，則不會影響極端降雨之變化趨勢與週期振盪特徵，而資料記錄時間的長度應是優先考量之因素。

關鍵字

颱風極端降雨 ； 颱風運動特性 ； 年代際變化 ； 西南氣流 ； 資料空間分布

並列摘要

In this paper, we reviewed the previous studies of Taiwan extreme rainfall variations under climate change. The hourly precipitation data from 21 Central Weather Bureau surface stations during the past 56 years (1960- 2015) and 91 westward typhoons are analyzed. The average precipitation rate was increased on past 30 years, and 10 of the top 15 rainfall cases occurred after 2004. The rain intensity was increasing about 15% for both pre-landing and overland periods; 34% for post-landfall period in last 30 years. The looping tracks extended the pre-landing duration of typhoons and caused more extreme typhoon rainfall. The precipitation during the overland period was related with the typhoon circulation and the terrain effect. This effect is resulted in topographic phase locked convections, which induce the asymmetric potential vorticity (PV) tendency, decelerate the typhoon translation speed, and make more changes for extreme rainfalls. The effects of typhoon duration time and the large-scale strength of southwesterly flow during the post-landfall period both increased the rainfall. It is also highly correlated with the Type-C track cases. The increasing trend of post-landfall typhoon extreme rainfall in past 30 years also reflected the variation of Type-C typhoon quantities. In addition to the increasing trend, both of extreme rainfall frequency and accumulated amount showed large inter-annual variations and a notable decadal variation. On the other hand, the signal of global warming may be superposed by the terrestrial effects and cannot be distinguished. Moreover, the spatial distribution of weather station networks may affect the observations of extreme rainfall. We analyzed the different data sets for 21 stations (56-year continuous record) and 150 stations (20-year continuous record) to explicate the affection of spatial distribution. The underestimate of extreme rainfall amount in the mountain regions for low resolution weather station network is broadly in line with expectations, but the long-term increasing trend and periodic variation features are consistent. As long as the data to maintain uniformity of spatial distribution, the extreme rainfall trend and periodic oscillation characteristics will not be affected. In the other words, the length of weather station record should be the first priority factor for the climate analysis.

並列關鍵字

Typhoon extreme rainfall ； Typhoon movements ； Decadal oscillation ； Southwesterly flow ； Spatial distribution of weather station networks