Southern Ocean SST Variability and Its Relationship with ENSO on Inter-Decadal Time Scales

YAN Li, DU Yan*, and ZHANG Lan

?

Southern Ocean SST Variability and Its Relationship with ENSO on Inter-Decadal Time Scales

YAN Li, DU Yan, and ZHANG Lan

,,,510301,

Empirical orthogonal function (EOF) analysis reveals a co-variability of Sea surface temperatures (SSTs) in the Southern Hemisphere (0?–60?S). In the South Indian and Atlantic Oceans, there is a subtropical dipole pattern slanted in the southwest-northeast direction. In the South Pacific Ocean, a meridional tripole structure emerges, whose middle pole co-varies with the dipoles in the South Indian and Atlantic Oceans and is used in this study to track subtropical Pacific variability. The South Indian and Atlantic Ocean dipoles and the subtropical Pacific variability are phase-locked in austral summer. On the inter-decadal time scales, the dipoles in the South Indian and Atlantic Oceans weaken in amplitude after 1979/1980. No such weakening is found in the subtropical South Pacific Ocean. Interestingly, despite the reduced amplitude, the correlation of the Indian Ocean and Atlantic dipoles with El Ni?o and Southern Oscillation (ENSO) are enhanced after 1979/1980. The same increase in correlation is found for subtropical South Pacific variability after 1979/1980. These inter-decadal modulations imply that the Southern Hemisphere participates in part of the climate shift in the late 1970s. The correlation between Southern Hemisphere SST and ENSO reduces after 2000.

Southern Ocean SST; ENSO; subtropical dipole; inter-decadal time scales

1 Introduction

In recent years, subtropical dipoles were found in the middle latitude of the Southern Ocean. One is the dipole pattern in the South Indian Ocean (SIOD) (Behera and Yamagata, 2001). The other is the sea surface temperature (SST) dipole pattern in the South Atlantic Ocean (SAOD) (Venegas, 1997, Nnamchi, 2011). Wang (2010a) found two sets of dipole-like pattern in the South Pacific. The eastern SST dipole is referred to as the subtropical dipole mode in the South Pacific Ocean (Huang and Shukla, 2006; Wang, 2010a). Note that this pattern in the South Pacific Ocean (Wang, 2010a) is a coupled mode derived from SST and wind field by the Singular Vector Decomposition (SVD). As the leading pattern, it explains about 10%/9% covariance in ERA-40/NCEP reanalysis. The dipole is the most important inter-annual signal in the southern subtropics. Such SST dipole events could have an impact on the rainfall over the subtropical continents by modulating atmospheric circulation and convection (Reason, 2001; De Almeida, 2007).

For the SST pattern, a few studies highlighted that dipoles only exist in the South Atlantic and Indian Oceans, concurring with each other (Fauchereau, 2003; Ter-ray, 2011). Meanwhile a meridional tripole pattern exists in the South Pacific Ocean (Terray, 2011), whose middle pole (hereafter the single pole) in this study is used to track subtropical Pacific variability. South Pacific SST tripole pattern is different from the dipole pattern in the South Atlantic and Indian Oceans. This is probably due to the wider Pacific zonal basin.

A positive (negative) phase of dipole mode is featured by warm (cool) SST anomalies in the southwest and by cool (warm) SST anomalies in the northeast of the South Indian and Atlantic Oceans, respectively. For the formation of dipole, sea-level air pressure anomalies modulate wind anomalies, which makes an SST dipole. In the South Indian/Atlantic Ocean, a positive dipole is produced by strengthened Mascarene/St Helena anticyclone, and vice versa for a negative dipole (Fauchereau, 2003). During the positive event, over the cool pole, stronger anticyclones lead to southeasterlies west of Australia and southern Afria, resulting in increase of evaporation and upper-ocean mixing. The anomalous winds induce offshore Ekman transport and coastal upwelling. All those factors tend to decrease SST in the region. Over the warm pole, anomalous anticyclones induce the decrease of westerlies, which reduces evaporation and upper-ocean mixing and then favors the SST warming (Hermes and Reason, 2005; Wang, 2010a).

So far, the relationshipbetweendipoleandENSOis still not clear. Early studies suggest that the SIODandENSOareindependent of each other(Behera andYamagata, 2001) or only a weak lead-lag relationship exists (Fauchereau, 2003). Hermes and Reason (2005) showed thatSSTsignal of SIOD/SAODleadsNi?o-3.4 index by about 4/1.5 months. Recent studiesrevealed that the Ni?o-3.4 SST index lags the subtropical dipole by 9–10 months after the mid 1970s’ climate shift (Terray and Dominiak, 2005; Terray, 2011). If so,the subtropicaldipolecan bea good predictor ofENSO.

However, almost no study pays attention to the subtropical SST variability in strength. Besides, the relationship between the subtropical South Pacific and ENSO has seldom been considered. This paper aims to answer the following questions: how the SST variability in strength varies on inter-decadal scales in these individual Oceans and the entire Southern Ocean? On what time scale, the South Pacific, South Atlantic and Indian Oceans relate with ENSO, and what is the relationship between them? Evidences from reanalysis data will be used to answer these questions in this paper.

2 Data and Methods

2.1 Data

The dataset of National Oceanic and Atmospheric Administration (NOAA) Extended Reconstructed Sea Surface Temperature (ERSST) V3b (Smith, 2008; Xue, 2003) is used in this study. It is a global monthly SST analysis from 1854 to the present, derived from the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) data (Woodruff, 2011) with missing data filled in by statistical methods, with a 2.0? latitude by 2.0? longitude resolution.

The National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis-1 surface wind (Kalnay, 1996; Kistler, 2001) is used in this paper. It is a global monthly, 0.995 sigma level wind data from 1948 to the present, with a 2.5?×2.5? horizontal resolution.

2.2 Methods

In order to extract the spatial patterns of variability, Empirical orthogonal function (EOF) analyses of the SST anomalies during 1948 and 2010 are performed on South Indian Ocean (30?E–120?E, 60?S–0?), South Atlantic Ocean (60?W–20?E, 60?S–0?), South Pacific Ocean (150?E–70?W, 60?S–0?), and the entire southern middle latitude ocean (0?–358?E, 60?S–0?). In the subtropics of the Southern Hemisphere, SST has the maximum variability in austral summer (Suzuki, 2004), under the weaker winds and stronger solar radiation over this region. In the light of such a seasonality, previous studies about subtropical SST variability focus on January-February-March (JFM, Fauchereau, 2003; Wang, 2010a; Terray, 2011). We follow this convention and do a seasonal EOF of SST anomalies and wind anomalies in the JFM season. An annual EOF of SST anomalies is performed to make a confirmation. By EOF, key areas of dipoles or single pole are identified.

A 21-year running root mean square (RMS) represents the strength of the dipoles or individual pole. The 21-year running correlation coefficients between Ni?o-3.4 index and each specific SST index are calculated to represent the lead-lag correlation between ENSO events and Southern Ocean SST variability. Regression coefficients of tropical Pacific SST with each specific Southern Ocean SST index before and after the climate shift are calculated respectively, too. Before the calculations (EOF, 21-year running RMS,), a detrend and a 4–120 months band pass filtering are applied.

3 Results

3.1 Spatial Pattern

Seasonal EOF (Fig.1) shows a SST dipole pattern in the Southern Indian (Fig.1a) and Atlantic Oceans (Fig.1b), respectively. There is a meridional tripole pattern in the Southern Pacific Ocean (Fig.1c). The SST EOF1 spatial patterns (JFM) are consistent with Terray (2011) EOF1 pattern for December-January (DJ) and February-March (FM) months. In the EOF analysis for the entire Southern Ocean (SO), the first mode (Fig.1d) exhibits the same spatial distribution as patterns derived from individual ocean basins. The correlation of the time coefficients illustrates that such SST anomaly patterns change synchronously (Table 1). Specifically, when there is a positive SIOD/SAOD, there is a warm SSTa pole in the subtropical southern Pacific and two cool SSTa poles in the equatorial region and higher latitudes (Fig.1d).

Table 1 Cross-correlation between the time coefficients of seasonal EOF1

Note: Cross-correlation between the time coefficients of the leading EOF modes (EOF1) of the South Indian (IO), Atlantic (AO), Pacific (PO) Oceans and the entire Southern Ocean (SO) SSTa during January-February-March season for the 1948–2010 period.

The synchronization of the three basins’ SST change reflects the oceanic response to the climatic wavenumber-3 anticyclonic winds and the anomalous wavenumber-4 anticyclonic winds (Fauchereau, 2003; Sterl and Hazeleger, 2003; Hermes and Reason, 2005; Wang, 2010a). Our results confirm this point. The leading mode of seasonal EOF for wind anomalies shows that SST dipole is mainly caused by the anticyclonic winds respectively in the South Indian (Fig.1a) and Atlantic (Fig.1b) Oceans. Meanwhile, in the South Pacific (Fig.1c), the single pole is not directly induced by wind anomalies. The distribution of EOF1 of wind anomalies in the Southern Ocean (Fig.1d) is consistent with that of each individual basin. Our results illustrated that the dipole and related anticyclonic winds are not only the dominant coupled mode as shown in previous work (Wang, 2010a), but also the dominant uncoupled mode in the subtropical latitudes. In addition, it is indicated that a positive Wind-Evaporation-SST (WES) feedback (Xie, 1993a, b) is partly responsible for the subtropical dipole SST pattern (Wang, 2010b). Note that Terray’s (2011) and Wang’s (2010a) EOF/SVD domains do not include the area 0?–10?S. In the present study, subtropical dipoles are still the most significant signal even if such tropical regions are included in the Indian and Atlantic Oceans. Basically, the annual EOF results show the same pattern as those from the seasonal EOF: dipole mode in the Indian and Atlantic Oceans, tripole mode in the Pacific Ocean. In the tropical Atlantic Ocean there is an ‘El Ni?o-like’ signal (Fig.2b) similar to that in the Pacific Ocean. In the Indian Ocean, the subtropical SST variability is stronger than that in tropics, since subtropical dipole accounts for a dominant mode throughout the year (Fig.2a).

The variance contributions of the two leading EOF modes for seasonal/annual EOF suggest that dipole is the typical mode in the South Indian Ocean and Atlantic Ocean (Tables 2 and 3). Furthermore, in the Indian Ocean, the dipole is more typical than in the Atlantic Ocean. In seasonal EOF analysis, for the South Indian Ocean, the dipole mode (, EOF1) accounts for 31% variance contribution, larger than EOF2’s 12% variance contribution, by 19% (Table 2). For the South Atlantic Ocean, variance contribution of EOF1 is larger than that of EOF2 by only 9% (Table 2). In annual EOF analysis, the variance difference between EOF1 and EOF2 is 8% (20% minus 12%) and 5% (19% minus 14%) for Indian and Atlantic Ocean respectively (Table 3). Hence, annual EOF analysis (Table 3) also supports that dipole in the South Indian Ocean is more typical than in the South Atlantic Ocean.

Fig.1 Spatial pattern of seasonal (January-February-March) EOF1 (leading mode) for (a) the southern Indian Ocean (b) southern Atlantic Ocean (c) southern Pacific Ocean and (d) southern Ocean SST anomalies (shadings and contours)/wind anomalies (vectors) during the 1948–2010 period. The positive-negative SST anomalies correspond to positive SIOD/SAOD/SPO events. Black boxes denote the area with large SST variability.

Fig.2 Same as in Fig.1, but for annual EOF1, without wind anomalies.

In order to quantify such dipoles and single pole, large SST variability (black boxes in Figs.1 and 2 denote the region) is averaged to form an index. SIOD index is obtained from the SST anomaly difference between the western (45?E–75?E, 50?S–33?S) and eastern (82?E–108?E, 36?S–18?S) subtropical Indian Ocean, then it is divided by 2. Similarly, SAOD index is obtained from the SST anomaly difference between the western (35?W–6?W, 52?S–32?S) and eastern (22?W–10?E, 30?S–16?S) subtropical Atlantic Ocean. South Pacific (SP) index is the average of the SST anomaly in the region (200?E–260?E, 38?S–24?S). We also construct a Southern Ocean (SO) index, which is the mean of SIOD, SAOD and SPO indices.

Table 2 Seasonal EOF variance contribution of EOF1 and EOF2

Note: Seasonal EOF variance contribution of the first and second leading EOF modes (EOF1 & EOF2) of the South Indian (IO), Atlantic (AO), Pacific (PO) Oceans and the entire Southern Ocean (SO) SSTa during January-February-March season for the 1948–2010 period.

Table 3 Annual EOF variance contribution of EOF1 and EOF2

Note: Same as in Table 2, but for annual EOF.

3.2 Decadal Variation in Amplitude of the Dipole and Single Pole

The 21-year running RMSs of SIOD, SAOD, and SPO indices indicate that the two dipoles are phase-locked in JFM,, the austral summer (Fig.3), consistent with previous studies and the EOF analysis above. Both SIOD and SAOD weaken in amplitude after the 1980s (Figs.3a and 3b). Strength of SSTa variability in the subtropical South Pacific (represented by a single pole area) has little change before and after the 1980s (Fig.3c). The maximum of RMS of SPO index exhibits a sign of moving to March-April after the 1980s (Fig.3c). Further, Fig.4 gives the 21-year running RMS of the western and eastern pole of the dipoles. Results show that the weakening of SIOD is mainly attributed to the weakening of its western pole (Fig.4a). Strength of SIOD’s eastern pole keeps stable (Fig.4b). Meanwhile both the western pole (Fig.4c) and eastern pole (Fig.4d) of SAOD are on the wane.

Fig.4 Twenty-one-year running RMS for (a) SIOD west pole, (b) SIOD east pole, (c) SAOD west pole, and (d) SAOD east pole indices. The years on x-axis denote the centers of sliding windows.

3.3 Decadal Shift of the Southern Ocean SST Relationship with ENSO

Fig.5 shows the 21-year running correlation of SIOD, SAOD, SPO, and SO indices with the November(0)–January(1) (ND(0)J(1)) Ni?o-3.4 SST index, as a function of year and calendar month. Note that all the months of-axis in Fig.5 belong to year(0). The years of-axis in Fig.5 denote the centers of sliding windows;, 1958 represents the correlation in the 21-year sliding window of 1948–1968. Fig.5a indicates that the correlation of SIOD leading ENSO by 9–11 months is enhanced after the 1980s. Subtropical Indian Ocean SSTs in January(0)–March(0) (JFM(0)) are highly correlated with ENSO after the 1980s; meanwhile there is no such relationship before the 1980s (Fig.5a). For SAOD and SPO, their relationship with ENSO has a similar decadal shift, except that the year SAOD/SPO-ENSO lead-lag relationship becomes closer slightly later than SIOD (Figs.5b and 5c). In addition, SAOD and ENSO events have a strong synchronous negative-correlation after the 1980s (Fig.5b). For all years, SPO and ENSO events naturally have a strong synchronous negative-correlation (Fig.5c). These reflect the basin differences among the three oceans.

Southern Ocean JFM(0) SST indices and ND(0)J(1) Ni?o-3.4 index are shown in Fig.6. Before the 1980s, SST variations in subtropical southern hemisphere seem independent of ENSO. After the 1980s, ENSO warm/cold events tend to lag by 9–11 months behind the positive/negative SIOD/SAOD/SPO events (Figs.6a, 6b and 6c). Take SIOD as an example, in years 1982, 1986, and 1997, positive SIOD appears in JFM(0), El Ni?o events occur later in ND(0)J(1); meanwhile in years 1984, 1988, and 1995, La Ni?a events occur 10 months later following negative SIOD events (Fig.6a). However, this strong lead-lag correlation weakens after year 2000. As for the amplitude of correlation, before the 1980s, SAOD and ENSO have a low value (only 0.02); afterwards, it rises to 0.43 (Fig.6b). Correlation of southern SST and Ni?o-3.4 SST index has a similar decadal shift,, from 0.15 to 0.36 for SIOD (Fig.6a), and from 0.1 to 0.5 for SPO (Fig.6c).

Fig.5 Twenty-one-year running correlation of ND(0)J(1) Ni?o-3.4 index with SST indices: (a) SIOD, (b) SAOD, (c) SPO, and (d) SO. On y-axis, the month 1 represents January(0), the month 12 represents December(0). The years on x-axis denote the centers of sliding windows.

Fig.6 ND(0)J(1) Ni?o-3.4 index (red line) and JFM(0) Southern Ocean SST indices (blue line): (a) SIOD index, (b) SAOD index, (c) SPO index, and (d) SO index. In the upper left corner of each sub-figure, correlation between Ni?o-3.4 index and southern SST index is shown for pre-epoch (1948–1979) and post-epoch (1980–2010).

Through the regressions, we found the impact of Southern Ocean SST on the tropical Pacific SST 9–11 months later. Assume regression equation is=, whererepresents the independent variable (, SIOD index),the dependent variable (tropical Pacific SST),the regression coefficient. Fig.7 shows that before the 1980s, regression is very low (<0.6) over the entire Pacific region whatever upon the SIOD, SAOD or SPO index; after the 1980s, regression can reach 2.5. In Ni?o-3.4 region, regression ranges from 1 to 2.5. In other words, after the climate shift, ENSO can be predicted 9–11 months in advance through the variation in Southern Ocean SST.

Fig.7 Regression of ND(0)J(1) tropical Pacific SST anomalies (℃) upon JFM(0) southern SST index: (a)–(b) SIOD index, (c)–(d) SAOD index, and (e)–(f) SPO index. The left column shows the pre-epoch (1948–1979), the right column the post-epoch (1980–2010).

4 Summary and Discussion

Our analysis found that after the abrupt climate change, subtropical dipoles lead ENSO by 9–11 months. This result confirms those from previous study of Terray (2011). In the post-climate-change epoch, the way for the southern SSTa to influence ENSO is not clear. Tropical southeast Indian Ocean is part of SIOD, this region may play an important role in Southern Ocean SSTa influencing ENSO. At the same time, it is worth noting that ENSO itself experiences an abrupt climate change. Before the climate shift, SST anomalies first emerge near the western coast of South American, and then propagate westward. After the climate shift, SST anomalies first emerge in the central equatorial Pacific, and then propagate eastward to reach the western coast of South American (Wang, 1995). In addition, after the mid-1970s, the frequency of the occurrence of El Ni?o Modoki (or Central Pacific El Ni?o) increases (Yeh, 2009). Those changes for ENSO as well as the SIOD decadal variations may be part of the global decadal variation.

For the dipole pattern in the southern Oceans, the SIOD shows the strongest variability (Table 2 and Fig.3). The EOF1 mode, SIOD, in Indian Ocean accounts for the largest variance contribution, 31%, while it is 24% for SAOD (Table 2). The former also has larger RMS variation (Figs.3a and 3b). There is an abrupt shift for the strength of SIOD and SAOD in the 1980s (Figs.3a and 3b). Meanwhile the strength of the single pole in the south Pacific is almost unchanged (Fig.3c). As a whole, the evolution of dipoles and single pole in the Southern Oceans is considered to be the oceanic response to the anti- cyclonic winds associated with the subtropical high-pressure system (Fauchereau, 2003; Sterl and Hazeleger, 2003; Hermes and Reason, 2005; Wang, 2010a).

Although the strength of SIOD and SAOD weaken after the late 1970s, their relationships with ENSO are enhanced (Figs.5, 6 and 7). In addition, South Pacific and ENSO correlation has a decadal shift, which is enhanced in recent decades. The lead-lag relationship between the Southern Ocean SST and ENSO helps to make climate prediction (Fig.7). Interestingly, such high correlation seems reduced after 2000 (Fig.6). Take the South Indian Ocean for an example, SIOD and ENSO events tend to occur independently after year 2000 (, 2001, 2005, 2006, and 2007) (Fig.6a). SAOD and SPO show similar tendency (Figs.6b and 6c). Such reduction of relationship may be associated with decadal background, such as Pacific Decadal Oscillation (PDO) (Mantua, 1997, Deser, 2010). For PDO mode, it has a negative-phase from 1947–1976, and turns to positive-phase from 1977–1998, then to a rather weaker negative phase (Deser, 2010). Note that PDO’s negative-phase corresponds to the period that the southern Ocean SST and ENSO have weak relationship, and vice versa. It is not clear whether PDO is modulated by the global warming. We will work on these interesting issues to enhance our understanding on air-sea interaction in the tropical and subtropical Indo-Pacific region.

Acknowledgements

This work is jointly supported by the National Basic Research Program (2012CB955603, 2010CB950302), National High Technology Research and Development Program of China (No. 2010AA012304), and the Knowledge Innovation Program of the Chinese Academy of Sciences (SQ201006 and XDA05090404).

Behera, S. K., and Yamagata, T., 2001. Subtropical SST dipole events in the southern Indian Ocean., 28: 327-330, DOI: 10.1029/2000GL011451.

De Almeida, R. A. F., Nobre, P., Haarsma, R. J., and Campos, E. J. D., 2007. Negative ocean–atmosphere feedback in the South Atlantic Convergence Zone., 34, L18809, DOI: 10.1029/2007GL030401.

Deser, C., Alexander, M. A., Xie, S.-P., and Phillips, A. S., 2010. Sea surface temperature variability: Patterns and mechanisms.,2: 115-43.

Fauchereau, N., Trzaska, S., Richard, Y., Roucou, P., and Camberlin, P., 2003. SST co-variability in the southern Atlantic and Indian Oceans and its connections with the atmospheric circulation in the Southern Hemisphere., 23: 663-677.

Hermes, J. C., and Reason, C. J. C., 2005. Ocean model diagnosis of interannual coevolving SST variability in the South Indian and South Atlantic Oceans., 18: 2864-2882.

Huang, B., and Shukla, J., 2006. Interannual SST variability in the southern subtropical and extra-tropical ocean.223. Center for Ocean-Land-Atmosphere Studies (COLA), Calverton, Maryland, 20pp.

Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Leetmaa, A., Reynolds, R., Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K. C., Ropelewski, C., Wang, J., Jenne, R., and Joseph, D., 1996. The NCEP/NCAR 40-year reanalysis project., 77: 437-470.

Kistler, R., Collins, W., Saha, S., White, G., Woollen, J., Kalnay, E., Chelliah, M., Ebisuzaki, W., Kanamitsu, M., Kousky, V.,

van den Dool, H., Jenne, R., and Fiorino, M., 2001. The NCEP-NCAR 50-year reanalysis: Monthly means CD-ROM and documentation., 82: 247-267.

Mantua, N., Hare, S. J., Zhang, Y., Wallace, J. M., and Francis, R. C., 1997. A Pacific interdecadal oscillation with impacts on salmon production., 78: 1069-1079.

Nnamchi, H. C., Li, J., and Anyadike, R. N. C., 2011. Does a dipole mode really exist in the South Atlantic Ocean?, 116, D15104, DOI: 10.1029/2010JD015579.

Reason, C. J. C., 2001. Subtropical Indian Ocean SST dipole events and southern African rainfall., 28: 2225-2227, DOI: 10.1029/2000GL012735.

Smith, T. M., Reynolds, R. W., Peterson, T. C., and Lawrimore, J., 2008. Improvements to NOAA’s historical merged land-ocean surface temperature analysis (1880–2006)., 21: 2283-2296.

Sterl, A., and Hazeleger, W., 2003. Coupled variability and air-sea interaction in the South Atlantic Ocean.,21: 559-571.

Suzuki, R., Behera, S. K., Iizuka, S., and Yamagata, T., 2004. Indian Ocean subtropical dipole simulated using a coupled general circulation model., 109, C09001, DOI: 10.1029/2003JC001974.

Terray, P., 2011. Southern Hemisphere extra-tropical forcing: a new paradigm for El Ni?o-Southern Oscillation., 36: 2171-2199.

Terray, P., and Dominiak S., 2005. Indian Ocean sea surface temperature and El Ni?o-Southern Oscillation: a new perspective., 18: 1351-1368.

Venegas, S. A., Mysak, L. A., and Straub, D. N., 1997. Atmosphere–ocean coupled variability in the South Atlantic., 10: 2904-2920.

Wang, B., 1995. Interdecadal changes in El Ni?o onset in the last four decades., 8: 267-285.

Wang, F., 2010a. Subtropical dipole mode in the Southern Hemisphere: A global view., 37, L10702, DOI: 10.1029/2010GL042750.

Wang, F., 2010b. Thermodynamical coupled modes in the tropical atmosphere–ocean: An analytical solution., 67: 1667-1677, DOI: 10.1175/2009JAS3262.1.

Woodruff, S. D., Worley, S. J., Lubker, S. J., Ji, Z., Freeman, J. E., Berry, D. I., Brohan, P., Kent, E. C., Reynolds, R. W., Smith, S. R., and Wilkinson, C., 2011. ICOADS release 2.5: Extensions and enhancements to the surface marine meteorological archive., 31: 951-967.

Xie, S. P., Kubokawa, A., and Hanawa, K., 1993a. Evaporation-wind feedback and the organizing of tropical convection on the planetary scale. Part I: Quasi-linear instability., 50: 3873-3883.

Xie, S. P., Kubokawa, A., and Hanawa, K., 1993b. Evaporation-wind feedback and the organizing of tropical convection on the planetary scale. Part II: Nonlinear evolution., 50: 3884-3893.

Xue, Y., Smith, T. M., and Reynolds, R. W., 2003. Interdecadal changes of 30-yr SST normals during 1871–2000., 16: 1601-1612.

Yeh, S. W., Kug, J. S., Dewitte, B., Kwon, M. H, Kirtman, B. P., and Jin, F. F., 2009. El Ni?o in a changing climate., 461: 511-514.

(Edited by Xie Jun)

10.1007/s11802-013-2262-1

ISSN 1672-5182, 2013 12 (2): 287-294

. Tel: 0086-20-89023180 E-mail:duyan@scsio.ac.cn

(January 5, 2013; revised February 16, 2013; accepted March 18, 2013)

? Ocean University of China, Science Press and Springer-Verlag Berlin Heidelberg 2013