Abstract

The impact of El Niño–Southern Oscillation (ENSO) on the South Atlantic subtropical dipole mode (SASD) is investigated using both observations and model simulations. The SASD is the dominant mode of coupled ocean–atmosphere variability in the South Atlantic. This study focuses on austral summer, when both ENSO and SASD peak. It is shown that negative SASD events are associated with central Pacific El Niño events by triggering the Pacific–South American wave train (PSA). The latter resembles the third leading mode of atmospheric variability in the Southern Hemisphere (PSA2) and causes a weakening and meridional shift of the South Atlantic subtropical high, which then generates the negative SASD events. On the other hand, a strengthening of the South Atlantic subtropical high related to central La Niña teleconnections causes positive SASD events. The results herein show that the PSA2, triggered by central Pacific ENSO events, connects the tropical Pacific to the Atlantic. This connection is absent from eastern Pacific ENSO events, which appear to initiate the second leading mode of atmospheric variability in the Southern Hemisphere (PSA1). It is for this reason that previous studies have found weak correlations between ENSO and SASD. These findings can improve the climate prediction of southeastern South America and southern Africa since these regions are affected by sea surface temperature anomalies of both the Pacific and Atlantic Oceans.

1. Introduction

The South Atlantic subtropical dipole mode (SASD) is the dominant mode of coupled ocean–atmosphere variability in the South Atlantic and is characterized by a dipole-like pattern of sea surface temperature (SST) anomalies oriented northeast–southwest and a monopole structure in sea level pressure (SLP) (Venegas et al. 1997). By convention, the positive phase consists of positive SLP anomalies centered at 45°S, 15°W, with negative SST anomalies in the northern pole and positive anomalies in the southern pole. The positive phase of the SASD is induced by the strengthening and poleward shift of South Atlantic subtropical high (SASH), which in turn strengthens (weakens) the southeasterlies (westerlies) over the SST northern (southern) pole, increasing (decreasing) the upper-ocean mixing and surface evaporation, and leading to cooling (warming) (Sterl and Hazeleger 2003; Fauchereau et al. 2003; Haarsma et al. 2005; Morioka et al. 2011). The opposite situation creates the negative phase. The modulation in warming of the surface mixed layer by the shortwave radiation is also important for the onset and decay of the SASD (Morioka et al. 2011). The changes in the atmospheric circulation over the South Atlantic that lead to the SASD start in late austral winter and are well established by austral spring. Because of the delay in ocean response to the atmosphere, the SASD peaks in austral summer (Sterl and Hazeleger 2003).

Understanding the mechanisms that generate the SASD is important on account of their impact on precipitation over southern Africa and South America. While studies have shown that precipitation over southern Africa is strongly linked to South Atlantic SST anomalies with a dipole structure (Vigaud et al. 2009; Morioka et al. 2011), the role of SASD with regard to precipitation over South America is less evident (Robertson and Mechoso 2000; Doyle and Barros 2002; Carvalho et al. 2004; De Almeida et al. 2007; Muza et al. 2009). However, Bombardi and Carvalho (2011) and Bombardi et al. (2014) have recently shown that the SASD affects precipitation over southeastern South America, in particular the South Atlantic convergence zone, which is known to be responsible for bringing heavy rainfall to the southeast of Brazil in austral summer.

Many studies have examined the mechanism by which the SASD SST anomalies are created by changes in the atmospheric circulation over the South Atlantic, but few have attempted to explain the causes of the atmospheric changes that trigger the SASD events. Some of these studies focuses on the effect of the Antarctic Oscillation (AAO) on the SASD (Hermes and Reason 2005; Morioka et al. 2014). Others have found a weak link between the SASD and ENSO (Venegas et al. 1997; Fauchereau et al. 2003; Hermes and Reason 2005). However, a close examination of the SASD shows that almost all events occur in ENSO years (see Table 1 and Fig. 1). For the period of 1950–2010, 14 out of 19 events occur simultaneously with ENSO. This suggests a missing mechanism, which may relate to the fact that the impact of the central Pacific has not been explored by the aforementioned studies. As we shall see, most of the SASD events are associated with central Pacific ENSO.

Table 1.

SASD events where negative and positive SASD events are defined as the years when the SASD index is equal or greater than ±0.8°C during austral summer (Morioka et al. 2011). The SASD index is given by SST averaged within 30°–40°S, 10°–30°W minus those averaged within 15°–25°S, 0°–20°W (see boxes in Fig. 2a). Highlighted in boldface are the negative events that occur in El Niño years and positive events that occur in La Niña years. (ENSO events are defined in Table 2.)

SASD events where negative and positive SASD events are defined as the years when the SASD index is equal or greater than ±0.8°C during austral summer (Morioka et al. 2011). The SASD index is given by SST averaged within 30°–40°S, 10°–30°W minus those averaged within 15°–25°S, 0°–20°W (see boxes in Fig. 2a). Highlighted in boldface are the negative events that occur in El Niño years and positive events that occur in La Niña years. (ENSO events are defined in Table 2.)
SASD events where negative and positive SASD events are defined as the years when the SASD index is equal or greater than ±0.8°C during austral summer (Morioka et al. 2011). The SASD index is given by SST averaged within 30°–40°S, 10°–30°W minus those averaged within 15°–25°S, 0°–20°W (see boxes in Fig. 2a). Highlighted in boldface are the negative events that occur in El Niño years and positive events that occur in La Niña years. (ENSO events are defined in Table 2.)
Fig. 1.

SASD index (°C) in DJF for the period of 1950–2010. The SASD index is SST anomalies averaged within 30°–40°S, 10°–30°W minus those averaged within 15°–25°S, 0°–20°W (see boxes in Fig. 2a). SASD events are considered those years when the SASD index is equal to or greater than ±0.8°C. Dark red (blue) lines represent negative (positive) SASD events that occur during El Niño (La Niña) years and light blue (red) lines negative (positive) SASD events that occur during neutral ENSO years (Table 1).

Fig. 1.

SASD index (°C) in DJF for the period of 1950–2010. The SASD index is SST anomalies averaged within 30°–40°S, 10°–30°W minus those averaged within 15°–25°S, 0°–20°W (see boxes in Fig. 2a). SASD events are considered those years when the SASD index is equal to or greater than ±0.8°C. Dark red (blue) lines represent negative (positive) SASD events that occur during El Niño (La Niña) years and light blue (red) lines negative (positive) SASD events that occur during neutral ENSO years (Table 1).

The importance of central Pacific ENSO in triggering teleconnection to the Southern Hemisphere has been the focus of several recent studies (Ding et al. 2011; Schneider et al. 2012; Ciasto et al. 2014; Wilson et al. 2014). For instance, Ding et al. (2011) show that central Pacific ENSO SST anomalies generate the strongest teleconnection in the Southern Hemisphere by triggering a stationary Rossby wave train. They are responsible for the greatest influence on the West Antarctic climate during austral winter and spring. However, Ciasto et al. (2014) found a weak link between central Pacific ENSO and atmospheric circulation in the Southern Hemisphere for austral summer. Nonetheless, we do believe that a strong connection to the South Atlantic exists in austral spring and is capable of forcing the SASD in austral summer.

Therefore, the objective of this study is to investigate possible causes of atmospheric circulation changes related to ENSO that ultimately generate SASD anomalies in austral summer. We find that SST anomalies in the tropical Pacific linked to ENSO are responsible for the teleconnections that trigger the SASD in the Atlantic. Moreover, we perform idealized model experiments to determine which heating/cooling location of ENSO is more efficient to create the teleconnection pattern that leads to the SASD.

2. Methodology

The SST data used in this study is from the monthly Extended Reconstructed SST dataset from 1950 to 2010 (Smith et al. 2008). Monthly mean atmospheric fields are taken from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis for the period 1950–2010 (Kalnay et al. 1996). For all datasets, monthly anomalies are estimated by subtracting the monthly climatological mean after removing a linear trend obtained from a least squares fit. Even though there are well-known problems with atmospheric reanalysis prior to the satellite era (1979), we use data from 1950 to 2010 for two main reasons: 1) to be comparable with previous literature (Venegas et al. 1997; Sterl and Hazeleger 2003; Morioka et al. 2011) and 2) to increase the number of events used for the composite analysis. Nevertheless, we have performed the same analysis for the period of 1979–2010 (not shown) and the results do not differ greatly from those using the longer period.

To investigate whether Pacific SST anomalies predetermine the SASD in the Atlantic, we perform a set of ensemble runs using the Simplified Parameterizations, Primitive Equation Dynamics (SPEEDY) model fully coupled to the Miami Isopycnal Coordinate Ocean Model (MICOM) in the Atlantic basin from 45°S to 60°N. SPEEDY has a vertical resolution of seven layers and a horizontal resolution of T30 spectral truncation (Molteni 2003). MICOM has 19 isopycnal layers and a horizontal resolution of 1°. [For more details, see Haarsma et al. (2005).] The SST anomalies are prescribed in the Pacific for 2 years (from January of the ENSO/SASD onset year until December of the decay year) and the ocean model in the Atlantic is allowed to respond to the atmospheric circulation caused by the Pacific teleconnection.

Three sets of ensemble runs are performed: 1) the CONTROL run with prescribed climatological SST in the Pacific, 2) the NEG run with prescribed SST anomalies in the Pacific from the composite of negative SASD events (Fig. 2a), and 3) the POS run with prescribed SST anomalies in the Pacific from the composite of positive SASD events (Fig. 3a). Note that these experiments are forced with SST anomalies in the Pacific from 20°S to 20°N; we have chosen to name the NEG and POS experiments because we extract the Pacific SST anomalies from the data composites that lead to negative and positive SASD in the Atlantic, respectively. Each ensemble consists of 20 model integrations that are initialized with slightly different atmospheric conditions to represent internal atmospheric variability and are integrated for 2 years (from January of the ENSO/SASD onset year until December of the decay year). The atmospheric initial conditions were extracted randomly from a SPEEDY control run in which SST climatology was used everywhere. The MICOM ocean initial state is obtained from a 100-yr climatological run. The modeled anomalies of the variables presented in the results section for the NEG and POS experiments are obtained by subtracting the ensemble mean of the CONTROL run variables from the ensemble mean of the NEG and POS run variables.

Fig. 2.

Composites of SST anomalies (°C) in DJF for (a) negative SASD events, (b) El Niño years without negative SASD, and (c) the difference between (a) and (b). (d)–(f) As in (a)–(c), but for sea level pressure (shading, hPa) and wind anomalies at 850 hPa (vectors, m s−1). There are 11 negative SASD and 11 El Niño events without negative SASD (Tables 1 and 2, respectively). Solid lines in (c) and (f) encompass areas where the difference between the means of the anomalies are statistically significant at the 95% confidence level given by a standard two-tailed t test. Boxes in (a) indicate the areas where SST is averaged to calculate the SASD index and those in (b) the areas where the artificial warming/cooling anomalies are imposed for the idealized experiments WP, CW, CE, and EP from west to east.

Fig. 2.

Composites of SST anomalies (°C) in DJF for (a) negative SASD events, (b) El Niño years without negative SASD, and (c) the difference between (a) and (b). (d)–(f) As in (a)–(c), but for sea level pressure (shading, hPa) and wind anomalies at 850 hPa (vectors, m s−1). There are 11 negative SASD and 11 El Niño events without negative SASD (Tables 1 and 2, respectively). Solid lines in (c) and (f) encompass areas where the difference between the means of the anomalies are statistically significant at the 95% confidence level given by a standard two-tailed t test. Boxes in (a) indicate the areas where SST is averaged to calculate the SASD index and those in (b) the areas where the artificial warming/cooling anomalies are imposed for the idealized experiments WP, CW, CE, and EP from west to east.

Fig. 3.

As in Fig. 2, but for positive SASD events and La Niña years without positive SASD. There are eight positive SASD events and 16 La Niña years without positive SASD (Tables 1 and 2, respectively).

Fig. 3.

As in Fig. 2, but for positive SASD events and La Niña years without positive SASD. There are eight positive SASD events and 16 La Niña years without positive SASD (Tables 1 and 2, respectively).

To investigate why some ENSO events create the atmospheric circulation patterns that lead to the SASD, an additional set of experiments was performed by forcing the coupled model with idealized warming/cooling anomalies imposed on different regions across the Pacific between 10°N–10°S and, respectively, the eastern Pacific, 120°–80°W (EP); central-eastern Pacific, 160°–120°W (CE); central-western Pacific, 160°E–160°W (CW); central-eastern and central western Pacific combined, 160°E–120°W (CEW); and western Pacific, 120°–160°E (WP). The SST anomalies are linearly damped from 1° to 0°C over a 10° latitude–longitude band and climatological monthly mean SST is prescribed elsewhere. Note that the imposition of these SST anomalies is greatly idealized and does not reflect the approximate position of anomalies associated with the different types of El Niño events. Instead the locations merely span the zonal extent of the equatorial Pacific and provide a way of comparing the atmospheric circulation resulting from the warm/cold SST anomalies in each region. The magnitudes of the anomalies are also idealized, as typically the eastern Pacific El Niño events are generally warmer than the central Pacific events. Nevertheless, this set of experiments can offer a better understanding of the sensitivity of the atmospheric circulation to the geographic location of forcing along the tropical Pacific.

3. Results

Following Morioka et al. (2011), we define the SASD index as SST anomalies averaged within 30°–40°S, 10°–30°W minus those averaged within 15°–25°S, 0°–20°W. This index shows a high correlation of 0.84 with the first principal component time series of observed SST anomalies in the South Atlantic. Negative and positive events are then defined as the years when the SASD index is equal or greater than ±0.8°C during austral summer [December–February (DJF)]. This leads to 11 negative events (Fig. 1), of which 10 occurred during El Niño years (Table 1). On the other hand, of 8 positive SASD events, 4 events occur during La Niña years (see Fig. 1 and Table 1). From a total of 19 SASD events, 14 occur during ENSO1 years, suggesting a possible link between ENSO and SASD that is further investigated by comparing composites of SST, SLP, and wind at 850 hPa for SASD events with those for ENSO events without SASD (Figs. 2 and 3).

The SST composites for negative SASD events (Figs. 2a) show that the SASD appears to be related to El Niño events with the strongest SST anomalies located toward the central Pacific (Niño-3.4 region). However, these are not necessarily El Niño Modoki events defined according to the Ashok et al. (2007) criteria. It is for this reason that we do not use the term “Modoki” for the central Pacific events associated with SASD. Rodrigues et al. (2011) also found that central Pacific El Niño events are associated with SST anomalies in the South Atlantic that resemble the SASD pattern. On the other hand, El Niño events that are not associated with SASD (Table 2) resemble the canonical El Niño events, which are events with the strongest SST anomalies in the eastern Pacific (Figs. 2b). The statistically significant differences in the SST anomalies between El Niño events with and without SASD appear over the tropical and South Atlantic, as well as in the equatorial Pacific (Figs. 2c). El Niño events linked to SASD are considerable weaker than those without SASD (represented by negative values in Fig. 2c).

Table 2.

ENSO years in which SASD events do not occur. ENSO events are defined using the oceanic Niño index (ONI), which is the 3-month running mean SST anomaly for the Niño-3.4 region (i.e., 5°N–5°S, 120°–170°W). El Niño events are defined as five consecutive overlapping 3-month periods at or above the +0.5°C anomaly and La Niña events at or below the −0.5°C anomaly. More details are available online (at http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml). Highlighted in boldface are eastern Pacific ENSO events and in italics central Pacific ENSO, according to Yeh et al. (2009) and Taschetto et al. (2014) for El Niño and Tedeschi et al. (2013) for La Niña events. ENSO years that were not classified in the literature as either eastern or central Pacific are not highlighted.

ENSO years in which SASD events do not occur. ENSO events are defined using the oceanic Niño index (ONI), which is the 3-month running mean SST anomaly for the Niño-3.4 region (i.e., 5°N–5°S, 120°–170°W). El Niño events are defined as five consecutive overlapping 3-month periods at or above the +0.5°C anomaly and La Niña events at or below the −0.5°C anomaly. More details are available online (at http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml). Highlighted in boldface are eastern Pacific ENSO events and in italics central Pacific ENSO, according to Yeh et al. (2009) and Taschetto et al. (2014) for El Niño and Tedeschi et al. (2013) for La Niña events. ENSO years that were not classified in the literature as either eastern or central Pacific are not highlighted.
ENSO years in which SASD events do not occur. ENSO events are defined using the oceanic Niño index (ONI), which is the 3-month running mean SST anomaly for the Niño-3.4 region (i.e., 5°N–5°S, 120°–170°W). El Niño events are defined as five consecutive overlapping 3-month periods at or above the +0.5°C anomaly and La Niña events at or below the −0.5°C anomaly. More details are available online (at http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml). Highlighted in boldface are eastern Pacific ENSO events and in italics central Pacific ENSO, according to Yeh et al. (2009) and Taschetto et al. (2014) for El Niño and Tedeschi et al. (2013) for La Niña events. ENSO years that were not classified in the literature as either eastern or central Pacific are not highlighted.

For negative SASD, the summer SLP patterns over the South Atlantic are similar to the first leading mode of variability in the South Atlantic with a SLP monopole centered at 45°S, 15°W (Fig. 2d). A meridional shift and weakening of the subtropical high for negative SASD weaken the trades generating the warm anomalies in the northern pole and intensify the westerlies, leading to cold anomalies in the southern pole (Fig. 2d). On the other hand, El Niño events without SASD present weaker SLP anomalies of opposite sign to those events with SASD (Figs. 2e). The wind anomalies are also weaker over the South Atlantic for those El Niño events without SASD. These differences are statistically significant and are shown in Fig. 2f.

Positive SASD events are associated with cold anomalies in the central Pacific (Fig. 3a). Even though only four positive SASD events occur during La Niña years, the other four events also exhibit cold anomalies in the Pacific that resembled La Niña but were below the requisite threshold. The SST anomalies are also significantly stronger in the central Pacific as well as in the eastern Pacific for La Niña events without SASD (Figs. 3b,c). Positive SASD events present an SLP monopole over the South Atlantic (Fig. 3d). In this case, however, the meridional shift is accompanied by a strengthening of the subtropical high that leads to an increase in the trades and cold anomalies in the northern pole, and a decrease in the westerlies and warm anomalies in the southern pole (Fig. 3d). These features are not present during La Niña years without SASD (Fig. 3e) and the differences are statically significant (Fig. 3f).

In summary, while Fig. 1 and Table 1 both highlight possible connections between ENSO and SASD events, composite analyses of SST anomalies during negative (positive) SASD events that coincide with El Niño (La Niña) occurrences illustrate that these do not largely occur in relation to canonical ENSO events, that is, events in which the strongest SST anomalies are located in the eastern Pacific (Figs. 2 and 3). By contrast, ENSO events that appear to be related to the SASD are largely linked to weaker SST anomalies in the central Pacific and the reason for this is investigated in the following sections.

a. ENSO–South Atlantic teleconnection

The systematic association of anomalies in the tropical Pacific and those in the South Atlantic suggests that SASD anomalies may result from an ENSO teleconnection particularly with the central Pacific (Figs. 13). In the Southern Hemisphere, this teleconnection is accomplished through the Pacific–South American wave train (PSA), which is most active from September to December and corresponds to the second and third leading patterns of atmospheric variability in the Southern Hemisphere (Mo 2000; Vera et al. 2004, and references therein; Grimm et al. 2007). The phase locking between ENSO and SASD is important. Sterl and Hazeleger (2003) show that the SLP/wind anomalies start around 6 months before the peak of the SST anomalies (i.e., the SASD mature phase in DJF; see their Fig. 5), but are well established 3 months in advance from September to November (SON) corresponding to when the ENSO teleconnection is most active. For this reason, we analyze composites of geopotential height anomalies at 200 hPa in SON.

For SASD events (Figs. 4a,d), the geopotential height anomalies reveal the presence of a pattern resembling the third leading mode (PSA2), described by Mo (2000) as a wave pattern related to the high-frequency part of ENSO (Fig. 1c in Mo 2000). For negative SASD events, negative anomalies of geopotential height are present over the South Atlantic (Fig. 4a) and are associated with an anomalous low-pressure center at 45°S, 15°W (Fig. 2d). This is consistent with a weakening and meridional shift of the climatological subtropical high in summer. For positive SASD events (associated with Pacific cold anomalies), an area of positive geopotential height (Fig. 4d) coincides with positive SLP anomalies also centered at 45°S, 15°W (Fig. 3d). We also analyzed geopotential height anomalies during the peak season of the SASD (not shown) and noticed that central ENSO teleconnection in DJF is very similar to that in SON, albeit weaker.

Fig. 4.

Composites of geopotential height anomalies at 200 hPa (shading, m) and Rossby wave activity flux (vectors, m2 s−2) in SON for (a) negative SASD events, (b) El Niño years without negative SASD, and (c) the difference between (a) and (b). (d)–(f) As in (a)–(c), but for positive SASD events and La Niña years without positive SASD. Solid lines in (c) and (f) encompass areas where the differences between the means of the geopotential anomalies are statistically significant at the 95% confidence level. Scaling for arrows is given near the lower-right corner (m2 s−2).

Fig. 4.

Composites of geopotential height anomalies at 200 hPa (shading, m) and Rossby wave activity flux (vectors, m2 s−2) in SON for (a) negative SASD events, (b) El Niño years without negative SASD, and (c) the difference between (a) and (b). (d)–(f) As in (a)–(c), but for positive SASD events and La Niña years without positive SASD. Solid lines in (c) and (f) encompass areas where the differences between the means of the geopotential anomalies are statistically significant at the 95% confidence level. Scaling for arrows is given near the lower-right corner (m2 s−2).

ENSO events without SASD (Figs. 4b,e), which includes most of canonical ENSO (Table 2), present a pattern similar to the second leading mode of variability (PSA1) related to the low-frequency part of ENSO (Fig. 1b in Mo 2000). This PSA does not have a center over the South Atlantic and consequently does not generate SASD events. Figures 4c and 4f depict the statistically significant differences between SASD events and ENSO events without SASD and they are more pronounced in the extratropics poleward of 30°S.

Since the atmospheric response in the Southern Hemisphere is a quasi-stationary Rossby wave train arching toward South America (PSA), we have also computed Rossby wave activity flux according to Takaya and Nakamura (2001) formulation and superimposed them on the geopotential height anomalies to highlight the PSA (Fig. 4). This formulation is an extension of the zonally averaged Eliassen–Palm flux (Andrews and McIntyre 1976) and the wave activity flux shows the direction of Rossby wave ray paths since it is parallel to the local group velocity. The PSA2 related to SASD events extends from the tropical Pacific poleward to the South Pacific before turning north into the Atlantic, passing through the tip of South America (Figs. 4a,d).

In contrast, the PSA1 related to ENSO without SASD extends farther poleward in the South Pacific. Negative (positive) geopotential height anomalies for El Niño (La Niña) events are over the Weddell Sea in the Southern Ocean, instead of over the South Atlantic (Figs. 4b,e). For El Niño events, the wave train splits into two, with the northern branch crossing over southern South America and reaching the South Atlantic farther north (Fig. 4b). For La Niña events, the split is not noticeable and the wave train misses the South Atlantic, reaching the southern tip of Africa (Fig. 4e). In any case, PSA1 is not capable of creating the atmospheric anomalies over the South Atlantic that generate the SASD.

Figure 4 thus suggests that central Pacific ENSO events trigger the PSA2, connecting the tropical Pacific to the Atlantic. This connection is absent from canonical ENSO events, which appear to initiate the PSA1 (the high-latitude response to ENSO). These links are further identifiable in Fig. 5a, which shows the PSA2 obtained from empirical orthogonal function (EOF) analysis for geopotential height anomalies at 500 hPa for the Southern Hemisphere from 1950 to 2010 (shading) with the composite of geopotential height anomalies at 200 hPa for El Niño events with SASD (contours). The latter is somehow displaced to the west over the Pacific with regards to the former, which might explain the weak spatial correlation between them of 0.4. However, the centers of negative geopotential height anomalies coincide over the South Atlantic and Indian Oceans. The PSA1 obtained from the EOF analysis is very similar to the composite for El Niño events without SASD, in particular over the South Pacific and South Atlantic, with a spatial correlation of 0.6. These PSA patterns are similar to those obtained by Ciasto et al. (2014) from June to September (see their Fig. 1).

Fig. 5.

(a) PSA2 and (b) PSA1 obtained from EOF analysis of geopotential height anomalies at 500 hPa for the Southern Hemisphere from 1950 to 2010 (shading, nondimensional units). Superimposed are contours of geopotential height anomalies at 200 hPa in SON for (a) negative SASD events and (b) El Niño years without negative SASD. Contour interval is every 10 m. Zero contours are omitted and solid (dashed) lines represent positive (negative) values.

Fig. 5.

(a) PSA2 and (b) PSA1 obtained from EOF analysis of geopotential height anomalies at 500 hPa for the Southern Hemisphere from 1950 to 2010 (shading, nondimensional units). Superimposed are contours of geopotential height anomalies at 200 hPa in SON for (a) negative SASD events and (b) El Niño years without negative SASD. Contour interval is every 10 m. Zero contours are omitted and solid (dashed) lines represent positive (negative) values.

Even though both PSA modes depict wave-3 patterns in quadrature with each other, the wavelength of the PSA1 associated with ENSO without SASD is longer near the source region (over the Pacific) than that of the PSA2 related to the SASD. Hoskins and Karoly (1981) show that longer wavelengths propagate farther poleward while shorter wavelengths appear to be trapped equatorward. Also, according to the basic Rossby wave theory, zonal elongation of the forcing (ENSO without SASD) enhances the meridional propagation (Ambrizzi and Hoskins 1997). This is consistent with the findings described here (Figs. 4 and 5). Thus, only the PSA2 triggered by SST anomalies in the central Pacific appears to connect the Pacific to the South Atlantic in order for SASD events to be initiated. The sensitivity to the location of the SST anomalies will now be investigated further with idealized model experiments.

b. Sensitivity of ENSO teleconnection to the location of heating/cooling source

To verify that the Pacific SST anomalies generate the changes in the atmospheric circulation over the South Atlantic that can trigger the SASD and check the robustness of the aforementioned results, we first examine the results from SPEEDY-MICOM runs in which the Atlantic basin is allowed to respond to changes in the atmospheric circulation caused by Pacific SST anomalies. Forced with the observed Pacific warm anomalies associated with the negative SASD events (NEG), the model generally reproduces the PSA teleconnection and the SASD well (Figs. 6a and 7a,g, respectively). The spatial correlation between the geopotential height anomalies from the observations and those from the NEG run is 0.55 (Table 3). The wave flux shows a PSA emanating from the central Pacific arching into the South Atlantic where negative geopotential anomalies are found, similar to the observations (Fig. 4a). However, the SLP and SST anomalies are somewhat weaker in magnitude compared to observations (cf. Figs. 2a,d). The negative SLP anomalies are displaced to the southeast and as a result the cooling in the southern pole is also shifted to the east. Nonetheless, the spatial correlations with regard to the observed SLP and SST are 0.63 and 0.66, respectively (Table 3).

Fig. 6.

Composites of geopotential height anomalies at 200 hPa (shading, m) and Rossby wave activity flux (vectors, m2 s−2) for El Niño/negative SASD experiments in SON: (a) NEG SASD, (b) EP, (c) CE, (d) CW, (e) CEW, and (f) WP. (g)–(l) As in (a)–(f), but for La Niña/positive SASD experiments.

Fig. 6.

Composites of geopotential height anomalies at 200 hPa (shading, m) and Rossby wave activity flux (vectors, m2 s−2) for El Niño/negative SASD experiments in SON: (a) NEG SASD, (b) EP, (c) CE, (d) CW, (e) CEW, and (f) WP. (g)–(l) As in (a)–(f), but for La Niña/positive SASD experiments.

Fig. 7.

Anomalies of SST (°C) from SPEEDY-MICOM for El Niño/negative SASD experiments in DJF: (a) NEG SASD, (b) EP, (c) CE, (d) CW, CEW, and (f) WP. (g)–(l) As in (a)–(f), but for SLP (shading, hPa) and wind at 850 hPa (vectors, m s−1). Solid lines in (a) and (g) encompass the regions where the differences between the NEG SASD and the climatological runs are statistically significant at the 95% confidence level. Solid lines in the others panels contain the regions where the differences between the idealized runs and the NEG SASD run are statistically significant at the 95% confidence level. Boxes in (a)–(f) indicate the areas of the SST poles and the stars in (g)–(l) the center of the SLP monopole, associated with the observed SASD.

Fig. 7.

Anomalies of SST (°C) from SPEEDY-MICOM for El Niño/negative SASD experiments in DJF: (a) NEG SASD, (b) EP, (c) CE, (d) CW, CEW, and (f) WP. (g)–(l) As in (a)–(f), but for SLP (shading, hPa) and wind at 850 hPa (vectors, m s−1). Solid lines in (a) and (g) encompass the regions where the differences between the NEG SASD and the climatological runs are statistically significant at the 95% confidence level. Solid lines in the others panels contain the regions where the differences between the idealized runs and the NEG SASD run are statistically significant at the 95% confidence level. Boxes in (a)–(f) indicate the areas of the SST poles and the stars in (g)–(l) the center of the SLP monopole, associated with the observed SASD.

Table 3.

Spatial correlation coefficients: geopotential height at 200 hPa (Z200), SLP, and SST fields extracted from the NEG and POS SASD runs are compared to those from the observations (OBS) and from their respective idealized experiments (EP, CE, CW, CEW, and WP). Highlighted in boldface are the highest coefficients.

Spatial correlation coefficients: geopotential height at 200 hPa (Z200), SLP, and SST fields extracted from the NEG and POS SASD runs are compared to those from the observations (OBS) and from their respective idealized experiments (EP, CE, CW, CEW, and WP). Highlighted in boldface are the highest coefficients.
Spatial correlation coefficients: geopotential height at 200 hPa (Z200), SLP, and SST fields extracted from the NEG and POS SASD runs are compared to those from the observations (OBS) and from their respective idealized experiments (EP, CE, CW, CEW, and WP). Highlighted in boldface are the highest coefficients.

For positive SASD events, the model also simulates the PSA teleconnections, the strengthening of the SASH, the associated cooling in the northern pole, and the warming in the southern pole (Figs. 6g, and 8a,g, respectively). The positive SLP anomalies are also displaced southward in the model but to the west, and as a consequence the warming in the southern pole is confined to the west of 10°W. In the observations, the SLP monopole expands to 20°E and the SST southern pole to 0°. The spatial correlation coefficients are 0.50, 0.45, and 0.34 for geopotential height, SLP, and SST, respectively (Table 3).

Fig. 8.

As in Fig. 7, but for La Niña/positive SASD experiments.

Fig. 8.

As in Fig. 7, but for La Niña/positive SASD experiments.

The differences in the southern pole between the model and observations may result from the proximity to the sponge layer in the model southern boundary. Morioka et al. (2012) also show that simulated SLP anomalies associated with the SASD in their coupled model are shifted to the south by 10°. However, given the general ability to replicate the primary features of SASD events, we have confidence in the model, and thus run further experiments to determine if particular regions of the Pacific are most vital in triggering teleconnections to the Atlantic via the PSA2.

To investigate why only some El Niño events are associated with negative SASD (Fig. 1 and Table 1), we analyze the response of the geopotential height at 200 hPa and Rossby wave activity flux to the varying locations of a 1°C SST perturbation along the equatorial Pacific (Figs. 6b–f). The solid gray lines indicate areas where the differences between the NEG run and the idealized run in question are statistically significant. The idea behind this analysis is that the more widespread the differences, the more poorly the idealized run simulates the SASD. Moreover, we have not shown the differences with regard to observations because we consider the NEG run the closest to reality the model can simulate. Spatial correlations between the NEG run and idealized experiments were also computed and are displayed in Table 3.

The geopotential height anomalies for the EP run (Fig. 6b) do not resemble those from the NEG run (Fig. 6a). The center of negative anomalies is displaced southeastward over the South Atlantic in comparison to the NEG run. In fact, the EP geopotential height anomalies and wave activity flux are similar to those for the El Niño events without SASD (Fig. 4b), which include most of the canonical El Niño events with heating in the eastern Pacific (Table 2). They are more elongated in the northeast and southeast direction with one center of positive anomalies flanked by two centers of negative anomalies. However, the centers are displaced slightly to the east in comparison to the observations, with the PSA reaching southern Africa. Similar to the observations, the SASD does not evolve in the South Atlantic as consequence of this teleconnection (Fig. 7b). The modeled SLP anomalies present the opposite sign to the observed El Niño events without SASD (Fig. 2e). This might be related to the coarse horizontal and vertical resolution of the atmospheric model.

In contrast to the EP run, the CE, CW, CEW, and WP experiments present more centers of negative and positive geopotential height anomalies in the Pacific–Atlantic sector (Figs. 6c–f) similar to El Niño events with SASD (Fig. 4a). CE teleconnection is a transition between the EP and CW runs. Judging by the statistically significant differences, CW and CEW better simulate the location of the centers of negative geopotential height over the South Atlantic (Figs. 6d,e). However, the CEW heating anomaly seems to have triggered two distinct wave trains in the Pacific. The CEW presents the highest spatial correlation compared to the NEG run (Table 3). The negative SLP anomalies over the South Atlantic for the CW and CEW (Figs. 7j–l) are also displaced to the southeast as are those in the NEG run (Fig. 7g), but this is weaker for the CEW run. As a consequence, the CW run better simulates the SST anomalies when compared to the NEG run and observed SASD (Figs. 7d, 7a, and 2a, respectively). SST and SLP patterns from the CW run present the largest spatial correlation with those from the NEG run (Table 3). In fact, the CW run is the only one that simulates the warming of the northern pole of the SASD. Thus, we conclude that the heating position in the central western Pacific (CW) seems to be important in generating a wave pattern similar to that from the observations (Fig. 4a) that leads to the low pressure center over the South Atlantic (Fig. 2d) and the correct SASD pattern (Fig. 2a), at least in the current version of the SPEEDY-MICOM coupled model.

Similar results are found for the idealized La Niña runs with cooling in the Pacific. The EP teleconnection (Fig. 6h) does not resemble that of the POS run but is similar to that generated by observed La Niña events without SASD (Figs. 6e and 4e, respectively). The PSA also reaches southern Africa, but positive geopotential height anomalies are over the South Atlantic between 30° and 60°S, and not over the Weddell Sea as in the observations. As a consequence, simulated SLP anomalies are correct in the tropical Atlantic, but have an opposite signal to the observations poleward of 30°S. The EP run erroneously generates a strengthening and meridional shift of the subtropical high, albeit weaker than that of the POS run and observations (Figs. 8h, 8g, and 3a, respectively), that leads to the cooling of the northern pole (Fig. 8b).

The PSA teleconnections are similar among the CE, CW, CEW, and WP runs, with stronger geopotential height anomalies for CEW and WP (Figs. 6i–l). Comparing them all to both POS run (Fig. 6g) and observations (Fig. 4d), the CW and CEW runs more realistically simulate the teleconnection (Figs. 6j,k) that leads to the SLP and wind anomalies over the South Atlantic (Figs. 8j,k). This is also corroborated by the spatial correlation analysis (Table 3). Both the CW and CEW runs simulate the cooling of the northern pole (Figs. 8d,e), although only CW simulates the warming of the southern pole. In general, imposing a cooling in the CW region of the Pacific is more effective in creating the high pressure monopole over the South Atlantic that leads to the positive SASD (Figs. 8d,j).

Therefore, the idealized runs and observations show that even though eastern Pacific ENSO events are generally stronger, central-western events are more efficient to generate the correct atmospheric response over the South Atlantic that leads to the SASD. The climatological SST state plays a crucial role and may explain why central Pacific ENSO events are more conducive to producing the strongest teleconnection in the Southern Hemisphere. SST exceeding 27.5°C gives rise to upper-level divergence perturbations that are capable of generating Rossby waves (Ciasto et al. 2014, and references therein). Similar analysis to theirs shows that much of this convective SST is observed in the central and western Pacific (Fig. 9a). Even though eastern Pacific ENSO SST anomalies are stronger, the fact that they occur in the eastern Pacific does not play a role on the Rossby wave excitation because convective SST (>27.5°C) is not achieved there. This can also explain why some eastern Pacific El Niños initiate SASD and some central Pacific El Niño events do not. For instance, the 1997/98 El Niño was the strongest on record, but the 1982/83 El Niño presented higher anomalies over the central western Pacific (Fig. 9b). Thus only the latter was capable of initiating a negative SASD in a similar fashion to the central Pacific El Niño events.

Fig. 9.

(a) Composites of total SST (°C) associated with El Niño years without negative SASD (green contours) and negative SASD events (brown contours) for SON. The contours are 27.5°, 28.5°, and 29.5°C, and the shading corresponds to the SST difference between El Niño years without negative SASD and negative SASD events. (b) As in (a), but for the 1997/98 minus 1982/83 El Niño years (green and brown contours, respectively).

Fig. 9.

(a) Composites of total SST (°C) associated with El Niño years without negative SASD (green contours) and negative SASD events (brown contours) for SON. The contours are 27.5°, 28.5°, and 29.5°C, and the shading corresponds to the SST difference between El Niño years without negative SASD and negative SASD events. (b) As in (a), but for the 1997/98 minus 1982/83 El Niño years (green and brown contours, respectively).

Note that we have not analyzed the effects of the AAO on the SASD, which Hermes and Reason (2005) and Morioka et al. (2014) found to be important for the development of SASD anomalies. Both studies investigate the initiation of subtropical dipoles in other oceans jointly with the South Atlantic. For instance, Morioka et al. (2014) have shown that prescribing tropical SST anomalies in a coupled model can simultaneously generate the subtropical dipoles of the Indian, South Atlantic, and South Pacific oceans. They also demonstrate that the AAO can trigger stationary Rossby waves that induce the SLP anomalies, which in turn lead to the subtropical dipoles, but only when the tropical SST anomalies are suppressed from their numerical experiments. This could explain some of the discrepancies between the observations and the coupled model results, although recent work by Ding et al. (2012) shows the AAO variability in austral summer is partially forced by ENSO and the PSA. Nevertheless, when forced solely by SST anomalies in the Pacific, the model generally reproduces the phase of the SASD, supporting our hypothesis that its initiation may be at least partially generated by Pacific SST anomalies.

4. Conclusions

In this study we have shown from observational and model results that central Pacific El Niño (La Niña) events cause, through a PSA teleconnection, a weakening (strengthening) and meridional shift of the South Atlantic subtropical high, which triggers the negative (positive) phase of the SASD. Previous studies have found weak correlations between SASD and Niño indices, since it is central Pacific El Niño and La Niña events that are most capable of inducing these atmospheric anomalies that generate the SASD, via the PSA2. Eastern Pacific ENSO events trigger the PSA1, which does not affect the atmospheric circulation over the South Atlantic in the way necessary to generate SASD events.

Observations show that the PSA1 associated with ENSO without SASD is longer near the source region (over the Pacific) and propagates farther poleward than the PSA2 related to ENSO with SASD. Also, zonal elongation of the forcing for the ENSO events without SASD enhances the meridional propagation. As a consequence of the different pathways, only central Pacific ENSO events connect the tropical Pacific to the South Atlantic. This is also corroborated by model results. Sensitivity experiments varying locations of a 1°C SST perturbation along the equatorial Pacific confirm that central-western events are more efficient in generating the correct atmospheric response over the South Atlantic that leads to the SASD because Rossby wave excitation is more sensitive to SST anomalies in the central Pacific, where convective SST (>27.5°C) is easily attained.

Understanding the link between SASD and ENSO could potentially improve our capability of predicting the SASD, particularly as ENSO-related SST anomalies evolve earlier than those in the Atlantic. This is corroborated by Yuan et al. (2014), who recently showed that SASD events can be predicted by their coupled model, partly as a result of the good skill in predicting the tropical Pacific. The results described here are also relevant to the climate of South America and southern Africa (Bombardi et al. 2014; Morioka et al. 2011). We recognize that there are some exceptions to the aforementioned rule that positive (negative) ENSO generates negative (positive) SASD and that we have not considered the effects of the AAO over the SASD (Hermes and Reason 2005; Morioka et al. 2014). However, specific differences regarding these exceptions can give us valuable clues about the mechanisms and is the objective of ongoing work.

Acknowledgments

We thank Dr. Brant Liebmann and the two anonymous reviewers for their valuable comments. RRR especially thanks CNPq for its support (Grant 477777/2011-2). EJDC thanks FAPESP (Grants 010/01943-8 and 2011/50552-4). This work is part of the research conducted by the INCT-MC, INCT-Mar COI, and Rede CLIMA.

REFERENCES

REFERENCES
Ambrizzi
,
T.
, and
B. J.
Hoskins
,
1997
:
Stationary Rossby-wave propagation in a baroclinic atmosphere
.
Quart. J. Roy. Meteor. Soc.
,
123
,
919
928
, doi:.
Andrews
,
D. G.
, and
M. E.
McIntyre
,
1976
:
Planetary waves in horizontal and vertical shear: The generalized Eliassen–Palm relation and the mean zonal acceleration
.
J. Atmos. Sci.
,
33
,
2031
2048
, doi:.
Ashok
,
K.
,
S. K.
Behera
,
S. A.
Rao
,
H.
Weng
, and
T.
Yamagata
,
2007
:
El Niño Modoki and its possible teleconnection
.
J. Geophys. Res.
,
112
, C11007, doi:.
Bombardi
,
R. J.
, and
L. M. V.
Carvalho
,
2011
:
The South Atlantic dipole and variations in the characteristics of the South American monsoon in the WCRP-CMIP3 multi-model simulations
.
Climate Dyn.
,
36
,
2091
2102
, doi:.
Bombardi
,
R. J.
,
L. M. V.
Carvalho
,
C.
Jones
, and
M. S.
Reboita
,
2014
:
Precipitation over eastern South America and the South Atlantic sea surface temperature during neutral ENSO periods
.
Climate Dyn.
,
42
,
1553
1568
, doi:.
Carvalho
,
L. M. V.
,
C.
Jones
, and
B.
Liebmann
,
2004
:
The South Atlantic convergence zone: Intensity, form, persistence, and relationships with intraseasonal to interannual activity and extreme rainfall
.
J. Climate
,
17
,
88
108
, doi:.
Ciasto
,
L. M.
,
G. R.
Simpkins
, and
M. H.
England
,
2014
:
Teleconnections between tropical Pacific SST anomalies and extratropical Southern Hemisphere climate
.
J. Climate
, 28, 56–65, doi:.
De Almeida
,
R. A. F.
,
P.
Nobre
,
R. J.
Haarsma
, and
E. J. D.
Campos
,
2007
:
Negative ocean–atmosphere feedback in the South Atlantic convergence zone
.
Geophys. Res. Lett.
,
34
, L18809, doi:.
Ding
,
Q.
,
E.
Steig
,
D.
Battisti
, and
M.
Küttel
,
2011
:
Winter warming in West Antarctica caused by central tropical Pacific warming
.
Nat. Geosci.
,
4
,
398
403
, doi:.
Ding
,
Q.
,
E.
Steig
,
D.
Battisti
, and
J.
Wallace
,
2012
:
Influence of the tropics on the southern annular mode
.
J. Climate
,
25
,
6330
6348
, doi:.
Doyle
,
M. E.
, and
V. R.
Barros
,
2002
:
Midsummer low-level circulation and precipitation in subtropical South America and related sea surface temperature anomalies in the South Atlantic
.
J. Climate
,
15
,
3394
3410
, doi:.
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 atmospheric circulation in the Southern Hemisphere
.
Int. J. Climatol.
,
23
,
663
677
, doi:.
Grimm
,
A. M.
,
J.
Pal
, and
F.
Giorgi
,
2007
:
Connection between spring conditions and peak summer monsoon rainfall in South America: Role of soil moisture, surface temperature, and topography in eastern Brazil
.
J. Climate
,
20
,
5929
5945
, doi:.
Haarsma
,
R. J.
,
E. J. D.
Campos
,
W.
Hazeleger
,
C.
Severijns
,
A. R.
Piola
, and
F.
Molteni
,
2005
:
Dominant modes of variability in the South Atlantic: A study with a hierarchy of ocean–atmosphere models
.
J. Climate
,
18
,
1719
1735
, doi:.
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
, doi:.
Hoskins
,
B. J.
, and
D. J.
Karoly
,
1981
:
The steady linear response of a spherical atmosphere to thermal and orographic forcing
.
J. Atmos. Sci.
,
38
,
1179
1196
, doi:.
Kalnay
,
E.
, and Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
, doi:.
Mo
,
K. C.
,
2000
:
Relationships between low-frequency variability in the Southern Hemisphere and sea surface temperature anomalies
.
J. Climate
,
13
,
3599
3610
, doi:.
Molteni
,
F.
,
2003
:
Atmospheric simulations using a GCM with simplified physical parameterization. I: Model climatology and variability in multi-decadal experiments
.
Climate Dyn.
,
20
,
175
191
, doi:.
Morioka
,
Y.
,
T.
Tozuka
, and
T.
Yamagata
,
2011
:
On the growth and decay of the subtropical dipole mode in the South Atlantic
.
J. Climate
,
24
,
5538
5554
, doi:.
Morioka
,
Y.
,
T.
Tozuka
,
S.
Masson
,
P.
Terray
,
J.-J.
Luo
, and
T.
Yamagata
,
2012
:
Subtropical dipole modes simulated in a coupled general circulation model
.
J. Climate
,
25
,
4029
4047
, doi:.
Morioka
,
Y.
,
S.
Masson
,
P.
Terray
,
C.
Prodhomme
,
S. K.
Behera
, and
Y.
Masumoto
,
2014
:
Role of tropical SST variability on the formation of subtropical dipoles
.
J. Climate
,
27
,
4486
4507
, doi:.
Muza
,
M. N.
,
L. M. V.
Carvalho
,
C.
Jones
, and
B.
Liebmann
,
2009
:
Intraseasonal and interannual variability of extreme dry and wet events over southeastern South America and subtropical Atlantic during the austral summer
.
J. Climate
,
22
,
1682
1699
, doi:.
Robertson
,
A. W.
, and
C. R.
Mechoso
,
2000
:
Interannual and interdecadal variability of the South Atlantic convergence zone
.
Mon. Wea. Rev.
,
128
,
2947
2957
, doi:.
Rodrigues
,
R. R.
,
R. J.
Haarsma
,
E. J. D.
Campos
, and
T.
Ambrizzi
,
2011
:
The impacts of inter–El Niño variability on the tropical Atlantic and Northeast Brazil climate
.
J. Climate
,
24
,
3402
3422
, doi:.
Schneider
,
D. P.
,
Y.
Okumura
, and
C.
Deser
,
2012
:
Observed Antarctic climate variability and tropical linkages
.
J. Climate
,
25
,
4048
4066
, doi:.
Smith
,
T. M.
,
R. W.
Reynolds
,
T. C.
Peterson
, and
J.
Lawrimore
,
2008
:
Improvements to NOAA’s historical merged land–ocean surface temperature analysis (1880–2006)
.
J. Climate
,
21
,
2283
2296
, doi:.
Sterl
,
A.
, and
W.
Hazeleger
,
2003
:
Coupled variability and air–sea interaction in the South Atlantic Ocean
.
Climate Dyn.
,
21
,
559
571
, doi:.
Takaya
,
K.
, and
H.
Nakamura
,
2001
:
A formulation of a phase-independent wave-activity flux for stationary and migratory quasigeostrophic eddies on a zonally varying basic flow
.
J. Atmos. Sci.
,
58
,
608
627
, doi:.
Taschetto
,
A. S.
,
A.
Sen Gupta
,
N. C.
Jourdain
,
A.
Santoso
,
C. C.
Ummenhofer
, and
M. H.
England
,
2014
:
Cold tongue and warm pool ENSO events in CMIP5: Mean state and future projections
.
J. Climate
,
27
,
2861
2885
, doi:.
Tedeschi
,
R. G.
,
I. F. A.
Cavalcanti
, and
A. M.
Grimm
,
2013
:
Influences of two types of ENSO on South American precipitation
.
Int. J. Climatol.
,
33
,
1382
1400
, doi:.
Venegas
,
S. A.
,
L. A.
Mysak
, and
D. N.
Straub
,
1997
:
Atmosphere–ocean coupled variability in the South Atlantic
.
J. Climate
,
10
,
2904
2920
, doi:.
Vera
,
C.
,
G.
Silvestre
,
V.
Barros
, and
A.
Carril
,
2004
:
Differences in El Niño response over the Southern Hemisphere
.
J. Climate
,
17
,
1741
1753
, doi:.
Vigaud
,
N.
,
Y.
Richard
,
M.
Rouault
, and
N.
Fauchereau
,
2009
:
Moisture transport between the South Atlantic Ocean and southern Africa: Relationships with summer rainfall and associated dynamics
.
Climate Dyn.
,
32
,
113
123
, doi:.
Wilson
,
A. B.
,
D. H.
Bromwich
,
K. M.
Hines
, and
S. H.
Wang
,
2014
:
El Niño flavors and their simulated impacts on atmospheric circulation in the high-southern latitudes
.
J. Climate
,
27
,
8934
8955
, doi:.
Yeh
,
S.-W.
,
J.-S.
Kug
,
B.
Dewitte
,
M. H.
Kwon
, and
B. P.
Kirtman
,
2009
:
El Niño in a changing climate
.
Nature
,
461
,
511
514
, doi:.
Yuan
,
C.
,
T.
Tozuka
,
J.-J.
Luo
, and
T.
Yamagata
,
2014
:
Predictability of the subtropical dipole modes in a coupled ocean–atmosphere model
.
Climate Dyn.
,
42
,
1291
1308
, doi:.

Footnotes

1

ENSO events are defined according to the NOAA Climate Prediction Center methodology; see Table 2 for more details.