利用Wavelet分析南海地區季內振盪與東亞夏季季風之研究

The Study of Intraseasonal Oscillation in South China Sea and East Asian Summer Monsoon by using Wavelet Analysis

鄒治華(Chih-Hua Tsou) ； 柯文雄(Wen-Shung Kau) ； 張卜仁(Pu-Jen Chang)

《大氣科學》 28卷1期 (2000/03) Pp. 27-46

摘要

季內振盪是全球大氣運動的一種普遍特徵，這種振盪是大氣本身內部所固有的。隨著大氣低頻振盪及其活動規律的揭露，季內振盪的研究開始被重視，研究大氣低頻振盪的重要目的，就是要將其結果用於長期天氣和短期氣候預報，以提高預報準確率。大氣的季內振盪最先在熱帶被發現，在熱帶地區，大氣季內振盪，具有明顯的地域特徵，同時在不同地方不同時間可有不同的傳播特徵。熱帶大氣季內振盪以亞州季風區最明顯。亞洲夏季季風首先在南海地區建立，南海位於東南亞，是連接東亞與南亞的橋樑，東亞季風與印度季風相互作用於此，因此南海地區由於其特殊地理位置，對東亞季風環流的演發和維持扮演重要的角色。因而深入研究南海地區大氣季內振盪及其活動將有助於東亞夏季季風的肇始及演發過程。目前有許多濾波方法可以用來分析季內振盪的特性，但要兼顧能在頻率及時間域中皆有良好的區域性則Wavelet Transform（小波轉換）爲一有利工具。本研究即採用以區域性特徵爲主的小波轉換進行南海地區資料的濾波分析。我們採用Daubechies（1988）的小波轉換，其基底函數具有正交函數的特性。利用CDC分析之1979～1995共17年，4月1日到8月8日130天之OLR資料進行分析。由CDC之OLR原始值顯示，熱帶地區出現三個強對流，分別在印度西岸、孟加拉灣與南海地區。透過小波轉換可得此三個地區也是五到七月OLR總變異量較大值地區。印度、孟加拉灣之高OLR變異量值，主要是由季節變遷而來。南海地區的OLR變化，30-60天振盪與季節變化同等重要，同時30-60天振盪之貢獻大於10-20天振盪，顯示30-60天振盪在南海地區爲主導。進一步分析南海地區低頻振盪的經向及緯向的傳播過程，我們可歸納出在東亞夏季季風肇始前後OLR低頻振盪的傳播方向有下列幾個特徵：（1）五月初由阿拉伯海東移的30-60天振盪移至孟加拉灣後加強深對流發展，並繼續東傳至五月中旬到達南海，（2）原位於160°E的30-60天振盪於五月初由東向西傳播，當傳至熱帶西太平洋時深對流加強發展，其後持續向西傳播至五月中旬與（1）由孟加拉灣東傳之30-60天振盪交會於南海地區，（3）於110－120°E之經向傳遞顯示出從五月上旬有一從22.5°N向南傳遞、及一從赤道向北傳播的30-60天振盪逐漸向南海地區接近，到了五月中旬交會於南海地區，及（4）10-20天振盪主要顯示爲在南海地區原地發展的特性，於五月中旬南海地區有一低OLR發展出來。這四個特徵的演變有其一致性，即在季風肇始時間同時交會於南海地區。伴隨南海夏季季風的肇始，對流層大氣環流的演變有急速的變化。對流層低層主要爲副高脊撤出南海，西南氣流控制南海，而對流層上層的變化反映了全球尺度的特性，即南海地區高層風場反轉爲較強東風及南亞高壓迅速北移。進一步探討30～60天振盪與東亞夏季季風肇始之關係。結果顯示東亞地區OLR 17年平均值在五月中旬突然下降，此與30～60天振盪及10～20天振盪轉入負相位的第一次交會時間是一致的。同時由NCEP資料17年平均顯示，南亞高壓突然北移至15°N以北地區、南海地區高層風場反轉較强東風及低層副高撤退出南海的時間亦發生在五月第四候，與南海地區OLR之30～60天振盪負值到達時間一致，顯示經小波轉換所處理的30～60天振盪低值僅發生在特定時間，具有區域性特徵，而且亦說明小波轉換得到的30～60天波動，確實可以掌握東亞夏季季風肇始的時間。

關鍵字

季內振盪 ； 夏季季風肇始 ； 小波轉換

並列摘要

Intraseasonal oscillation is a common characteristic of atmospheric motion. This oscillation is part of the atmospheric internal dynamics. Through the observations and studies of the low-frequency oscillation of atmospheric activity, the study of intraseasonal oscillation (ISO) has gradually been emphasized. The purpose of the study of low frequency oscillations is to improve the forecast of long term weather、short term climate and to provide theoretical understanding and numerical weather prediction guidance.
The ISO is first found in tropical areas. The activity of ISO convection is not the same everywhere in the tropics; it has very localized characteristics. At different locations and time periods, the propagation and intensity of ISO can be very different. The most significant area of the ISO in the tropics is in the Asian Monsoon region. The Asian summer monsoon is first established in the South China Sea(SCS).The SCS is located in southeastern Asia, it connects east and south Asia, while at the same time, the East Asian monsoon and south Asian monsoon both are interactive at this area. Therefore, SCS is important not only as a key area to study the onset of the Southeast Monsoon but also the evolution of the Asian summer monsoon. In this paper, we will study the ISO in SCS area in order to understand the onset and evolution of the East Asian Summer Monsoon (EASM).
There are several filter methods can be used to study ISO. However, the ISO in the monsoon area is very localized and takes only place at special instance. Therefore, a localized signal needs to be isolated. To reach this purpose, we use a wavelet analysis developed by Dubechies (1988).
The primary dataset used in the present analysis is satellite-derived outgoing long wave radiation (OLR) by CDC. The OLR data available to us cover 17 years (from 1979 to 1995) in the period of the 1st of April to the 8th of August. The climatologically mean OLR in the Northern Hemisphere summer monsoon region show 3 strong convective areas. These areas are, the west coast of India, Bay of Bengal (BB), and SCS. By using wavelet analysis we can also find the 3 maximum total variances areas, which are, coincide with the above areas. The high variance values of India and BB are contributed by the seasonal variance, however in the SCS, the seasonal variance and 30-60 day variance are equally important(at the same time 30-60 day variance is greater than 10-20 day variance).By more detailed analysis of the propagation of ISO in SCS, we can conclude the following features during the onset of EASM. These are:1)the 30-60 day oscillations propagate eastward at the beginning of May, from Arab Sea. They enhance at BB when they arrive, and then move continually eastward to SCS around Mid-May.(2)The 30-60 day oscillations originated at 160°Emove continually westward since the beginning of May. Deep convection will develop when they arrive to Western Pacific. Afterward, these oscillations will keep on moving westward and join the eastward propagating 30-60 day oscillations from BB at SCS.(3)At 110-120°E,there are 30-60 day oscillations propagate southward from higher latitude and northward from equator, and both are joined together at SCS during Mid-May.(4)The 10-20 day oscillations in SCS become evident and develop locally at Mid-May. All the above four features are joined together at SCS during the onset period of EASM.
The onset of EASM is signaled by an abrupt change in large-scale circulation. The subtropical Pacific anticyclone will suddenly retreat eastward from SCS. While at the same time, the low-level cyclonic circulation and the upper-level anticyclone establish in south Asia. By more detailed studying the relationship between the 30-60 day oscillations and the onset of EASM, we can find that the climatologically averaged unfiltered OLR abruptly decreased in Mid-May corresponding to the climatologically onset date of EASM. A prominent negative signal occurring in Mid-May, can be easily detected from both the 30-60 day and 10-20 day oscillations. From NCEP reanalysis data we can also see that the upper-level anticyclone in South Asia rapidly moves northward to the north of 15°N.A rapid intensification of the lower westerly and upper easterly monsoon circulation also occur at the same period as the 30-60 day oscillations turns negative. This implies that the 30-60 day oscillations play an important role on the onset of EASM. To improve the predictability of the summer monsoon, the propagation and intensity of the ISO may need to be adequately studied and a wavelet transform which provided a local value of amplitude and phase is a suitable tool for this purpose.

並列關鍵字

Intraseasonal oscillation ； Onset of Summer Monsoon ； Wavelet transform