Quasi-Biennial Oscillation Modes Appearing in the Tropical Sea Water Temperature and 700mb Zonal Wind* By Ryuichi Kawamura

December 1988 R. Kawamura 955 Quasi-Biennial Oscillation Modes Appearing in the Tropical Sea Water Temperature and 700mb Zonal Wind* By Ryuichi Kawamura Environmental Research Center University of Tsukuba Tsukuba, Ibaraki 305, Japan (Manuscript received 15 June 1988, in revised form 11 October 1988) Abstract The interrelationships among the 700mb zonal wind in the tropics, the sea surface temperature (SST) in the tropical Indian and Pacific Oceans, and the sea water temperature (SWT) in the tropical western Pacific are discussed on a characteristic interannual time scale. The interannual variability of the main thermocline in low latitudes along 137*E longitude line is reflected primarily on the quasi-biennial oscillation (QBO) time scale. To investigate its physical process, each dominant QBO mode propagating eastward is deduced by applying the complex EOF analysis to monthly mean SST and 700mb tonal wind data over tropical Indian and Pacific Oceans. The QBO mode of the 700mb zonal wind propagates eastward with uniform phase speed. On the other hand, the QBO mode of SST does not propagate with uniform phase speed and a marked phase difference is observed at the area around 150-160*E, where the zonal wind anomalies at 700mb level have the largest amplitudes. Thus, the QBO mode of tropical SST does not always propagate in parallel with that of the tropical troposphere and it is suggested that the tropical tropospheric QBO mode propagating eastward from the Indian Ocean brings about the remarkable phase shift of SST anomalies around 150-160*E. Since the SWT anomalies located in the main thermocline fluctuate in phase with those in the near surface layer (0-50m depth), it is inferred that these variations, which are influenced by the accumulation and release of warm water east of the Philippines, result from the dynamic response of the ocean to the wind stress with the QBO variability. 1. Introduction The importance of the air-sea coupling in the tropical western Pacific, which is connected with triggering E1 Nino/Southern Oscillation (ENSO) phenomenon, has been pointed out recently (e.g., Lau and Chan, 1986; Nitta and Motoki, 1987). It is expected that the large-scale cumulus convective activity is very sensitive to small fluctuations of high sea surface temperature (SST) in this region, although the amplitude of *This paper was presented at the Jacob Bjerknes Symposium on Air-Sea Interactions during the 68th annual meeting of the American Meteorological Society (Los Angeles, January 1988). 1988, Meteorological Society of Japan SST variation is much smaller than those in other regions. The SST over the tropical Pacific has the distinct interannual variability of the ENSO (4-5 years) and the quasi-biennial oscillation (QBO) time scales (see, e.g., Yasunari, 1986; Yoshino and Kawamura, 1987). The QBO time scale is more periodic than the ENSO one. It is also possible that the ENSO is considered as the amplitude modulation of the QBO (Lau and Sheu, 1988). The existence of the QBO in the tropical lower troposphere has been confirmed by Trenberth (1975) and others. Yasunari (1985) showed that the tropospheric QBO can be regarded as an eastward-propagating mode of the global-scale east-west circulation in the tropics. Meehl (1987) noticed the biennial tendency of

956 Journal of the Meteorological Society of Japan Vol. 66, No. 6 the monsoon rainfall in the tropical Pacific and Indian Ocean regions. Thus it is suggested that the QBO, especially in the tropical western Pacific, is the important key for understanding the air-sea coupled system in the tropics on the interannual time scale. However, it is still uncertain how the tropospheric QBO is coupled with the "oceanic" QBO in the tropics. The purpose of this paper is to investigate the relationships between the tropospheric and oceanic QBOs and explore a possible air-sea interaction. First, data sources and analysis procedure are reviewed in section 2. In section 3 and 4, we discuss the QBO-like interannual variation appearing in the SWT and SST associated with the anomalous east-west circulation in low latitudes, utilizing time filtering and complex empirical orthogonal function (CEOF) analyses. Conclusions are presented in section 5. 2. Data and analysis procedure 2.1 Data Table 1 exhibits the specifications of the data used. To investigate the overall features of the interannual variation of the western Pacific SST, monthly averaged data were utilized. We also used the SWT data along 137*E longitude line reflecting a meridional fraction of the vertical thermal structure of the mixed layer in the western Pacific. We need some significant information on the thermal conditions in the mixed layer of the tropical Ocean to discuss the origin of QBO appearing in the tropical SST anomaly field. The SWT data along this longitude line are regularly collected twice yearly in January and July by the marine observation ship (Ryofu-Maru) of Japan Meteorological Agency. However, since this data is not time averaged, we are to be careful in interpreting the analyzed results. Furthermore, we utilized the 700mb zonal wind data as representative of the lower tropospheric circulation in low latitudes to investigate the air-sea interaction, with the QBO time scale. 2.2 Analysis technique The EOF analysis was employed to identify the major modes of the SWT anomaly pattern of the latitude-vertical section along 137*E line. The SWT anomaly is evaluated as departures from the long-term mean of 14 years (1972-1985) in which both January and July SWT data are available. The complex EOF (CEOF) analysis extended to an imaginary domain was also performed with SST and 700mb zonal wind data to clarify the possible propagating features of the air-sea coupled QBO mode. The usual EOF analysis clearly extracts the cross-relationships that phase difference of each station is 0 or * and hence it is quite a useful technique for phenomena like the standing oscillation. However, the above technique cannot easily identify the propagating phenomena with phase shifts in space and time. The CEOF analysis, providing time lag information, enables us to examine the time-space structure of the quasi-periodic propagating mode. A concise description of this analysis technique is given as follows. We should refer to Barnett (1983) if more detailed explanation is needed. Table 1. Periods of record, time means, spatial resolutions, objective regions and sources of the data used in this study.

December 1988 R. Kawamura 957 The frequency component of the QBO mode appearing in the SST and 700mb wind data was extracted using the band-pass time filter, which is presented by Murakami (1979), before we apply the CEOF analysis to their data. By Hilbert transform the filtered data are extended to the complex variables. We solve the eigenvalue problem of the complex covariance matrix (Hermite matrix) and as a consequence can obtain the real eigenvalues and the complex eigenvectors of Bm (J), where m represents the m-th mode and J is the spatial component. Using Bm (J), the amplitude function of SAm(J) and phase function of SPm(J) in space domain are defined by respectively. These two spatial functions can denote where each principal mode enlarges and how it propagates in space. On the other hand, the complex temporal function Fm(t) is given by where X(J, t) represents the complex data extended by the Hilbert transform. The amplitude and phase function in the time domain can be Fig. 1. Vertical section of sea water temperature along 137*E longitude line in January and July during 1972-1985. Contour interval is 1.0*. The standard deviation is also plotted with a thick line. Note that the scale of depth is not uniform.

958 Journal of the Meteorological Society of Japan Vol. 66, No. 6 similarly provided using Fm(t). The above analysis has the advantage of separating plural propagating modes with near frequency domain, although the cross-spectral analysis can be used for various wave phenomena. 3. Dominant modes of SWT anomaly field Figure 1 indicates the vertical structure of January and July SWT (not monthly mean) for 14 years from 1972 through 1985. The mean depth of the main thermocline (i. e., the layer in which vertical gradient of the SWT is steepest) is about 200*300m near the equator, shallower to a depth of 100-150m in the tropics (5-10*N) and deeper again in mid-latitudes. The maximum depth reaches roughly to 400-500m near 30-32*N. The standard deviation of SWT variation has two dominant peaks in space; one is the thermocline in the tropics, the other is that in mid-latitudes. The latter seems to correspond to the variability of the Kuroshio Current axis. In order to investigate the interannual variability of the SWT anomaly field, firstly, we employed the usual EOF analysis to January and July SWT data latitudinally along 137*E longitude line. Figures 2 and 3 reveal the spatial anomaly patterns and their temporal variations for the first two EOF modes. The first and the second EOF modes account for 24.1% and 10.6% of the total variance, respectively. The spatial fig. 2. Spatial patterns of the first two EOF modes deduced by applying the EOF analysis to the SWT date along 137*E longitude line in January and July during the period 1972-1985. Contours are drawn in relative units. Light and heavy shadings denote areas with marked positive and negative anomalies, respectively. Fig. 3. Temporal variations (time coefficients) of two EOF modes indicated in Fig. 2.

December 1988 R. Kawamura 959 structure of the first mode is dominated by a remarkable north-south pattern with opposite phase in tropical and mid-latitude regions. A center of significant anomalies is located around the layer of 100-150m depth at 5-10*N, which is equivalent to the main thermocline in low latitudes. The other anomalies with the same sign as those in the layer of 100-150m depth are also found in the surface layer in the tropics and subtropics, although significant anomalies do not exist at 50m depth in this latitude zone. This suggests that the SWTs in the surface layer fluctuate in phase with those near the main thermocline. As shown in Fig. 3, which is composed of twice yearly (January and July) data, it is noteworthy that this mode has an interannual variation with 2 to 2.5 year period (i. e., QBO time scale). The variability of the main thermocline in low latitudes must be reflected primarily on the QBO time scale. The QBO-like feature of tropical SWT is also shown by Yasunari (1988), who discussed the impacts of :Indian monsoons on the western tropical Pacific SWT, and by Saiki (1987) who investigated the mterannual variation of the subtropical Gyre in the western North Pacific. In the second mode negative anomalies are found in the surface layer in mid-latitudes, while positive anomalies are near the layer of 100-200m depth in the tropics. The interannual variation is not particularly pronounced in this mode. It appears that the second mode mainly explains thermal conditions of the surface layer above 50m depth in middle latitudes. 4. Complex EOF modes of SST and 700mb zonal wind We could obviously confirm the QBO-like interannual variability of the SWT anomaly in the main thermocline even in the western tropical Pacific where the variance of SST is so small. The air-sea coupled QBO in the tropics involving the Indian Ocean sector is further investigated in connection with the origin of the SST anomaly there. The spatial amplitude and phase functions of the first CEOF mode in the SST and the 700mb zonal wind fields are exhibited respectively in Fig. 4. When we first notice the amplitude distribution of SST, there exist regions of somewhat large amplitude not only in the eastern tropical Pacific but also in the tropics from the South China Sea to the Indian Ocean. The phase leads primarily in regions west of India and near Borneo, although the regions from tropical Indian Ocean to South China Sea are roughly in phase, and then propagates from South China Sea, east of the Philippines, across the equator to the eastern Pacific. However, its phase speed is not constant and a pronounced phase gap occurs in the longitude sector 150 to 160*E. The major region of 700mb zonal wind variability, on the other hand, is characterized by a zonally elongated pattern with a maximum amplitude over the western equatorial Pacific near 150 to 160*E longitude line. The 704mb zonal wind anomalies with the QBO component propagate from the tropical Indian Ocean and the southwest maritime continent eastward over the equatorial regions. To demonstrate the features of zonal propagation with seasonal variation, each of the SST and 700mb zonal winds was reconstructed by only the first CEOF mode such that Next, composites were performed with reference to the 700mb zonal wind anomaly values reconstructed in a key region (0-10*N, 150-160*E), where a maximum amplitudee of the 700mb zonal wind exists. One QBO cycle is composed of categories 1 to 8. Category 3 denotes the westerly (positive) maximum phase in the key region, while category 7 implies the easterly (negative) maximum phase. Other categories correspond with the intermediate phases. Composite anomaly maps of the 700mb zonal wind for categories 1 to 4 are shown in Fig. 5. Composite anomaly maps for categories 5 to 8 are omitted because they are equivalent to the composite maps for categories 1 to 4 but for an opposite sign. Composite SST anomaly maps are also made with reference to the categories of the zonal wind at 700mb and similarly indicated in Fig. 6.

960 Journal of the Meteorological Society of Japan Vol. 66, No. 6 Fig. 4. Spatial amplitude (relative units) and phase functions (degrees) of the first complex EOF mode of the QBO filtered SST and 700mb zonal wind anomalies. At the category 1 of 700mb zonal wind, there exist positive (westerly) anomalies from the tropical Indian Ocean to the maritime continent, while negative (easterly) anomalies are observed over the tropical eastern Pacific. The regions of westerly anomaly propagate eastward and cover the maritime continent and the western Pacific at category 2. At the category 2 of SST, however, there remain weak negative anomalies in the tropical eastern Pacific though the westerly anomalies appear over the western Pacific. At category 3, nearly corresponding with June to August, westerly anomalies develop, reach their maximum around 150*160*E in the tropics and extend further eastward over the tropical Pacific Ocean. On the other hand, the positive SST anomalies appear drastically in the tropical eastern Pacific and, in contrast, negative anomalies are found in the western Pacific. After category 3, the regions of large westerly anomaly move toward the eastern Pacific and easterly anomalies are gradually established over the tropical Indian Ocean and the maritime continent, while positive SST anomalies are intensified in the eastern Pacific. Figure 7 displays the time-longitude cross

December 1988 R. Kawamura 961 Fig. 5. Each category of QBO filtered 700mb zonal wind anomalies reconstructed only by the first CEOF mode. Contour interval is 0.2m/s. Negative values are shaded and imply easterly anomalies. section of QBO filtered zonal wind anomalies at 700mb along the equatorial zone (0*-10*N) reconstructed only by the first CEOF mode. Although a discontinuity of phase appears at 80*E, it can be found that zonal wind anomalies east of this longitude propagate eastward with nearly uniform phase speed. The large amplitude modulations of these anomalies can be seen conspicuously over the equatorial Pacific east of Philippines. The maximum anomalies appear during two periods from 1973 through 1976 and from 1980 through 1983. This QBO mode is nearly phase-locked to the seasonal cycle and shows the maximum amplitudes over the regions from 140 to 160*E, where the variance of the anomaly is largest among areas along the latitude belt, in the warm seasons from May to October. It seems that the first CEOF mode deduced by the present study accounts for the more fundamental QBO mode propagating eastwards with an unchanging phase speed than that at 700mb level presented by Yasunari (1985). The time-longitude cross section of QBO filtered SST anomalies along 2 to 10*N is also indicated in Fig. 8 similar to Fig. 7. The periods of large amplitude almost coincide with those of the 700mb zonal wind anomaly field. However, the phase speed of SST anomalies is not uniform and the phase differences are observed most significantly in the regions from 120 to 130*E

962 Journal of the Meteorological Society of Japan Vol. 66, No. 6 Fig. 6. Each category of QBO filtered SST anomalies reconstructed similar to Fig. 5. Contour interval is 0.1* and negative values are shaded. (corresponding to the Philippines) and from 150 to 160*E. Compared to Fig. 7, it is found that the area around 150-160*E, where the marked phase shift of SST anomalies exists, agrees well with the large amplitude area of the 700mb zonal wind anomaly field. The westerly (easterly) anomalies at longitudes around 150-160*E east of the Philippines predominate from May to October and subsequently the SST anomalies centered at 140*E have negative (positive) values. These SST anomalies are the prevailing tendency for reaching the maximum amplitudes about three months later. The above results suggest that a dominant eastward-propagating QBO mode, which may be a type of the normal mode in the tropical troposphere, does not always propagate in parallel with the QBO mode of SST. Although there naturally exist interactions between the tropospheric and oceanic QBOs, it can be considered that the pronounced phase shift of SST anomalies near 150-160*E primarily results from the forcing due to the anomalous wind stress with the QBO variability. Because the QBO mode of the 700mb zonal wind propagates eastward from the tropical Indian Ocean as in Fig. 7 and hence it is difficult to think that the zonal wind field at 700mb level over the Indian Ocean is directly influenced by the thermal forcing of the ocean accompanied with the marked phase shift

December 1988 R. Kawamura 963 Fig. 7. Time-longitude cross sectin of QBO filtered 700mb zonal wind anomalies (0*-10*N) reconstructed only by the first CEOF mode. Contour interval is 0.5m/s and positive values imply westerly anomalies. Fig. 8. Time-longitude cross section of QBO filtered SST anomalies (2-10*N) reconstructed similar to Fig. 7. Contour interval is 0.1*. Light and heavy shadings denote areas with easterly and westerly anomalies at the 700mb level more than 1.0m/s. of SST anomalies around 150.160*E. Since the SST anomalies around 137*E longitude line seen in Fig. 8 correspond to the time coefficients of the first EOF mode of SWT anomaly, it is understood that two QBO modes of the SST and SWT are an identical phenomenon. It is expected, thus, that the SWT anomaly in the main thermocline in low latitudes, which has a QBOlike features results from the dynamic response of the ocean to the wind stress with the QBO mode. This idea is also supported by the results of Yasunari (1988). Nitta and Motoki (1987) emphasized the westerly bursts accompanying equatorial twin cyclones as a trigger of El Nino. The eastward-propagating QBO mode in the lower troposphere may play an important role in producing favorable conditions for the occurrence of the westerly burst over the western Pacific. If so, it is deduced that the tropospheric QBO mode presented in this paper affects the ENSO onset. The pronounced SST anomalies over the South China Sea in the cold season, on the other hand, are not presumably formed by the above process. It is found that these anomalies are sensitive to the major cold surges (Hanawa et al., 1988; Kawamura, 1988). Since the cold surges play a role in intensifying convective, activities in low latitudes if we notice their high-frequency variations (e.g., Chang and Lum, 1985; Wang and Murakami, 1987), it is necessary that we also investigate their effect associated with the enhanced air-sea coupling of the QBO mode. This line will be further studied. 5. Concluding remarks The SST and SWT in the western tropical Pacific have a QBO-like interannual variability coupled with the tropical lower troposphere. We discussed the relationships between the tropo-

964 Journal of the Meteorological Society of Japan Vol. 66, No. 6 spheric and oceanic QBOs. We employed the usual EOF analysis to January and July SWT data latitudinally along 137*E longitude line and obtained the dominant modes of the SWT anomaly field there. The first mode has the interannual variability with the QBO time scale. It is found that the tropical SWTs near the main thermocline of this mode fluctuate in phase with those in the surface layer above a depth of 50m. By applying the complex EOF analysis to the SST and 700mb zonal wind anomaly fields over tropical Indian and Pacific Oceans, each dominant QBO mode propagating eastward is extracted. The QBO mode of SST does not propagate with uniform phase speed and the phase differences are observed in the regions around the Philippines and around 150-160*E along the equatorial zone. The area near 150-160*E coincides with the large amplitude area of the QBO mode of 700mb zonal wind, which propagates eastward with uniform phase speed. Thus it is inferred that marked phase shift of SST anomalies around 150-160*E is brought about by the tropical tropospheric QBO mode propagating eastward from the tropical Indian Ocean. It is suggested that the SWT anomalies in the main thermocline in the tropical western Pacific result from the dynamic response of the ocean to the wind stress with the QBO variability because two QBO modes of the SST and SWT are basically an identical phenomenon. Acknowledgements The author is grateful to Prof. M. Yoshino and Dr. T. Yasunari for their valuable suggestions and encouragement throughout this study. He also expresses his thanks to two anonymous reviewers for invaluable continents. Thanks are extended to the staff of the Oceanographical Division of the Japan Meteorological Agency for making available SWT data used in this study. This research was financially supported by the research project of University of Tsukuba for 1988. Computations were performed with the FACOM M-780/20 at the Science Information and Processing Center of University of Tsukuba. References Barnett, T.P., 1983: Interaction of the monsoon and Pacific trade wind system at interannual time scales, Part I, The equatorial zone. Mon. Wea. Rev., 111, 756-773. Chang, C.-P. and K.G. Lum, 1985: Tropical-midlatitude interactions over Asia and the western Pacific Ocean during the 1983/84 northern winter. Mon. Wea. Rev., 113, 1345-1358. Hanawa, K., T. Watanabe, N. Iwasaka, T. Suga and Y. Toba, 1988: Surface thermal condition in the western North Pacific during the ENSO events. J. Meteor. Soc. Japan, 66, 444-455. Kawamura, R., 1988: The interaction between winter monsoon activities in East Asia and sea surface temperature variations over the western Pacific Ocean (in Japanese with English abstract). Geographical Review of Japan, 61 (Ser. A), 469-484. Lau, K.-M. and P.H. Chan, 1986: The 40-50 day oscillation and the El Nino/Southern Oscillation: A new perspective. Bull. Amer. Meteor. Soc., 67, 533-534. - and P.J. Sheu, 1988: Annual cycle, quasibiennial oscillation and southern oscillation in global precipitation. J. Geophys. Res. (to be published) Meehl, GA., 1987: The annual cycle and interannual variability in the tropical Pacific and Indian Ocean regions. Mon. Wea. Rev., 115, 27-50. Murakami, M., 1979: Large-scale aspects of deep convective activity over the GATE area. Mon. Wea. Rev., 107, 994-1013. Nitta, T. and T. Motoki, 1987: Abrupt enhancement of convective activity and low-level westerly burst during the onset phase of the 1986-87 El Nino. J. Meteor. Soc. Japan, 65, 497-506. Saiki, M., 1987: Interannual variation of the subtropical gyre in the western North Pacific (in Japanese with English Abstract). Umi to Sora, 63,1-13. Trenberth, K.E., 1975: A quasi-biennial standing wave in the southern hemisphere and interrelations with sea surface temperature. Quart. J. Roy. Met. Soc., 101, 55-74. Wang, X.-L. and T. Murakami, 1987: Intraseasonal meridional surges and equatorial convections during the southern hemisphere summer. J. Meteor. Soc. Japan, 65, 727-736. Yasunari, T., 1985: Zonally propagating modes of the global east-west circulation associated with southern oscillation. J. Meteor. Soc. Japan, 63,1013-1029. 1986: A possible link of QBOs -, among the stratosphere, the troposphere and sea surface temperature in the tropics. In workshop on 40 day mode & ENSO, Meteor. Res. Rep., 80-85. 1988: Impact of Indian monsoon -, on the coupled atmosphere/ocean system in the tropical Pacific. J. Climate (to be published).

December 1988 R. Kawamura 965 Yoshino, M. and R. Kawamura, 1987: Periodicity and propagation of sea surface temperature fluctuations in the equatorial Pacific. Beitr. Phys. Atmosph., 60, 283-293.