Subsurface temperature anomalies in the North Pacific Ocean associated with the ENSO cycle*

Transcription

1 Chinese Journal of Oceanology and Limnology Vol. 28 No. 6, P , 2010 DOI: /s Subsurface temperature anomalies in the North Pacific Ocean associated with the ENSO cycle* CHEN Yongli ( 陈永利 ) 1, 2, **, ZHAO Yongping ( 赵永平 ) 1, 2, WANG Fan ( 王凡 ) 1, 2, HAO Jiajia ( 郝佳佳 ) 3, FENG Junqiao ( 冯俊乔 ) 1, 2 1 Institute of Oceanology, Chinese Academy of Sciences, Qingdao , China 2 Key Laboratory of Ocean Circulation and Wave Studies, Chinese Academy of Sciences, Qingdao , China 3 Yantai Institute of Costal Zone Research, Chinese Academy of Sciences, Yantai , China Received Nov. 18, 2009; revision accepted Jul. 3, 2010 Chinese Society for Oceanology and Limnology, Science Press, and Springer-Verlag Berlin Heidelberg 2010 Abstract Multi-year Simple Ocean Data Assimilation (SODA) and National Centers for Environmental Prediction (NCEP) datasets were used to investigate the leading patterns of subsurface ocean temperature anomalies (SOTA) and the corresponding atmospheric circulation structure in the Pacific Ocean (20 S 60 N). In this paper, the evolution of North Pacific SOTA associated with El Niño-southern oscillation (ENSO), and their relationship with the overlying zonal/meridional atmospheric circulations were elucidated. The results indicate that: (1) there are two dominant modes for the interannual variability of the North Pacific SOTA. The primary mode is the dipole pattern of the central and western North Pacific SOTA associated with the leading mode of ENSO, and the second mode is the zonal pattern related to the second mode of ENSO. These two modes consist of the temporal-spatial variation of the SOTA in the North Pacific. (2) During the developing phase of the El Niño event, positive (negative) SOTA appears in the western (central) portion of the North Pacific Ocean. During the mature and decaying phase of the El Niño event, the western positive center and the central negative center continue to be maintained and enhanced. Meanwhile, the position of the western positive center slightly changes, and the central negative center moves eastward slowly. After the El Niño event vanishes, the positive SOTA disappears, and the central negative SOTA becomes weak and remains in the northeastern Pacific Ocean. The results for La Niña are generally the opposite. (3) During the El Niño/La Niña cycle, formation and evolution of the SOTA, with opposite signs in central and western North Pacific Ocean, resulted from vertical movement of the two northern branches of the Hadley Cell with opposite direction, as well as the positive feedback of the air-sea interaction induced by dynamic processes in the mid-latitudes. The former gives rise to the initial SOTA, and the latter intensifies SOTA. Under the forcing of these two processes, SOTA evolution is formed and sustained during the El Niño/La Niña events. Also discussed herein as background for the ENSO cycle are the possible connections among the West Pacific subtropical high, the strength of the Kuroshio near the East China Sea, the Kuroshio meanders south of Japan, the Aleutian Low, and cold advection in the central North Pacific Ocean. Keyword: ENSO cycle; North Pacific; subsurface ocean temperature anomalies; Hadley Cell 1 INTRODUCTION North Pacific Ocean (NPO) sea temperature anomalies have a significant influence on the atmospheric circulation system in the northern hemisphere. The climatic changes in the NPO and surrounding areas have been the focus of many studies regarding NPO formation and its influencing processes (Frankignoul, 1985; Miller et al., 1997; Liu et al., 2006; Gong et al., 2006). Previous observations show that, during the ENSO cycle period, a significantly large sea temperature anomaly area appeared consistently in the NPO. It is believed that NPO sea temperature anomalies are the result of meridional heat transport of low-latitude tropical ocean energy anomalies, which are caused by atmospheric Supported by the Knowledge Innovation Program of Chinese Academy of Sciences (No. KZCX2-YW-Q11-02) and the National Basic Research Program of China (973 Program) (Nos. 2007CB and 2006CB403601) Corresponding author: ylchen@qdio.ac.cn

2 No.6 CHEN et al.: Subsurface variability during ENSO cycle 1305 circulation anomalies in the mid and high latitudes and ENSO events, through the Atmospheric Bridge, or the North Pacific western boundary current. Previous studies have mainly focused on the teleconnection of the ENSO with the NPO sea surface temperature anomalies (SSTA). These studies have identified that the SSTA is mainly affected by sea surface heat flux. To identify the generating mechanisms of the NPO sea temperature anomalies, the present study investigates the characteristics and formative processes of the NPO subsurface ocean temperature anomalies (SOTA) during the ENSO cycle. This work also explores the physical mechanism of the tropical ocean influence on NPO sea temperature anomalies. 2 DATA AND METHOD The Oceanic and atmospheric data utilized in this study includes the Simple Ocean Data Assimilation (SODA ) (Carton et al., 2000a; 2000b) and the National Centers for Environmental Prediction (NCEP) reanalysis datasets (Kalnay et al., 1996). The analysis region and time period extend from 20 S to 60 N, and from January 1958 to December 2007, respectively. Based on the observations that the maximum ocean temperature anomalies generally appear near the thermocline, scientists have used temperature anomalies on the thermocline curved surface to describe SOTA. There are many methods to estimate thermocline depth anomaly. Usually, the depth of the 20 С or 14 С isotherms in the tropical ocean, and the temperature anomalies at 400 m or 200 m in the mid and high latitudes oceans are utilized to represent the thermocline depth. Chao et al. (2003) used a curved surface of depth in which the subsurface temperature anomalies reached their maxima to replace the thermocline surface. In addition, some researchers have used the heat content anomalies in the ocean upper layer, or sea level height anomalies, as thermocline depth variations. On the basis of the above results, and using the 50-year data of ocean temperature anomalies within the upper 600 m in the Pacific Ocean, we analyzed the thermocline depth distribution. The distribution of the climatological thermocline surface in the Pacific Ocean is shown in Fig.1. The thermocline is shaped like a trapezoidal mesa in the tropical Pacific (20 S 20 N), which becomes shallower and wider from west to east. The depth in the warm pool can reach 140 m in the West Pacific, but it is smaller than 60 m in the East Pacific. On the north and south sides of the mesa, outside the equatorial region, there is a deep trough on each side with maximum depths greater than 200 m. In NPO, an east-west trough with depths >200 m appears on the east side of Japan. In addition, the thermocline on the south and east side of the trough is in the direction of southwest to northeast, with a gradually increasing width and decreasing depth (<60 m). Compared with the circulation system in the Pacific Ocean, the deep troughs in the tropical regions on both sides of the equator and east of Japan are closely related to the powerful equatorial currents in the north and south Pacific and the Kuroshio south of Japan, respectively. To understand the formation mechanism and evolution of the SOTA in the Pacific during the ENSO cycle, we first determined the climatological thermocline surface in the Pacific using the SODA data, and then calculated the temperature anomaly series at the thermocline surface as a SOTA index. Since the ENSO cycle has inter-annual variability over a 2 7-year period, we conduct the 1 8-year band pass filtering to the SOTA index to separate inter-annual variability, followed by EOF analysis. The long-term trends were eliminated with the filtering process. A similar procedure was performed for the NCEP data. Fig.1 Distribution of the annually averaged thermocline depth (m) in the Pacific (contour interval is 40 m)

3 1306 CHIN. J. OCEANOL. LIMNOL., 28(6), 2010 Vol.28 3 MAIN MODES OF INTER-ANNUAL VARIABILITY OF THE SOTA IN THE NPO Fig.2 shows the first (a) and second (b) modes, and their time series (c) obtained by the EOF analysis (un-standardized) of the inter-annual variability of the SOTA in the Pacific. They accounted for 27.2% and 14.7% of the total variance, respectively. The main features of the first mode in space were: in the tropical area between 20 S and 20 N, there is a dipole pattern of SOTA in the zonal direction, with a longitude of 160 W as its axis. One of the extreme SOTA was located near the warm pool region and 150 E north of the equator in the western tropical Pacific. The other extreme SOTA was located between 90 and 120 W south of the equator in the eastern tropical Pacific with slightly larger amplitude than the western center. In the areas north of 20 N, the SOTA anomalies showed a weak dipole pattern oriented southwest-northeast, which was opposite to that in the tropical areas. The main feature of spatial distribution of the second mode was that there was a seesaw shaped SOTA distribution in the tropical Pacific, with 6 8 N as its zonal axis, on average, north and south of which were SOTA regions with large values of opposite signs. These regions were located in the central and eastern North Pacific around N (12 N on average), and the central equatorial Pacific, respectively. In the areas north of 20 N, the SOTA only showed zonal pattern connected to the tropical Pacific, and formed a negative-positive-negative or positive-negativepositive SOTA zonal distribution from south to north. Spectral analysis (Fig.2d) shows that both the first and second modes possessed two significant periods-56 and 44 months. There was an optimum simultaneous correlation between the time series of the first mode and Niño 3 (the SST index in the region 5 S 5 N, W) with r=0.83; and an optimum delayed correlation between the time series of the second mode and Niño 3 with r=0.59 and the optimum with a delayed time being 6 months. Except for the two weak El Niño events in 1993 and , the above mentioned modes manifested in all the ENSO events in the past 50 years, such as El Niño events in 1963/1964, 1965/1966, 1968/1969, 1972/1973, 1976/1977, 1982/1983, 1986/1987, 1991/1992, 1997/1998, 2002/03 and the La Niña events in 1964/1965, 1967/1968, 1970/1971, 1973/1974, 1975/1976, 1984/1985, 1988/1989, 1995/1996 and 1998/2000. The contributions of the third or higher SOTA modes accounted for less than 3.7% of the total variance, and there was no reliable correlation between their time series and Niño 3. These imply that modes 1 and 2 were major modes for the SOTA in the Pacific, and the combination of the two modes produced tropical ocean El Niño/La Niña events (Zhao et al., 2007), which lead to synchronal evolution of the SOTA in the NPO. Fig.2 Time-space distribution of the first (a) and second (b) modes of the inter-annual variability of SOTA in the Pacific (solid lines show positive values, dashed lines show negative values, contour interval is 0.005), corresponding time series (c) and spectral analysis results (d) (thick solid line for the first mode; dashed line for the second mode, thin solid line shows Niño 3)

4 No.6 CHEN et al.: Subsurface variability during ENSO cycle EVOLUTION OF THE SOTA IN THE NPO DURING ENSO CYCLE We combined the first and second modes to obtain the reconstructed fields of inter-annual variability of the SOTA in the NPO. During the strongest El Niño events in 1997/98, the SOTA in the tropical ocean reached 5.0 C, and it was also up to 2.0 C in the NPO. Since 1958, during the seven strong El Niño events (1965/66, 1972/73, 1982/83, 1986/87, 1991/92, 1997/98 and 2002/03) and six strong La Niña events (1964/65, 1970/71, 1973/74, 1974/75, 1988/89, 1998/99), the larger values of the SOTA all appeared in winter, except the 1986/87 El Niño event. Fig.3 shows the evolution process of the SOTA in the Pacific Ocean over 6 months before and after the peak of ENSO event, which was composed of the six El Niño events (except 1986/87) and six La Niña events. This trend shows that, during the formation of the El Niño, the strong positive center of SOTA in the tropical western Pacific moved eastward to the equatorial eastern Pacific Ocean along the equator with the strengthened value, and then expanded northward along the coast of North America, leading to the El Niño events. In the initial stage of El Niño events, a weaker positive SOTA signal remained northeast of the NPO. However, there was no clear SOTA in the central NPO. With the development of El Niño events, a significant positive SOTA center appeared in south of Japan, in the western NPO (20 30 N), and a large negative SOTA center appeared in the central NPO. Following these events, these two centers gradually strengthened with the slowly eastward movement toward the latter one. The weak positive SOTA which originally existed in the northeast moved to the north coast of North America, and associated with the positive SOTA originated from the tropical eastern Pacific Ocean along the north coast of North America. With gradually weakening El Niño events, the positive SOTA in the tropical eastern Pacific moved westward along N, and the positive SOTA center in the western NPO and the central negative SOTA strengthened the latter maintained during its eastward propagation. After the demise of El Niño events, the western SOTA center disappeared, while the central SOTA center weakened and remnants appeared in the Gulf of Alaska. During La Niña events, the result was generally the opposite, but its strength was weaker than that of El Niño events. If the SOTA variation during El Niño (La Niña) was maintained as the initial field of that during La Niña (El Niño), a complete ENSO cycle would have been formed. Compared with the synchronous SSTA pattern (data not shown), their abnormal distribution and evolution were very consistent. The main difference was that the SSTA in the tropical eastern Pacific was distributed further west compared to the SOTA pattern, with the western boundary shifted to 160 E. During the development of El Niño, the SSTA centers in the western NPO were located further east compared to the SOTA, the SSTA centers located at the central NPO were further northwest than the SOTA, and these two SSTA centers appeared several months earlier than those of SOTA. The comparisons show that SSTA and SOTA may be subject to the same dynamic process of the atmosphere and ocean coupled system, while the SSTA also was affected by the important impact of the sea surface heat flux. There were many factors resulting from the SOTA in the NPO, such as local air-sea feedback and marine internal dynamic process. However, it was unclear what percentage of the ENSO event accounted for the SOTA in the NPO. We defined the ratio between the variance explained by the first 2 EOF modes of the SOTA and the total variance (1 8-year band filtering was applied to the original data) as the contribution rate of ENSO (Fig.4). We identified that in the tropical oceans, SOTA mainly was caused by ENSO processes, and its contribution was more than 40% in most areas, especially in the tropical central and western Pacific. In the NPO, the area with a contribution > 40% was not as large as in the tropics, and mainly lay south of Japan in the western NPO, as well as in the band shaped area from central NPO to the Aleutian and coastal areas of North America, with a contribution > 50% in central areas. However, contributions of the local air-sea feedback and the marine internal dynamic process were dominant for most areas in the North Pacific. Based on observational data, White (1995) defined the mixed layer depth as the depth where the temperature is 1 C lower than the SST. He also found that the mixed layer depth anomaly in the NPO during the mature ENSO had the same distribution as the SOTA distribution during the ENSO peak periods (Fig.3), except for a slightly larger area for the northern coastal area of North America. This also confirms that the analysis method in this paper is reliable from another perspective. 5 ZONAL AND MERIDIONAL CIRCULATION FEATURES OF THE NORTH PACIFIC ATMOSPHERIC ANOMALY LINKED TO THE ENSO CYCLE Previous studies (e.g. Alexander et al., 2002) have

5 1308 CHIN. J. OCEANOL. LIMNOL., 28(6), 2010 Vol.28 Fig.3 Evolution of the SOTA in the Pacific for 6 months before and after the El Niño (left) and La Niña (right) event peaks

6 No.6 CHEN et al.: Subsurface variability during ENSO cycle 1309 Fig.4 Percentage distribution of the contribution rate of the ENSO to the total variance of the SOTA interannual variability Fig.5 Spatial pattern of the first (left) and second (right) EOF modes of the 500 hpa vertical velocity anomaly fields (hpa/s, positive value indicates the updraft) shown that, during El Niño/La Niña events, the thermal effect of the tropical Pacific Ocean firstly led to Walker circulation anomalies over the equator, and then induced subtropical Hadley circulation anomalies over the NPO, thereby transmitting the energy of the tropical sea temperature anomalies to the mid and high latitudes. This resulted in high-latitude atmospheric circulation change, and further influenced the state of the ocean underneath. However, it is not clear how to implement the impact on the NPO through the Atmospheric Bridge. Fig.5 shows the spatial distribution of the first and second EOF modes of the 500 hpa vertical velocity anomaly fields (positive and negative values indicate updraft and downdraft, respectively), which contributed 19.8% and 6.8% to the total variance, respectively. The spatial distributions of the corresponding eigenvector of SOTA were basically similar, and their time series showed a good relationship. Correlation coefficients were and 0.62 for the first and second modes, respectively. For the first vector field, during El Niño events, there were updrafts and downdrafts to the east and west of 160 E over the tropical Pacific Ocean, respectively. Particularly, over the NPO, it was positive-negative-positive from west to east, and at the same time there were downdrafts in the central NPO and updrafts on both sides. The corresponding subsurface ocean temperature increased in the eastern tropical Pacific, decreased in the central and western tropical regions, and in the NPO it decreased in central areas and increased on both sides. Furthermore, the situation was opposite during La Niña events. There were similar distribution characteristics of the second mode in the NPO, indicating that the abnormal vertical motion of atmospheric circulation caused by an ENSO event was likely to have a major impact on the SOTA in NPO. Wang (2002) analyzed atmospheric circulation features associated with the ENSO, and pointed out that there were significant anomalies of the Walker, Hadley circulation and mid-latitude zonal circulation (MZC) over the Pacific Ocean during the ENSO mature stage. Based on his results, the composite analysis of the abnormal vertical flow over the equator, which is closely related to the six El Niño events, showed zonally-averaged abnormal vertical flow from E and 160 E 140 W, and meridionally-averaged abnormal vertical flow along

7 1310 CHIN. J. OCEANOL. LIMNOL., 28(6), 2010 Vol N. These events characterize abnormal Walker circulation, abnormal Hadley circulation and abnormal MZC (Fig.6), respectively. This shows that during El Niño events, the abnormal Walker circulation had formed in the 8 months before the El Niño peak, and had been markedly enhanced in the initial 6 months. Furthermore, it reached its strongest 2 months before the peak. Then, it significantly decreased 4 months after the peak. During this period, the updraft appeared east of 160 E with the extreme near the 160 W and 500 hpa. The downdraft appeared west of 160 E with the extreme near the 120 E and 350 hpa. The abnormal Hadley circulation in the western North Pacific took shape during the 6 months before the El Niño peak, and had been markedly enhanced in the initial 4 months. It also reached its strongest at the El Niño peak, and then significantly decreased during the 6 months after the peak. Its main feature was that the downdraft appeared over the tropical oceans with its extreme near the 10 E and 300 hpa. The updraft appeared at N with its extreme near the 30 N and 700 hpa. In the NPO, abnormal Hadley circulations in the central-eastern and western areas were synchronously changed, and the pattern showed that the updraft appeared over the equatorial oceans with its extreme at the 300 hpa. The downdraft appeared at 8 35 N with the extreme near N and 500 hpa. In addition, Fig. 6 also shows that the abnormal MZC truly exists over mid-latitude waters, simultaneously connected to the North Pacific Hadley circulation anomalies, and reached the strongest within 2 months before and after the El Niño mature stage, but it was still evident 6 months after the El Niño mature stage. The distribution showed that the downdraft appeared in the central NPO with its extreme near 180 E and 400 hpa. The updraft appeared in the western NPO with its extreme near 125 E and 500 hpa, and the updraft basically appeared in the eastern NPO with its extreme at 300 hpa over the North American coast. In addition, a weak updraft also appeared in the Gulf of Alaska. It was clear that abnormal Walker circulation over the equatorial region was firstly strengthened, and then Hadley circulation to the east and west of the NPO and the MZC in the mid-latitudes were strengthened during the following two months. The abnormal Walker circulation reached Fig.6 Structures composited by zonal-vertical and meridional-vertical cross sections of vertical velocity anomalies (10-3 hpa/s) during El Niño. a, Walker circulation; b, western NPO Hadley circulation; c, central and eastern NPO Hadley circulation; d, NPO mid-latitude MZC

8 No.6 CHEN et al.: Subsurface variability during ENSO cycle 1311 the strongest at 2 months before El Niño peaked. However, the other circulation loops reached their strongest within 2 months before and after El Niño peaked. In addition, the abnormal MZC was actually the north vertical air branch of the abnormal Hadley circulations in the eastern and western NPO, resulting from ENSO forcing. Compared with the temporal and spatial distributions of the SOTA during El Niño, it is clear that the equatorial abnormal Walker circulation and the tropical SOTA respond to one another. The abnormal updraft over the western NPO was consistent with the positive SOTA center in the waters south of Japan, and the abnormal downdraft over the central NPO was basically consistent with the negative SOTA center there. Meanwhile, we also mapped the structures composited by the zonal and meridional vertical velocity anomalies during La Niña events (data not shown), which showed similar process and distribution during El Niño events, but with the opposite sign and weaker intensity. That is, the two opposite phases of the abnormal Hadley circulations caused by the SOTA and ENSO in the NPO were closely related. 6 PHYSICAL PROCESSES OF FORMATION AND EVOLUTION OF THE SOTA IN THE NPO ASSOCIATED WITH THE ENSO Based on data analysis and numerical simulation of a 50-year dataset, Alexander et al. (2002) reviewed the previous work on the impact and physical processes of the ENSO events on the SSTA in NPO, and revealed the basic distribution pattern of the SSTA in NPO influenced by the ENSO events. Alexander et al. (2002) pointed out that the tropical ocean transferred its energy to the mid and high latitudes by the Atmospheric Bridge, and then changed the distributions of the wind, temperature, humidity and cloud there. Thus, it influenced the SSTA through the sea surface heat flux anomaly, abnormal Ekman pumping and ocean advection transport to influence the SST, sea surface salinity and mixed layer depth. However, its specific physical process is still unclear. According to the results of the previous section, the Atmospheric Bridge mentioned here is the Hadley circulation. Fig.7 shows the evolution of the SOTA in the Pacific, horizontal wind field and the corresponding divergence at 850 hpa and 200 hpa during the 1982/83 El Niño events. What needs to be clarified now is that when the energy of the tropical waters has been transferred to the mid-latitudes, how it is transferred down to the ocean? The SOTA is more subject to ocean dynamic processes. Thus, it is important to elucidate its dynamic process. The 1982/83 El Niño event formed from May In the tropical Pacific, it showed a significant positive SOTA which propagated to the equatorial eastern Pacific Ocean in the west-east direction on the thermocline surface along the equator. In the NPO, it showed that the large-scale positive SOTA with zonal distribution appeared in high latitudes north of 35 N, and the larger value was located in the Gulf of Alaska. However, the weak negative SOTA appeared on its southern side in the east-west direction. This is the distribution of the SOTA in the NPO during La Niña state before El Niño events occur. At the corresponding period, the abnormal anti-cyclonic circulation appeared in the lower troposphere (850 hpa) over the NPO. In August 1982, the Walker circulation was significantly strengthened, and two abnormal Hadley circulations in the eastern and western NPO were basically established in October. At the same time, abnormal patterns at 850 hpa in the NPO had been ultimately changed, and anti-cyclonic circulation anomalies appeared on its west side, with a strong divergence center and with prevailing NW airflow. However, at its east side, it had transformed to the abnormal cyclonic circulation with a strong convergence center off the coast of the North America. Meanwhile, an abnormal convergence center appeared over the sea south of Japan, in the western NPO. This center was covered by a strong southwest airflow. During the El Niño peak in December, abnormal cyclonic circulation occupied high latitudes in the NPO. The circulation also had significantly strengthened Aleutian low pressure, and the significantly abnormal divergent centers. The associated abnormal anti-cyclonic circulation appeared in the NPO south of 30 N and the waters south of Japan in the western NPO south of 20 N, respectively. During this period, the abnormal convergence and divergence centers appeared at the upper troposphere (200 hpa) over the central NPO and the sea south of Japan in the western NPO. These centers, together with the troposphere divergence center in the tropical Pacific, formed two strong abnormal Hadley circulation cells in the NPO. Clearly, the 200 hpa divergence centers were located more northerly than that at 850 hpa. At the same time, the SOTA peak in the tropical Pacific reached the east coast, and then moved northward along the coast of North America as Kelvin waves. A wide range of negative SOTA appeared in the central NPO. The

9 1312 CHIN. J. OCEANOL. LIMNOL., 28(6), 2010 Vol.28 Fig.7 Evolution of the horizontal wind field (m/s) and corresponding divergence field (the shading bar is s -1 ) at 850 hpa (a) (left column), and 200 hpa (b) (mid column), and the SOTA (c) (right column) in the Pacific during the 1982/83 El Niño event

10 No.6 CHEN et al.: Subsurface variability during ENSO cycle 1313 positive SOTA, originating from high latitude, weakened and moved eastward to link with the positive SOTA coming from the eastern equator. Subsequently, El Niño events decayed from mature to vanishing. However, the positive SOTA center over the sea south of Japan in western NPO showed sustained strengthening with a increasing range (+0.8 C in June 1983). The SOTA center in central NPO was strengthened and slowly shifted eastward, and moved to the Gulf of Alaska with an intensity of up to -1.4 C in August After the demise of El Niño, the SOTA center in western NPO rapidly disappeared, and the one in central NPO was reduced significantly. At the same time, the 850 and 200 hpa wind fields and the corresponding divergence fields kept the distribution pattern during the El Niño peak. However, the intensity was gradually weakened until August During El Niño events, the sea surface wind stress field and the corresponding divergence field (Fig.omitted) also showed similar evolutionary characteristics. The most obvious feature was that of the Aleutian low pressure, the NW airflow in the central NPO, and the SW airflow in the waters south of Japan, which was strengthened. However, the Western Pacific subtropical high was weakened. Alexander et al. (2002) indicated similar results with the SLP composite analysis from the 9 El Niño and La Niña events during His results showed that the significant SOTA in the central and western NPO were formed during the developing stage of El Niño events, and were strengthened and maintained during El Niño events, during mature and decay periods, and disappearing after the vanishing of El Niño. The above analysis shows that, during El Niño strengthening periods in the central NPO, the troposphere was controlled by the abnormal downdraft of the Hadley circulation with a high-level convergence and low-level divergence, corresponding to the abnormal updraft branch of the tropical Hadley circulation. The lower atmosphere in the central NPO had a strong northerly airflow and conducted a southerly airflow blowing northward along the coast of North America, and the Aleutian low pressure was strengthened. At the same time, a strong northwest airflow over the central NPO resulted in cold ocean advection, leading to a wide range of negative SOTA in that area. During El Niño strengthening periods in the western NPO, the troposphere was controlled by the abnormal updraft of the Hadley circulation with high-level divergence and low-level convergence, corresponding to its southern side tropical Hadley circulation downdraft branch. The lower atmosphere had a strong southerly airflow resulting in the warm ocean advection and the appearance of the significant positive SOTA in that area. Owing to the fact that the abnormal Hadley circulation cannot be formed in the initial phase of El Niño, the SOTA in the NPO appeared only during its developing period. It is still unclear how to interpret the intensification and maintenance of the SOTA in the NPO during the decay of El Niño events. Previous studies show that there is a strong air-sea interaction process in the mid-latitudes, especially in winter. In the central NPO, once sub-surface ocean temperature declines, conditions will be more conducive to the low-level atmospheric divergence and the development of the downdraft. Meanwhile, the northerly airflow behind the Aleutian trough and the southerly airflow ahead of the Aleutian trough would be strengthened, such that the Aleutian Low will be strengthened. As a reaction, the latter will promote cold ocean advection in the central NPO, resulting in reduction of sub-surface ocean temperature. This process is possible because of the strengthening Aleutian Low, where the divergence center located in the lower troposphere near 30 N deviates southward. This positive air-sea feedback process may be the main power source resulting in the intensification and maintenance of the SOTA in the central NPO during El Niño decay periods. This also compensates the energy reduction caused by the gradual weakening intensity of the Hadley circulation. Then, the SOTA center moves eastwards to the Gulf of Alaska along the North Pacific sub-polar front until the opposite change in atmospheric circulation. The opposite positive feedback process appears in the western NPO, when the SOTA increases in the seas south of Japan. This situation is conducive to the updraft, and when the southerly flow is strengthened, warm ocean advection occurs, and sub-surface ocean temperature rises again. Thus, the SOTA is strengthened and maintained there during El Niño decay periods. As this feedback process finishes locally, and the SOTA center disappears, a subsequent opposite change in atmospheric circulation occurs. Thus, the formation of the SOTA in the NPO, along with the ENSO process is firstly introduced by the northern branch of the abnormal Hadley circulation, and is strengthened and maintained by abnormal Hadley circulation and the positive feedback of the mid-latitude air-sea interaction.

11 1314 CHIN. J. OCEANOL. LIMNOL., 28(6), 2010 Vol.28 However, the direct factor resulting in SOTA is mainly the cold/warm ocean advection effects. 7 DISCUSSION AND CONCLUSIONS Previous studies have focused mainly on the formation and evolution of the SSTA in the NPO associated with the ENSO. Such studies have indicated that this process is caused by the sea surface heat flux changes in the NPO, resulting from the Atmospheric Bridge. Wang et al. (2000) pointed out that abnormal warming in the eastern equatorial Pacific produces an abnormal anti-cyclone in the lower troposphere of the northwest Pacific Philippine Sea. The positive and negative SSTA appears in this area and about 20 longitude east of this area, respectively. On the other hand, the negative SSTA can strengthen the anti-cyclone, and a positive feedback mechanism will be formed through the sea surface heat flux. The maintenance of the abnormal anticyclone in the western NPO during El Niño decay periods can be explained by the Kelvin wave in the Western Pacific generated by the tropical Indian Ocean warming (Xie et al., 2009). The Kelvin wave leads to reduced equatorial sea level pressure, and airflow divergence north of equator suppresses Northwest Pacific Ocean convection. These results reveal the same phenomenon as in the present paper, but their physical interpretation was based on a thermal air-sea interaction which is opposite to that in the present paper. In fact, SSTA is not only controlled by sea surface thermal processes, but also is affected by thermocline variations (i.e., when the thermocline rises, the SST decreases, and vice versa). This response was particularly apparent in this study in the central NPO and the waters south of Japan, where the thermocline was shallow. The changes in subsurface ocean temperature were different from the SST, and were achieved mainly through the air-sea dynamic interaction and the air-sea thermal interaction, respectively. In this paper we compared the SOTA and SSTA in the NPO. The results show that their spatial and temporal distributions were consistent, indicating SOTA significantly affects the SSTA, and the differences in their magnitude and scope mainly result from the sea surface thermal effect. The results of previous research (Peng et al., 2000; Ying et al., 2000) and this paper show that, during El Niño events, the Western Pacific subtropical high is often weaker than in normal year and its ridge line lies to the south. At these times, the Aleutian Low is stronger than normal years, and northwest airflow in the central NPO is stronger. The weakened Western Pacific subtropical high was closely related to abnormal Hadley circulation in the western NPO during El Niño which promotes weakened tropical convection and strengthened convection over the seas south of Japan. According to the studies of Huang et al. (1994) and Zhang et al. (1998), the West Pacific subtropical high is weak and the ridge line lies relatively southerly. At the same time, from the East China Sea to south of Japan, where the mainstream area of the East China Sea Kuroshio is located, the SOTA mainly is transported by the SW warm advection. Clearly, the SOTA in that area is closely related with the abnormal strength of the East China Sea Kuroshio. Research on the Kuroshio suggests that during the outbreak of El Niño or around the outbreak time, the Kuroshio volume often has surged, and is associated with the Large Meanders of Kuroshio (Japanese Kuroshio), at E south of Honshu and the large cold masses. Our results indicate the possible mechanism is that during El Niño years, the Western Pacific subtropical high weakens, its ridge line lies to the south, the SW airflow prevails in the East China Sea and south of Japan, the East China Sea Kuroshio warm advection is strengthened, and a positive SOTA center appears in the seas south of Japan on the south side of the Kuroshio. Meanwhile, owing to the strengthened Aleutian Low, cold advection in the central NPO and the Oyashio are strengthened. Therefore, during El Niño events, the volume of the East China Sea Kuroshio increases, and converges in the seas south of Japan, restraining the volume and direction of the Japanese Kuroshio. This is conducive to the Meanders of the Kuroshio. At the same time, the strengthened Oyashio north of the Japanese Kuroshio is beneficial to the formation of the cold masses. The combined effect is likely to have some influence on the Large Meanders of Kuroshio south of Japan and the associated large-scale cold masses. The opposite situation in La Niña years also can be explained accordingly. During El Niño, abnormal MZC indeed exists in the mid-latitude troposphere. The results in the present paper show that the zonal circulation circles with the opposite phase in the central and western NPO are actually composed of two Hadley northern branches in the NPO. However, the other circulation circle in the Gulf of Alaska may result from the downstream effects of the atmospheric circulation. The subsiding divergence in the NPO leads to a rise in pressure behind the Aleutian trough,

12 No.6 CHEN et al.: Subsurface variability during ENSO cycle 1315 intensification of the northerly wind and development of the Aleutian Low. Then, the southerly airflow ahead of the pressure trough is strengthened, resulting in the convergent updraft there. As a result of ENSO events, the abnormal MZC also is an atmospheric phenomenon in the NPO following the ENSO cycle. The formation and evolution of the SOTA in the NPO associated with the ENSO can be summarized as the following three processes: 1. El Niño (La Niña) developing period: The abnormal Walker circulation and abnormal Hadley circulation over the western and central NPO are strengthened (weakened). The Western Pacific subtropical high is weakened (strengthened), its ridge line has a southerly (northerly) shift, the SW air flow is strengthened (weakened) in the East China Sea to southern waters of Japan, +(-)SOTA appears. The North Pacific subtropical high is strengthened (weakened), its ridge has a northerly (southerly) shift, NW airflow is strengthened (weakened) in the Central NPO, -(+)SOTA is generated. 2. El Niño (La Niña) mature and decay periods: The abnormal Hadley circulation in the western and central NPO is weakened, and the northern branch of the lower troposphere Hadley moves southward. The mid-latitude air-sea positive feedback is strengthened. At the same time, the Western Pacific subtropical high is weakened (strengthened), resulting in the strengthening (weakening) SW airflow in the East China Sea to southern waters of Japan. On the contrary, the large scope of strengthened +(-)SOTA weakens (strengthens) the Western Pacific subtropical high. Meanwhile, the strengthening (weakening) Aleutian Low results in the strengthening (weakening) NW airflow in the central NPO, a wide range of -(+)SOTA is strengthened and slowly moves eastward, and further strengthens (weakens) the Aleutian Low. 3. After El Niño (La Niña) vanishes: The +(-)SOTA disappears in the western NPO; the -(+)SOTA is weakened in the eastern NPO. References Alexander M A, Blade I, Newman M et al The atmospheric bridge: the influence of ENSO teleconnections on air-sea interaction over the global oceans. J. Climate, 15: Carton J A, Chepurin G, Cao X et al. 2000a. A simple ocean data assimilation analysis of the global upper ocean , Part 1: methodology. J. Phys. Oceanogr., 30: Carton J A, Chepurin G, Cao X. 2000b. A Simple Ocean Data Assimilation analysis of the global upper ocean , Part 2: results. J. Phys. Oceanogr., 30: Chao J P, Yuan S Y, Chao Q C et al The origin of warm water in warm pool subsurface of the western tropical Pacific-the analysis of the El Niño. Chin. J. Atmos. Sci., 27(2): Frankignoul C Sea surface temperature anomalies, planetary wave, and air-sea feedback in the middle latitudes. Rev. Geophys., 23: Gong Y F, He J H, Duan Y Y, et al Numerical experiment on the influences of minus-ssta over mid-latitude Northern Pacific on the subtropical anticyclone. J. Trop. Meteor., 22(4): (in Chinese with English abstract) Huang R H, Sun F Y Impacts of the thermal state and the convective activities in the tropical western Pacific warm pool on the summer climate anomalies in East Asia. Scientia Atmospherica Sinica, 18(2): (in Chinese with English abstract) Kalnay E, Kanamitsu M, Kistler R et al The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc., 77: Liu Q, Wen N, Liu Z An observational study of the impact of the North Pacific SST on the atmosphere. Geophys. Res. Lett., 33: L Miller A J, White W B, Cayan D R North Pacific thermocline variations on ENSO timescales. J. Phys. Oceanogr., 27: Peng J Y, Sun Z B Influence of Spring Equatorial Eastern Pacific SSTA on Western Pacific Subtropical High. Journal of Nanjing Institute of Meteorology, 23: (in Chinese with English abstract) Wang B, Wu R G, Fu X H Pacific-east Asian teleconnection: how does ENSO affect Asian climate. J. Climate, 13: Wang C Atmospheric circulation cells associated with the El Niño-Southern Oscillation. J. Climate, 15: White W B Design of a global observing system for gyre-scale upper ocean temperature variability. Progress in Oceanography, 36: Xie S P, Hu K, Hafner J et al Indian Ocean capacitor effect on Indo-western Pacific climate during the summer following El Niño. J. Climate, 22: Ying M, Sun S Q A Study on the Response of Subtropical High over the Western Pacific on the SST anomaly. Acta. Oceanol. Sin., 24: (in Chinese with English abstract) Zhang Q Y, Tao S Y Tropical and subtropical monsoon over East Asia and its influence on the rainfall over eastern China in summer. Quarterly Journal of Applied Meteorology, 9: Zhao Y P, Chen Y L, Wang F et al Mixed-layer water oscillations in tropical Pacific for ENSO cycle. Sci. China Ser.: D-Earth Sci., 50(12):