1. Introduction
Southern African rainfall reaches its peak during austral summer owing to the southward migration of the south Indian convergence zone (Cook 2000). Since it affects the rain-fed agriculture and the public health in southern Africa, its interannual variations are very important in the regional society. Several studies discussed its link with El Niño–Southern Oscillation (Lindesay 1988; Richard et al. 2000) and the interannual sea surface temperature (SST) variations in the southern Indian Ocean (Reason and Mulenga 1999; Behera and Yamagata 2001; Reason 2001), and some other studies focused on impacts of interannual SST variations in the South Atlantic Ocean (Walker 1990; Mason 1995; Reason 1998).
Following Richard et al. (2000), we introduce the southern African rainfall index (SARI) defined as the area-averaged land rainfall anomalies south of 10°S. Interestingly, the correlation between the SARI and SST anomalies during 1960–2008 in the South Atlantic shows a remarkable dipole pattern (Fig. 1); the SST anomalies in the southwestern (northeastern) region are positively (negatively) correlated with the SARI. Among studies devoted to the understanding of interannual SST variations in the South Atlantic, Venegas et al. (1996) was the first to discuss this SST dipole. After the similar phenomenon in the southern Indian Ocean named the Indian Ocean subtropical dipole (IOSD; Behera et al. 2000; Behera and Yamagata 2001; Suzuki et al. 2004), it is called the South Atlantic subtropical dipole (SASD). Venegas et al. (1997) showed that the atmosphere leads the ocean by 1–4 months by calculating a lag correlation between the principal components of the sea level pressure (SLP) and SST anomalies. Furthermore, Fauchereau et al. (2003) suggested that the SST anomalies are related to the latent heat flux anomalies associated with the southward migration as well as strengthening of the subtropical high. This variation in the subtropical high was suggested to have a link with the atmospheric pattern of zonal wavenumber 3 or 4 in the Southern Hemisphere. Only a few studies discussed the oceanic roles in the generation of the SASD, however. In this regard, Sterl and Hazeleger (2003) examined the mixed layer heat balance and showed that the anomalous Ekman heat transport contributes to the growth of SST anomalies even though the latent heat flux anomaly plays the dominant role. Because of the lack of ocean data, they used the climatological mean mixed layer depth derived from an ocean reanalysis in the heat balance calculation. Also, Hermes and Reason (2005) showed that the Ekman upwelling anomaly affects the evolution of SST anomalies using outputs from an ocean general circulation model (OGCM). Since the mixed layer depth undergoes significant seasonal and interannual variations in the subtropics, those studies that do not take the variations into account have serious flaws as we show later. In fact, Morioka et al. (2010) have recently discussed the importance of the interannual variations in the mixed layer depth on the growth of SST anomalies associated with the IOSD.
Correlation between the SARI and SST anomalies during 1960–2008. The contour interval is 0.1, and the values exceeding 90% confidence level in a two-tailed t test are shaded. The box over the southwestern (northeastern) pole is defined between 10° and 30°W and between 30° and 40°S (0°–20°W and 15°–25°S).
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
This paper investigates the mechanism of the SASD using outputs from an OGCM. It is organized as follows. A brief description of the observational data and an OGCM is given in the next section. In section 3, we discuss the annual cycle of the SST and the mixed layer depth in the subtropical South Atlantic. In section 4, we define the SASD events by introducing an SASD index (SASDI) and examine the growth and decay mechanisms of the SST anomalies associated with the SASD. The cause of the interannual variations in the mixed layer depth is also discussed by calculating the Monin–Obukhov depth. The final section summarizes the main results.
2. Observational data and OGCM design
We use the monthly mean observed SST data from the Hadley Centre sea ice and sea surface temperature (HadISST; Rayner et al. 2003). They are gridded data with 1° × 1° resolution, and we analyze the period of 1960–2008 because there were few observations in the southern part of the South Atlantic before the 1960s. Monthly SST anomalies are calculated by subtracting the monthly mean climatology after removing a linear trend using a least squares fit. To calculate the mixed layer depth, we use the monthly climatology of ocean temperature from the World Ocean Atlas 2009 (WOA09; http://www.nodc.noaa.gov). It has 24 levels in the vertical with 1° × 1° horizontal resolution. We also use the surface heat flux, specific humidity at 2-m height, and wind speed at 10 m above the surface from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis dataset (Kalnay et al. 1996). It covers the same period on a T62 Gaussian grid. Since qualitatively similar results are obtained using the European Centre for Medium Range Weather Forecasts reanalysis dataset (Uppala et al. 2005), we only show results from the NCEP–NCAR reanalysis dataset in this paper. For precipitation, we use the gridded monthly rainfall data provided by the University of Delaware (Legates and Willmott 1990) from 1960 to 2008 with a horizontal resolution of 0.5° in both longitude and latitude.
To calculate the mixed layer heat balance, we use outputs from an OGCM. The ocean model is based on the Modular Ocean Model, version 3.0 (MOM 3.0), developed at the National Oceanic and Atmospheric Administration/Geophysical Fluid Dynamics Laboratory (Pacanowski and Griffies 1999) and covers the global ocean from 65°S to 30°N. It has a horizontal resolution of 0.5° in both longitude and latitude and 25 levels in the vertical with eight levels in the upper 100 m. The bottom topography and coastlines adopted in this model are based on the 5-min earth topography (ETOPO5) dataset. The vertical eddy viscosity and diffusivity are calculated using the parameterization of Pacanowski and Philander (1981), while the lateral eddy viscosity and diffusivity are based on the formula given by Smagorinsky (1963). Near the southern and northern boundaries (poleward of 62°S and 27°N), values of these coefficients are increased steadily so that the damping time scale reaches one day at 65°S and 30°N. The temperature and salinity are relaxed to the monthly mean climatology (Levitus and Boyer 1994; Levitus et al. 1994) within the sponge layer so that artificial wall effects are reduced. All measures described above are conventional.
The model is first spun up for 20 yr by the monthly mean climatology of the wind stress from the NCEP–NCAR reanalysis dataset and by the surface heat flux calculated by the bulk formula using the simulated SST and the atmospheric variables obtained from the reanalysis data. The sea surface salinity (SSS) is restored to the monthly climatology with the relaxation time scale of 30 days. The initial condition is the annual mean climatology (Levitus and Boyer 1994; Levitus et al. 1994) with no motion. Then, the model is further integrated for 59 yr from 1950 to 2008 using the daily mean wind stress from the NCEP–NCAR reanalysis data and the daily surface heat flux calculated by the bulk formula using the simulated SST and the atmospheric variables obtained from the reanalysis data. The model SSS is again restored to the monthly mean climatology. Considering the oceanic adjustment time to the interannually varying forcing, we only analyze the outputs after 1960.
3. Annual cycle of the SST and the mixed layer depth in the subtropical South Atlantic
Annual march of (a),(b) the SST (°C), (c),(d) each term in Eq. (1) (10−7 °C s−1), and (e),(f) the mixed layer depth (m) at the (left) southwestern and (right) northeastern poles defined in Fig. 1. In (a) and (b), the SST from HADISST (solid line) and the model (dashed line) are shown. In (c) and (d), MLT tend., NSHF, Hor. adv., Ent., and Res. indicate the mixed-layer temperature tendency (thick solid line), net surface heat flux (thick dashed line), horizontal advection (thick dotted line), entrainment (thin solid line), and residual (thin dashed line) terms, respectively. In (e) and (f), the mixed-layer depths derived from the WOA09 (solid line) and the model (dashed line) are shown.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
Figures 2c and 2d show the monthly climatology of each term in Eq. (1) over the southwestern and northeastern poles. The mixed layer temperature tendency over both poles is positive (negative) from October to February (from March to September). This is explained mostly by the net surface heat flux dominated by the shortwave radiation (figure not shown). Also, the entrainment contributes to the cooling when the mixed layer deepens as shown in Figs. 2e and 2f; the mixed layer depth at both poles becomes deepest from July to September.
4. SASD events
The first empirical orthogonal function (EOF) mode of the observed SST anomalies in the South Atlantic shows a dipole pattern and explains 20.4% of the total variance (Fig. 3a). The second EOF mode with a monopole pattern explains 12.9% of the total variance and is well separated from the first mode (North et al. 1982). This dipole pattern of the first mode corresponds well to that in Fig. 1 derived from the correlation with the SARI. To capture this interesting SST variability in a simple way, we introduce the SASDI, which is defined by the difference in the SST anomalies between the southwestern pole and the northeastern pole. As expected, the SASDI shows a high correlation of 0.92 with the principal component of the first EOF mode presented in Fig. 3b.
(a) Spatial pattern (0.1°C) and (b) principal component of the first EOF mode of SST anomalies during 1960–2008. Negative values in (a) are shaded, and the boxes over the positive and negative poles are defined as in Fig. 1. The values in (b) are normalized by the standard deviation.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
The monthly standard deviation of the SASDI undergoes significant variations, with values greater than 0.8°C from December to March (Fig. 4a). This suggests that the interannual variations of the SASDI are mostly locked to the austral summer. Therefore, we define years in which the SASDI exceeds 1 standard deviation during December–March as SASD-event years (Fig. 4b). This procedure leads to 8 (11) positive (negative) SASD events (Table 1).
(a) Monthly standard deviation of the observed (bar) and simulated (line) SASDI (°C). (b) Time series of the observed (solid line) and simulated (dashed line) SASDI averaged from December to March during 1960–2008. The values in (b) are normalized by the standard deviation.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
Positive and negative SASD years used in the composite analysis.
The phase-locking nature of the SASD events is well reproduced in the simulation model as seen in Fig. 4a, and the time series of the simulated SASDI are highly correlated (0.95) with the observed SASDI (Fig. 4b). Hereinafter, we analyze the positive and negative SASD events in detail using both data and simulation results.
a. Positive SASD
Figure 5 shows composites of the observed and simulated SST anomalies for the positive SASD events over four seasons from austral spring to winter: September(0)–November(0), December(0)–February(1), March(1)–May(1), and June(1)–August(1). Both positive and negative SST anomaly poles start to develop synchronously from spring, reach their peak during summer, and decay after autumn. Since the pattern correlation between the observed and simulated SST anomalies of the peak (decay) phase in summer (autumn) is 0.94 (0.94), we may conclude that the model reproduces the evolution of composite SST anomalies very well except for some slight differences in magnitude.
Evolution of (left) observed and (right) simulated composite SST anomalies for the positive SASD. The contour interval is 0.2°C. The boxes over the positive and negative poles are defined as in Fig. 1. The shading indicates anomalies exceeding 90% confidence level in a two-tailed t test.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
The upper panels in Fig. 6 show composite anomalies of each term in Eq. (1) over the positive and negative poles. The mixed layer temperature anomalies over both poles develop significantly from November(0) to January(1). It is mostly due to the anomalous contribution from the net surface heat flux. The anomaly in the horizontal advection also contributes to the growth of the positive pole. This is dominated by the anomalous meridional advection owing to the anomalous southward Ekman and geostrophic flows (Sterl and Hazeleger 2003). Contributions from four components of the net surface heat flux are shown in Figs. 6c and 6d. It is clear that the anomalous contribution from the shortwave radiation plays the dominant role. This is in marked contrast with the previous results from Fauchereau et al. (2003) and Hermes and Reason (2005), who suggested the importance of the latent heat flux anomaly among four components of the net surface heat flux anomalies as shown in Figs. 6e and 6f. This interesting discrepancy needs to be solved.
Time series of composite anomalies of (a),(b) the mixed layer temperature tendency terms in Eq. (1) (10−7 °C s−1), (c),(d) components of the net surface heat flux term in Eq. (1) (10−7 °C s−1), and (e),(f) components of the net surface heat flux (W m−2) over the positive and negative SST anomaly poles for the positive SASD. A 3-month running mean is applied to smooth the time series. Filled (open) circle shows anomalies exceeding 90% (80%) confidence level in a two-tailed t test. In (a) and (b), MLT tend., NSHF, Hor. adv., Ent., and Res. indicate the mixed-layer temperature tendency (thick solid line), net surface heat flux (thick dashed line), horizontal advection (thick dotted line), entrainment (thin solid line), and residual (thin dashed line) terms, respectively. In (c) and (d), SW, LW, LH, and SH indicate the shortwave radiation (thick solid line), longwave radiation (thick dotted line), latent heat flux (thin solid line), and sensible heat flux (thin dashed line) terms, respectively. In (e) and (f), Qnet, QSW, QLW, QLH, and QSH indicate the net surface heat flux (thick dashed line), shortwave radiation (thick solid line), longwave radiation (thick dotted line), latent heat flux (thin solid line), and sensible heat flux (thin dashed line) anomalies, respectively.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
Time series of composite anomalies of (a),(b) the shortwave radiation terms in Eq. (3) (10−7 °C s−1) and (c),(d) the mixed layer depth (m) at the positive and negative SST anomaly poles for the positive SASD. Filled (open) circle shows anomalies exceeding 90% (80%) confidence level in a two-tailed t test. A 3-month running mean is applied to smooth the time series. In (a) and (b), the net shortwave radiation term (thick solid line), first term (thick dashed line), second term (thick dotted line), and residual term (thin solid line) are shown.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
Composite anomalies of the contributions from the wind stirring (black bar), surface heat flux (gray bar), and residual (white bar) terms in Eq. (5) at the (left) positive and (right) negative SST anomaly poles for the positive SASD (m). Here, ND(0) represents the average from November(0) to December(0), and JF(1) represents the average from January(1) to February(1).
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
Composite anomalies of sea level pressure (thick contours with interval of 0.5 hPa) and latent heat flux (thin contours with interval of 3 W m−2) averaged in (left) November(0)–December(0) and (right) January(1)–February(1) of the positive SASD. Latent heat flux anomalies exceeding 90% confidence level in a two-tailed t test are shaded.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
Furthermore, a significant difference in the mixed layer depth (Monin–Obukhov depth) anomaly between two poles is found especially in November(0)–December(0) as shown in Figs. 7c and 7d (Fig. 8). Besides the fact that the specific humidity difference near the surface over the northeastern pole is slightly larger than that over the southwestern pole, the wind speed at 10-m height over the northeastern flank of the subtropical high is nearly 2 times that over the southwestern flank (figure not shown). This causes larger latent heat loss over the negative pole in November(0)–December(0), leading to the smaller net surface heat flux in Eq. (5) than that over the positive pole. In addition, the absolute value of the latent heat anomaly over the negative pole is 2 times that over the positive pole, leading to the larger absolute value of the net surface heat flux anomaly (Figs. 6e,f). Both of these contribute to the asymmetry of the mixed layer depth (Monin–Obukhov depth) anomaly between the poles.
Time series of composite anomalies of (a),(b) the entrainment terms (10−7 °C s−1) in Eq. (6) and (c),(d) the latent heat flux terms (10−7 °C s−1) in Eq. (3) over the (left) positive and (right) negative SST anomaly poles for the positive SASD. Filled (open) circle shows anomalies exceeding 90% (80%) confidence level in a two-tailed t test. A 3-month running mean is applied to smooth the time series. In (a) and (b), the net entrainment term (thick solid line), first term (thick dashed line), second term (thick dotted line), third term (thin solid line), and residual term (thin dashed line) are shown. In (c) and (d), the net latent heat flux term (thick solid line), first term (thick dashed line), second term (thick dotted line), and residual term (thin solid line) are shown.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
b. Negative SASD
The evolution of the SST anomalies for the negative SASD is shown in Fig. 11. Both positive and negative SST anomaly poles start to develop synchronously from austral spring, reach their peaks during summer, and decay after autumn. The pattern correlation between the observed and simulated SST anomalies during the peak (decay) phase of summer (autumn) is 0.95 (0.93), and the evolution of the negative SASD is also well reproduced in the model.
As in Fig. 5, but for the negative SASD.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
The mixed layer temperature anomalies over both poles develop significantly from November(0) to January(1) mostly owing to the anomalous contribution from the net surface heat flux (Figs. 12a,b). In particular, the contribution from the shortwave radiation is dominant (Figs. 12c,d). This contribution is mostly explained by the second term on the right-hand side of Eq. (3) (Figs. 13a,b), indicating that, over the positive (negative) pole, the warming of the mixed layer by the contribution from the climatological shortwave radiation is enhanced (suppressed) by the thinner (thicker) mixed layer as shown in Fig. 13c (Fig. 13d). As indicated in Fig. 14, the quick decay as well as the amplitude of the mixed layer depth anomaly is well diagnosed by those of the Monin–Obukhov depth anomaly, which is mostly due to the contribution from the surface heat flux anomaly [the second and third terms on the right-hand side of Eq. (5)]. In particular, the latent heat flux anomaly plays the dominant role (Figs. 12e,f). This latent heat flux anomaly may also be linked with the weakening of the subtropical high in the South Atlantic (Fig. 15).
As in Fig. 6, but for the negative SASD.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
As in Fig. 7, but for the negative SASD.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
As in Fig. 8, but for the negative SASD.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
As in Fig. 9, but for the negative SASD.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
The mixed layer temperature anomalies in both poles decay during early (late) autumn owing to the anomalous contribution from the entrainment (net surface heat flux) in Figs. 12a and 12b. Over the positive (negative) pole, the cooling of the mixed layer by the entrainment is enhanced (suppressed) by the larger (smaller) temperature difference between the mixed layer and the entrained water in early autumn as shown in Fig. 16a (Fig. 16b). In particular, the positive (negative) mixed layer temperature anomaly mostly contributes to the temperature difference. In late autumn, the anomalous contribution from the net surface heat flux is mostly explained by that from the latent heat flux (Figs. 12c,d). The cooling of the mixed layer by the contribution from the climatological latent heat flux is enhanced by the thinner mixed layer at the positive pole (Fig. 16c). On the other hand, over the negative pole, the contribution from the positive latent heat flux anomaly contributes mostly to the damping of the negative mixed layer temperature anomaly (Fig. 16d). Although the mean mixed layer depth and latent heat flux over the negative pole are similar to those over the positive pole, the mixed layer depth anomaly at the negative pole is 0.2 m, much smaller than −1.8 m at the positive pole (Figs. 13c,d). This is why the contribution from the climatological latent heat flux over the negative pole is much smaller than that over the positive pole.
As in Fig. 10, but for the negative SASD.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
5. Conclusions
Using outputs from observational data and OGCM simulation results, the interannual SST variations in the South Atlantic are investigated. The new mechanism of the positive SASD demonstrated in this study is summarized schematically in Fig. 17. During the growth phase (Fig. 17a), the anomalous southward migration and strengthening of the subtropical high in late spring cause the positive (negative) latent heat flux anomaly over the positive (negative) SST anomaly pole. This leads to the anomalous shoaling (deepening) of the mixed layer in early summer. As a result, the warming of the mixed layer by the contribution from the climatological shortwave radiation is enhanced (suppressed) by the thinner (thicker) mixed layer and the positive (negative) SST anomaly pole develops. Thus, the latent heat flux anomaly contributes to the growth of both poles through its influence on the mixed layer depth. This is in contrast to previous studies on the SASD, which suggested that the latent heat flux anomaly directly generates the SST anomaly. This important influence by the interannual mixed layer depth anomaly on the contribution from the climatological shortwave radiation is also discussed to explain the evolution of the IOSD (Morioka et al. 2010).
Schematic of (a) growth and (b) decay mechanisms of the positive and negative SST anomaly poles for the positive SASD. Red (blue) color indicates the positive (negative) mixed layer temperature anomaly.
Citation: Journal of Climate 24, 21; 10.1175/2011JCLI4010.1
On the other hand, during the decay phase (Fig. 17b) the anomalous contributions from the net surface heat flux and entrainment damp the above SST anomalies. In early autumn, the cooling of the mixed layer by the entrainment is enhanced (suppressed) by the anomalously large (small) temperature difference between the mixed layer and the entrained water. In particular, the positive (negative) temperature anomaly in the mixed layer is responsible for this anomalous temperature difference. In addition, the cooling of the mixed layer by the contribution from the climatogical latent heat flux is enhanced (suppressed) by the thinner (thicker) mixed layer in late autumn. The almost similar mechanism is obtained for the growth and decay of the negative SASD.
This study reveals the important roles of interannual variation of the mixed layer depth on the formation as well as the decay of the SASD. The above results are based on the outputs from the OGCM, however, in which the surface heat flux is calculated by the bulk formula using the atmospheric reanalysis data and the simulated SST. Further studies using an ocean–atmosphere coupled model are necessary to investigate the air–sea interaction processes involving the SASD in more detail.
Acknowledgments
The authors thank Dr. Yukio Masumoto for his helpful comments. They also thank two anonymous reviewers for their helpful comments. The OGCM was run on the HITACHI SR11000/J1 of the Information Technology Center at the University of Tokyo as part of cooperative research with the Atmosphere and Ocean Research Institute of the University of Tokyo. This research is supported by the Japan Science and Technology Agency/Japan International Cooperation Agency through the Science and Technology Research Partnership for Sustainable Development (SATREPS). Also, the first author is supported by both the Sasakawa Scientific Research Grant from the Japan Science Society and the Research Fellowship of the Japan Society for the Promotion of Science.
REFERENCES
Behera, S. K., and T. Yamagata, 2001: Subtropical SST dipole events in the southern Indian Ocean. Geophys. Res. Lett., 28, 327–330.
Behera, S. K., P. S. Salvekar, and T. Yamagata, 2000: Simulation of interannual SST variability in the tropical Indian Ocean. J. Climate, 13, 3487–3499.
Cook, K. H., 2000: The South Indian convergence zone and interannual rainfall variability over southern Africa. J. Climate, 13, 3789–3804.
Davis, R. E., R. de Szoeke, and P. Niiler, 1981: Variability in the upper ocean during MILE. Part II: Modeling the mixed layer response. Deep-Sea Res., 28, 1453–1475.
Fauchereau, N., S. Trzaska, Y. Richard, P. Roucou, and P. Camberlin, 2003: Sea-surface temperature co-variability in the southern Atlantic and Indian Oceans and its connections with the atmospheric circulation in the Southern Hemisphere. Int. J. Climatol., 23, 663–677.
Hermes, J. C., and C. J. C. Reason, 2005: Ocean model diagnosis of interannual coevolving SST variability in the South Indian and South Atlantic Oceans. J. Climate, 18, 2864–2882.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.
Kraus, E. B., and J. S. Turner, 1967: A one-dimensional model of the seasonal thermocline: II. The general theory and its consequences. Tellus, 19, 98–106.
Legates, D. R., and C. J. Willmott, 1990: Mean seasonal and spatial variability in gauge-corrected, global precipitation. Int. J. Climatol., 10, 111–127.
Levitus, S., and T. P. Boyer, 1994: Temperature. Vol. 4, World Ocean Atlas 1994, NOAA Atlas NESDIS 4, 117 pp.
Levitus, S., R. Burgett, and T. P. Boyer, 1994: Salinity. Vol. 3, World Ocean Atlas 1994, NOAA Atlas NESDIS 3, 99 pp.
Lindesay, J. A., 1988: South African rainfall, the Southern Oscillation, and a Southern Hemisphere semi-annual cycle. J. Climatol., 8, 17–30.
Mason, S. J., 1995: Sea-surface temperature—South African rainfall associations, 1910–89. Int. J. Climatol., 15, 119–135.
Moisan, J. R., and P. P. Niiler, 1998: The seasonal heat budget of the North Pacific: Net heat flux and heat storage rates (1950–90). J. Phys. Oceanogr., 28, 401–421.
Morioka, Y., T. Tozuka, and T. Yamagata, 2010: Climate variability in the southern Indian Ocean as revealed by self-organizing maps. Climate Dyn., 35, 1059–1072.
North, G. R., T. L. Bell, R. F. Cahalan, and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110, 699–706.
Pacanowski, R. C., and S. G. H. Philander, 1981: Parameterization of vertical mixing in numerical models of tropical oceans. J. Phys. Oceanogr., 11, 1443–1451.
Pacanowski, R. C., and S. M. Griffies, 1999: MOM 3.0 manual. NOAA/GFDL, 680 pp.
Paulson, C. A., and J. J. Simpson, 1977: Irradiance measurements in the upper ocean. J. Phys. Oceanogr., 7, 952–956.
Qiu, B., and K. A. Kelly, 1993: Upper-ocean heat balance in the Kuroshio Extension region. J. Phys. Oceanogr., 23, 2027–2041.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analysis of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.
Reason, C. J. C., 1998: Warm and cold events in the southeast Atlantic/southwest Indian Ocean region and potential impacts on circulation and rainfall over southern Africa. Meteor. Atmos. Phys., 69, 49–65.
Reason, C. J. C., 2001: Subtropical Indian Ocean SST dipole events and southern African rainfall. Geophys. Res. Lett., 28, 2225–2227.
Reason, C. J. C., and H. Mulenga, 1999: Relationships between South African rainfall and SST anomalies in the southwest Indian Ocean. Int. J. Climatol., 19, 1651–1673.
Richard, Y., S. Trzaska, P. Roucou, and M. Rouault, 2000: Modification of the southern African rainfall variability/ENSO relationship since the late 1960s. Climate Dyn., 16, 883–895.
Smagorinsky, J., 1963: General circulation experiments with the primitive equations. Part I: The basic experiment. Mon. Wea. Rev., 91, 99–164.
Sterl, A., and W. Hazeleger, 2003: Coupled variability and air–sea interaction in the South Atlantic Ocean. Climate Dyn., 21, 559–571.
Suzuki, R., S. K. Behera, S. Iizuka, and T. Yamagata, 2004: Indian Ocean subtropical dipole simulated using a coupled general circulation model. J. Geophys. Res., 109, C09001, doi:10.1029/2003JC001974.
Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 2961–3012.
Venegas, S. A., L. A. Mysak, and D. N. Straub, 1996: Evidence for interannual and interdecadal climate variability in the South Atlantic. Geophys. Res. Lett., 23, 2673–2676.
Venegas, S. A., L. A. Mysak, and D. N. Straub, 1997: Atmosphere–ocean coupled variability in the South Atlantic. J. Climate, 10, 2904–2920.
Walker, N. D., 1990: Links between South African summer rainfall and temperature variability of the Agulhas and Benguela current systems. J. Geophys. Res., 95, 3297–3319.
Yasuda, I., T. Tozuka, M. Noto, and S. Kouketsu, 2000: Heat balance and regime shifts of the mixed layer in the Kuroshio Extension. Prog. Oceanogr., 47, 257–278.
