SEARCH

SEARCH BY CITATION

Keywords:

  • climatology;
  • climate variability;
  • Atlantic multidecadal oscillation;
  • El Niño-Southern Oscillation

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Results
  6. 4. Conclusion and summary
  7. Acknowledgements
  8. References

This article analyses the relations of the Atlantic multidecadal oscillation (AMO) and the El Niño-Southern Oscillation (ENSO) and their influence on the South American rainfall. The ENSO-related precipitation anomalous composites over South America show more (less) organized patterns with the significant anomalies occupying extensive (reduced) areas when ENSO and AMO are in the opposite (same) phase. The El Niño (La Niña) events in the cold (warm) AMO phase are, in general, stronger than those in the warm (cold) AMO phase. The strong El Niño (La Niña) events in the cold (warm) AMO phase are due to the presence of a negative (positive) inter-Pacific-Atlantic sea surface temperature (SST). The negative (positive) SST anomalies in the equatorial Atlantic reinforce the El Niño (La Niña) in the tropical Pacific through an anomalous Atlantic Walker circulation. In consequence, the ENSO-related precipitation anomalies over South America are more intense and with less horizontal structure under the existence of this connection between the climate variability of the tropical Atlantic and Pacific Oceans. As far as the authors know, the results presented here have not been discussed before and have important implications for regional climate monitoring, as well as for modelling studies. Copyright © 2013 Royal Meteorological Society

1. Introduction

  1. Top of page
  2. ABSTRACT
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Results
  6. 4. Conclusion and summary
  7. Acknowledgements
  8. References

El Niño-Southern Oscillation (ENSO), the major ocean–atmosphere coupled inter-annual mode in the tropics, causes significant inter-annual climate variations in large portions of the globe, in particular over South America, where the ENSO-related rainfall anomalies have been documented in innumerous papers (Kousky et al., 1984, Ropelewski and Halpert, 1987, Aceituno, 1988, Kayano et al., 1988, Kiladis and Diaz, 1989, Ropelewski and Halpert, 1989, Rao and Hada, 1990, Pisciottano et al., 1994, Grimm et al., 1998, Grimm et al., 2000, Montecinos et al., 2000, Souza et al., 2000, Barros and Silvestri, 2002, Cazes-Boezio et al., 2003, Grimm, 2003, Kayano, 2003, Vera et al., 2004, Andreoli and Kayano, 2005, Kayano and Andreoli, 2006, among others). The arguments to explain the El Niño-related (La Niña-related) dry (wet) conditions over northern and northeastern South America and nearly simultaneous opposite climate conditions over the southeastern part of this continent are well known and based on atmospheric circulation anomalies associated with an anomalous Walker circulation (Kousky et al., 1984, Kayano et al., 1988, Kousky and Ropelewski, 1989) and anomalous Hadley cells on both sides of the equator (Zhou and Lau, 2001).

Observational evidence that the ocean–atmosphere coupled modes of the Atlantic variability also play an important role in causing significant inter-annual climate variations over the surrounding continental areas has been shown by many authors. One of these modes for the sea surface temperature (SST) features a dipole or inter-hemispheric gradient pattern in the tropical Atlantic, which connects dynamically with a thermal direct meridional cell related to the Atlantic intertropical convergence zone (ITCZ). The mechanistic relation between the dipole mode and the ITCZ north–south migration in the tropical Atlantic is a argument commonly used to explain the inter-annual rainfall variability over northeastern Brazil (NEB) (e.g. Hastenrath and Heller, 1977, Weare, 1977, Hastenrath, 1978, Moura and Shukla, 1981, Hastenrath, 1990, Hastenrath and Greischar, 1993, Souza et al., 2000). Other modes contributing to the South American climate variability are those manifesting themselves dominantly in the South Atlantic Ocean (Markham and Mclain, 1977, Diaz et al., 1998, Andreoli and Kayano, 2006; Kayano et al., 2012). Diaz et al. (1998) found that the SST anomalies in the southwestern subtropical Atlantic and rainfall anomalies over Uruguay and southern Brazil during austral summer are simultaneous and positively correlated. Kayano et al. (2012) showed that this relation occurs through an anomalous SST mode with one centre in the southwestern South Atlantic and another in the southern mid-latitudes, which is in turn closely related to the ENSO.

Thus, both the Pacific and Atlantic Oceans, acting separately or together, play an important role in the rainfall inter-annual variability over South America. Dictated by the regions with ENSO-coherent rainfall anomalies, studies on the ENSO effects focused mainly on NEB (Giannini et al., 2004, Andreoli and Kayano, 2006, Kayano and Andreoli, 2006), northern Brazil (Andreoli et al., 2012) and southeastern South America (Diaz et al., 1998, Barros and Silvestri, 2002, Kayano, 2003). The continental view of the combined and independent effects of the Pacific and Atlantic Oceans on the South American rainfall variability is also found in the literature (Kayano et al., 2009, 2011, 2012, among others).

The above-mentioned studies analysed the effects of dominant coupled ocean–atmosphere modes of the inter-annual variability in the tropical Pacific, tropical Atlantic and South Atlantic without taking into account the existence of a low-frequency SST variability mode in the North Atlantic Ocean, called the Atlantic multidecadal oscillation (AMO) (Kerr, 2000) or a tripole mode (Czaja and Frankignoul, 2002, Peng et al., 2003). The AMO features almost the same sign SST anomalies in the North Atlantic with a main centre at 55°N and another one at 15°N. First identified during the 1990s (Delworth et al., 1993, Deser and Blackmon, 1993, Kushnir, 1994), it was more accurately characterized when longer and better-quality SST data became available (Enfield and Mestas-Nuñez, 1999, Mestas-Nuñez and Enfield, 1999, Enfield et al., 2001, Goldenberg et al., 2001, Peng et al., 2003, among others). Enfield et al. (2001) found a 65- to 80-year period oscillation for an index defined as the 10-year running mean of the detrended monthly SST anomaly time series in the entire North Atlantic. They found that the warm (or positive) AMO phase occurred during the 1860–1885 and 1925–1965 periods and the cold (or negative) AMO phase, during the 1895–1924 and 1970–1990 periods.

Knight et al. (2006), using a 1400-year control simulation of the HadCM3 climate model, argued that the northward displaced Atlantic ITCZ during the warm AMO phase might reduce the rainfall over NEB. The anomalous southward positioning of the ITCZ after 1955 documented by Rao et al. (2006) is likely due to the cold phase of the AMO. However, they did not associate the multidecadal variations of the ITCZ positioning with the AMO.

Evidence on the relations between the AMO and the SST variations in the tropical Pacific, was provided in some studies (Dong and Sutton, 2002, Dong et al., 2006, Dong and Sutton, 2007, Sutton and Hodson, 2007, Timmermann et al., 2007). These relations are summarized in Figure 11 by Timmermann et al. (2007) who showed that the cold (warm) AMO phase is associated with strong (weak) ENSO variability and reduced (increased) magnitude of annual cycle of the SST in the eastern tropical Pacific. Because the AMO affects the ENSO variability, it might also affect the ENSO effect on the South American rainfall. So, this aspect is focused here by stratifying the ENSO extreme years according to the AMO phases. For these years, ENSO-related atmospheric circulation and associated rainfall anomalies over South America are examined. The following section describes data and methodology. Composites for the precipitation, SST, sea level pressure (SLP) and vertical velocity are presented and discussed in Section 'Results'. Conclusions are given in Section 'Conclusion and summary'.

2. Data and methodology

  1. Top of page
  2. ABSTRACT
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Results
  6. 4. Conclusion and summary
  7. Acknowledgements
  8. References

Monthly gridded precipitation time series at 1° resolution derived from the gauge-based reconstructions of the Global Precipitation Climatology Center (GPCC) (Beck et al., 2005, Rudolf and Rubel, 2005) are used here. The GPCC full data reanalysis version 6 is used. Precipitation time series are selected for the 1901–2006 period over South America which encompasses the area bounded at 15.5°N, 55.5°S, 84.5°W and 29.5°W. The GPCC gridded precipitation data were generated by an operational analysis system which integrated the data from different sources, performed quality-control and calculated the area-averaged precipitation on the gridcells (Rudolf and Schneider, 2005). Errors that might exist in this dataset are due to deficiencies in the measuring process or to regionally sparse data (Rudolf and Schneider, 2005). Despite this, the gridded precipitation data is of high quality.

Other variables used here are obtained for the same period as that of the precipitation data, the 1901–2006 period. Among them, the monthly SST data reconstructed by Smith et al. (2008), as well as the monthly SLP, and vertical velocity at 19 vertical pressure levels derived from the Twentieth Century Reanalysis (Compo et al., 2011) are also used. These data are at 2° by 2° latitude–longitude resolution grids. The SST dataset consists of the version 3b of the monthly gridded SST data which is available at http://jisao.washington.edu/data/ersst/. The reanalysis data are from the version 2 of the Twentieth Century Reanalysis and are available at www.esrl.noaa.gov/psd/data/gridded/data.20thC_ReanV2. The SST and SLP time series are obtained in the global domain between 70°N and 70°S, and the vertical velocity time series, at 19 pressure levels from 1000 to 100 hPa in the global domain between 6°N and 6°S.

The analyses used here focus on the El Niño and La Niña events of the 1901–2006 period stratified according to the AMO phases. The onset years of these events are obtained using the Niño-3 SST index and following the criterion suggested by Trenberth (1997). The Niño-3 SST index is defined as the 5-month running mean of the monthly SST anomalies averaged in the area bounded at 4°N, 4°S, 150°W and 90°W. According to Trenberth (1997), an El Niño (a La Niña) event is identified when the Niño-3 SST index exceeds (is less than) the threshold of 0.5°C (−0.5 °C) for at least six consecutive months. Because the definition of the Niño-3 SST index includes a 5-month running mean smoothing, the SST time series used to calculate this index are selected for the 1900–2007 period. The Niño-3 SST index for the 1901–2006 period is detrended and depicted in Figure 9. The onset years of the ENSO extremes during the 1901–2006 period stratified according to the AMO phases are listed in Table 1. Hereinafter, the symbols (−1), (0) and (+1) are used to refer to the previous, onset and mature stage years of the El Niño or La Niña events. Interpretation of Table 1 is given in the next section.

Table 1. Onset years of the ENSO extremes stratified according to the AMO phases
 El NiñoLa Niña
WARM AMO1930, 1939, 1940, 1951, 1957, 2002, 2003, 20041933, 1938, 1942, 1949, 1955, 1999
COLD AMO1902, 1904, 1911, 1914, 1918, 1925, 1965, 1968, 1972, 1976, 1982, 1986, 19911906, 1908, 1916, 1920, 1922, 1924, 1964, 1967, 1970, 1973, 1975, 1984, 1988

The AMO index defined by Enfield et al. (2001) is recalculated using the SST time series in the North Atlantic area bounded at 66°N, equator, 80°W and Greenwich longitude for the 1866–2011 period. These time series are detrended and their monthly anomalies are obtained considering the means of the 1866–2011 period. According to these authors, the AMO index is defined as the 121-month running mean of the monthly detrended SST anomalies averaged in the North Atlantic area. The AMO index spanning from January 1871 to December 2006 is then obtained. The reason why the AMO index is calculated for a longer period (1871–2006) than that (1901–2006) with availability of precipitation data is to include balanced numbers of warm and cold AMO phases. In order to define the warm and cold periods, the last year before and the first year after the sign reversal of AMO index are considered neutral phase. Using this criterion and considering the 1901–2006 period, the warm AMO phase corresponds to the 1929–1958 and 1998–2006 periods, and the cold AMO phase, to the 1901–1925 and 1962–1994 periods.

The linear trends are removed only from the SST and SLP time series. Monthly anomalies of the detrended SST and SLP, and vertical velocity and precipitation are obtained as departures from the means of the 1901–2006 period. The monthly anomaly series at each grid point are normalized by the corresponding monthly standard deviations.

Monthly composites of the SST and SLP anomalies, and of the vertical velocity anomalies averaged between 6°N and 6°S are obtained as average of these anomalies for the months from July (0) to May (+1) of the selected years of each case. These composites are shown for every 2 months. Monthly composites of SST anomalies averaged between 4°N and 4°S are also obtained for the months of the period from January (−1) to October (+1) of the selected years of each case. To smooth the rainfall composite patterns, the ENSO-related rainfall anomalies are obtained for 3-month running composites. These composites are indicated by the initials of the months and are obtained from July–August–September(0) [JAS(0)] to April–May–June(+1) [AMJ(+1)], which are the trimesters spanning from the onset to decay stages of the ENSO extremes.

Because the number of years in each case is <30, the Student's t-test is the one recommended in the text books to assess the significance of the composite (mean) of small sample sizes. This test does not require the normal distribution of the sample. Thus, the statistical significance of the composites is assessed with the Student's t-test and with the number of the selected years for the composites as the number of degrees of freedom (n). So, only composites with absolute values exceeding inline image are statistically significant (Panofsky and Brier, 1968). In this formula, S is the standard deviation of the sample, the confidence level of 90%, and tα(n−1) is given in a Student's t-distribution table.

3. Results

  1. Top of page
  2. ABSTRACT
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Results
  6. 4. Conclusion and summary
  7. Acknowledgements
  8. References

3.1. ENSO years according to the AMO phases

Table 1 lists the ENSO years stratified according to the AMO phases. A higher frequency of the years with ENSO extremes during the cold AMO phase compared to the warm AMO phase is evident. The 26 years with ENSO extremes, equally distributed in El Niño and La Niña years, noted during the cold AMO phase represent 45% of the total number of years (58) of this AMO phase. On the other hand, the 14 years with ENSO extremes noted during the warm AMO phase account for 36% of the total number of years (39) of the warm AMO phase. Among these 14 years, eight are El Niño years and six are La Niña years. This result confirms the previous findings on the relation of the AMO and the ENSO variability (Dong and Sutton, 2002, Dong et al., 2006, Dong and Sutton, 2007, Sutton and Hodson, 2007, Timmermann et al., 2007). Because the ENSO variability depends on the AMO phases, the ENSO effects on the South American rainfall may also show differences depending on the AMO phases. These differences and the accompanied oceanic and atmospheric anomalies are illustrated in the composites of the precipitation, SST, SLP and vertical velocity at 19 levels.

3.2. El Niño composites

The precipitation anomaly composites of the El Niño years show differences between the cold and warm phases of the AMO (Figure 1(a) and (b)). In the cold AMO phase, the El Niño-related precipitation anomalous patterns show mostly a northwestern-southeastern dipolar structure with its centres occupying relatively large areas and evolving quite smoothly (Figure 1(a)). In JAS (0) and ASO (0), the significant negative anomalies are found in small areas over northern and northwestern South America, and the positive ones, in a large area of central and eastern Brazil. While the significant negative anomalies become more intense and extend over a large area including Venezuela, Guyana, Suriname, French Guiana and northern Brazil, the positive ones extend over southeastern South America in NDJ (0) and JFM (+1). Gradually, as the magnitude of the significant precipitation anomalies are reduced they are confined to relatively small areas, with the negative anomalies in the western Amazon in AMJ (+1), and the positive ones in southeastern South America (southern Brazil, Uruguay and central eastern Argentina) in FMA (+1), and in southeastern and central Brazil in AMJ (+1).

image

Figure 1. Composites of precipitation anomalies over South America from JAS (0) to AMJ (+1) of the El Niño years in the: (a) cold AMO phase; (b) warm AMO phase. Contour interval is 0.3 SD. Continuous (dashed) lines are positive (negative). The zero line is omitted. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

The El Niño-related precipitation patterns in the warm AMO phase show less organized and less extensive structure in most trimesters (Figure 1(b)) than those in the cold AMO phase (Figure 1(a)). In fact, in the warm AMO phase, the dipolar structure with the significant negative anomalies in northern South America and the positive ones in southeastern South America become well-established in SON (0). This pattern is intensified in OND (0), remains quite strong in NDJ (0), weakens in DJF (+1) and becomes less organized showing more horizontal structure in the subsequent trimesters.

Consistent with the differences in the El Niño-related precipitation anomalies between the AMO phases, differences in the associated composites for the other variables are also noted. Figures 2 and 3 show the El Niño-related SST and SLP anomalies, respectively, in the cold and warm phases of the AMO. Both the figures show El Niño coherent patterns for SST and SLP, but some differences are evident between the AMO phases. Among them, the most conspicuous difference is the larger magnitude of the SST anomalies in the eastern tropical Pacific and of the SLP anomalies in the tropical belt in the cold than in the warm phase of the AMO.

image

Figure 2. El Niño composites in the cold AMO phase of the: (a) SST anomalies; (b) SLP anomalies. Continuous (dashed) lines are positive (negative). The zero line is omitted. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

image

Figure 3. El Niño composites in the warm AMO phase of the: (a) SST anomalies; (b) SLP anomalies. Continuous (dashed) lines are positive (negative). The zero line is omitted. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

The determining factor for these differences seems to be the climate conditions in the equatorial Atlantic. The El Niño-related SST pattern in the cold AMO phase features significant negative anomalies in the equatorial Atlantic and positive ones along the eastern tropical Pacific in July (+0) (Figure 2(a)). This pattern is indeed the inter-Pacific-Atlantic east–west SST gradient (hereinafter, referred to as inter-Pacific-Atlantic SST mode), which is part of a feedback mechanism for the climate variability of tropical Pacific and Atlantic basins, as previously shown by many authors (Wang, 2006, Kucharski et al., 2007, Kucharski et al., 2008, Kucharski et al., 2009, Wang et al., 2009, Losada et al., 2010, Kayano et al., 2011). The negative inter-Pacific-Atlantic SST mode remains quite strong up to January (+1) and is maintained through the atmospheric bridge between the two tropical oceanic basins. In fact, the corresponding SLP composites in most months feature a well-defined zonal wavenumber one pattern in the tropics with significant positive SLP anomalies extending from the Atlantic eastward to western Pacific and the negative ones in the eastern Pacific (Figure 2(b)).

It is interesting to note that the positive SST anomalies extend over tropical North Atlantic (TNA) overwhelming the AMO-related negative SST anomalies in this region from March (+1) to May (+1). The positive SST anomalies in the TNA are likely a remote El Niño effect. Enfield and Mayer (1997) showed that an El Niño yields a warming in the TNA during late austral summer and autumn. It is also noticeable that a positive inter-hemispheric SST gradient manifests in the tropical Atlantic in May (+1). Consistent with this SST anomalous pattern in the tropical Atlantic and due to the El Niño SST mode in the tropical Pacific, negative precipitation anomalies are noted over northern NEB in MAM (+1) and AMJ (+1) (Figure 1(a)).

The El Niño-related SST composites in the warm AMO phase show significant positive SST anomalies in the eastern tropical Pacific from July (0) to May (+1) (Figure 3(a)). However, the magnitudes of these anomalies are smaller than those in the cold AMO phase. Consistently, the El Niño-coherent SLP patterns in the warm AMO phase (Figure 3(b)) are weaker than those in the cold AMO phase (Figure 2(b)). In the case of the warm AMO composites, the mechanism to reinforce the SST and SLP anomalies in the tropics is absent. This is evident in the Atlantic sector. In fact, the El Niño-related SST pattern in the warm AMO phase features significant positive anomalies in the subtropical and extratropical North Atlantic which are associated with the warm AMO phase, and neutral SST anomalies in the equatorial and most of the tropical South Atlantic (TSA) in July (0) (Figure 3(a)). Gradually, the positive anomalies enhance and extend southward occupying the TNA, equatorial Atlantic and most of the TSA from January (+1) to May (+1). Under these conditions, the inter-Pacific-Atlantic SST mode is not formed and the feedback mechanism for the climate variability of the tropical Pacific and Atlantic basins is also absent. So, the El Niño-related SST and SLP anomalies in warm AMO phase (Figure 3(a) and (b)) show smaller magnitudes than those in the cold AMO phase (Figure 2(a) and (b)).

The above results show that the El Niño events in the cold AMO phase are more intense than those in the warm AMO phase. In order to better illustrate this aspect, the El Niño composites in the cold and warm phases of the AMO of the vertical velocity anomalies averaged between 6°N and 6°S are shown in vertical zonal sections (Figure 4(a) and (b)). In both AMO phases, the upward motions in the central and eastern Pacific longitudes, as shown by significant negative vertical velocity anomalies, are flanked to the east and west by sinking motions in July (0). Gradually, the vertical velocity anomalies evolve, with the negative anomalies reaching their maximum magnitudes in January (+1), and weakening in the following months. In all analysed months, the vertical velocity anomalies show larger magnitudes in the cold than in the warm AMO phase. This reinforces that the AMO affects the El Niño events in such a way that more (less) intense El Niño events occur during the cold (warm) AMO phase due to the presence (absence) of the feedback mechanism of the climate variability of tropical Pacific and Atlantic basins associated with the negative inter-Pacific-Atlantic SST mode.

image

Figure 4. Composites of the vertical velocity anomalies averaged between 6°N and 6°S for the El Niño years in the: (a) cold AMO phase; (b) warm AMO phase. The zero line is omitted. Continuous (dashed) lines encompass positive (negative) significant values at the 90% confidence level using the Student's t-test. The colour bar indicates the shading intervals in standard deviations.

Download figure to PowerPoint

3.3. La Niña composites

Similarly to the El Niño composites, the La Niña-related precipitation anomalies show differences between the AMO phases (Figure 5(a) and (b)). In most trimesters, the precipitation anomaly composites of the La Niña years in the cold AMO phase do not show a well-defined northern/southeastern dipolar structure (Figure 5(a)). In fact, significant positive precipitation anomalies are found over northern and northwestern South America and the negative ones scattered in small areas of central and eastern Brazil during JAS (0) and ASO (0). The significant positive precipitation anomalies remain in small areas along the northern coast of South America, and the negative ones become more intense and extend over Uruguay, southern Brazil and central eastern Argentina from SON (0) to NDJ (0). While the significant negative precipitation anomalies weaken, the positive ones enhance and extend over northern and northeastern South America from DJF (+1) to FMA (+1). The significant positive precipitation anomalies remain over NEB and central eastern Amazon, and the significant negative precipitation anomalies become more intense over southeastern South America in MAM (+1) and AMJ (+1).

image

Figure 5. Composites of precipitation anomalies over South America from JAS (0) to AMJ (+1) of the La Niña years in the: (a) cold AMO phase; (b) warm AMO phase. Continuous (dashed) lines are positive (negative). The zero line is omitted. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

Conversely, the dipolar configuration of precipitation anomaly patterns of the La Niña years in the warm AMO phase noted from JAS (0) to DJF (+1) evolve quite smoothly (Figure 5(b)). Indeed, significant positive precipitation anomalies are found over northern South America, and the negative ones, in a large area of central South America between 10°S and 30°S during JAS (0). While the significant positive anomalies extend southeastward occupying NEB, the negative ones split into two areas, one over southeastern South America and another over the western Amazon in SON (0) and OND (0). Gradually, the positive anomalies are confined over northern South America and the negative ones weaken considerably in DJF (+1). The precipitation anomalies weaken further and the significant positive precipitation anomalies remain over western central Amazon by AMJ (+1).

The La Niña-related precipitation anomalies in the cold (warm) AMO phase show a less (more) organized horizontal structure and extend over small (extensive) areas. The composites of the other variables show consistent differences between the cold and warm phases of the AMO. The monthly SST and SLP anomaly composites of the La Niña years in the cold and warm phases of the AMO are depicted in Figures 6 and 7, respectively. The composites of these variables show La Niña coherent patterns. However, the SST anomalies in the eastern tropical Pacific and the SLP anomalies in the tropical belt are more intense in the warm AMO phase than in the cold AMO phase (Figures 6 and 7).

image

Figure 6. La Niña composites in the cold AMO phase of the: (a) SST anomalies; (b) SLP anomalies. Continuous (dashed) lines are positive (negative). The zero line is omitted. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

image

Figure 7. La Niña composites in the warm AMO phase of the: (a) SST anomalies; (b) SLP anomalies. Continuous (dashed) lines are positive (negative). The zero line is omitted. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

As with the El Niño composites, the SST anomaly patterns in the equatorial Atlantic have a crucial role in determining the intensities of the La Niña events. In July (0), the La Niña SST composites in the cold AMO phase show a La Niña coherent pattern in the eastern tropical Pacific, significant negative anomalies in the TNA which are the tropical part of AMO in its cold phase, and neutral conditions in most of the TSA (Figure 6(a)). These SST anomalous conditions in the eastern tropical Pacific and tropical Atlantic remain up to March (+1). Therefore, the inter-Pacific-Atlantic SST mode is not formed in the La Niña SST composites in the cold AMO phase from July (0) to March (+1). In consequence, the feedback mechanism connecting the climate variability of the tropical Pacific and Atlantic basins is also absent. However, the SST anomaly pattern in the tropical Atlantic features a negative inter-hemispheric SST mode, which is particularly conspicuous in May (+1) (Figure 6(a)). Consistent with this pattern and due to the presence of a La Niña in the tropical Pacific, positive precipitation anomalies are noted over northern NEB from NDJ (0) to AMJ (+1).

On the other hand, the La Niña-related SST pattern in the warm AMO phase features significant negative anomalies in the eastern tropical Pacific and the positive ones in the North Atlantic including the TNA and the equatorial Atlantic, and negative SST anomalies in the subtropical South Atlantic in July (0) (Figure 7(a)). The positive SST anomalies in the TNA and equatorial Atlantic sectors are part of the warm AMO phase, and together with the negative SST anomalies in the eastern tropical Pacific feature a positive inter-Pacific-Atlantic SST mode. This mode, as part of a feedback mechanism for the climate variability of tropical Pacific and Atlantic basins (Wang, 2006, Kucharski et al., 2007, Kucharski et al., 2008, Kucharski et al., 2009, Wang et al., 2009, Losada et al., 2010), remains quite strong up to January (+1). The associated atmospheric bridge between the two oceanic tropical basins maintains this mode as revealed in the corresponding SLP composites (Figure 7(b)). The SLP composites show a zonal wavenumber one pattern with significant positive anomalies in the eastern tropical Pacific flanked by opposite sign SLP anomalies to the east and west (Figure 7(b)). This pattern is conspicuous from November (0) to January (+1). It is interesting to note that the negative SST anomalies extend over TNA overwhelming the AMO-related positive anomalies in this region from March (+1) to May (+1). In this case, the negative SST anomalies in the TNA are likely due to the La Niña effect. In fact, Enfield and Mayer (1997) showed that a La Niña yields cooling in the TNA during the austral late summer and autumn months due to the wind/evaporation/SST feedback mechanism (Chang et al., 1997).

The La Niña-related vertical velocity anomalies averaged between 6°N and 6°S in the cold and warm phases of the AMO are shown in Figure 8(a) and (b). In both AMO phases, strong sinking motions in the central and eastern Pacific longitudes are flanked by rising motions to the east and west in all analysed months. The vertical velocity anomalies reach their maximum magnitudes in January (+1), and weaken in the following months. It is remarkable that the largest vertical velocity anomalies occur in the warm AMO phase in all analysed months. Thus, the more (less) intense La Niña events during the warm (cold) AMO phase are due to the presence (absence) of a positive inter-Pacific-Atlantic SST mode and the associated feedback mechanism of the climate variability of tropical Pacific and Atlantic basins.

image

Figure 8. Composites of the vertical velocity anomalies averaged between 6°N and 6°S for the La Niña years in the: (a) cold AMO phase; (b) warm AMO phase. Continuous (dashed) lines encompass positive (negative) significant values at the 90% confidence level using the Student's t-test. The colour bar indicates the shading intervals in standard deviations.

Download figure to PowerPoint

4. Conclusion and summary

  1. Top of page
  2. ABSTRACT
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Results
  6. 4. Conclusion and summary
  7. Acknowledgements
  8. References

Previous papers found evidence of the relations between the AMO and the SST variations in the tropical Pacific (Dong and Sutton, 2002, Dong et al., 2006, Dong and Sutton, 2007, Sutton and Hodson, 2007, Timmermann et al., 2007). Timmermann et al. (2007) showed that the cold (warm) AMO phase is associated with strong (weak) ENSO variability and reduced (increased) magnitude of annual cycle of the SST in the eastern tropical Pacific. Here, we analyse further the relations between the ENSO and the AMO by stratifying the ENSO extreme years according to the AMO phases, and obtaining through the composite technique the ENSO-related atmospheric circulation and associated rainfall anomalies over South America. The analyses are done for the 1901–2006 period using SST, SLP, vertical velocity at 19 pressure level and precipitation.

The most remarkable result is that the ENSO-related precipitation anomalous composites over South America show more (less) organized patterns with the significant anomalies occupying extensive (reduced) areas when ENSO and AMO are in the opposite (same) phase. In other words, the El Niño- (La Niña-) related precipitation anomalies in the cold (warm) AMO phase are more accentuated and show less horizontal structure occupying larger areas than those in the warm (cold) AMO phase. These rainfall anomalous aspects over South America are due to stronger El Niño (La Niña) events in the cold (warm) AMO phase than in the warm (cold) AMO phase.

Concerning the occurrence of stronger La Niña events noted here during the warm AMO than during the cold AMO phase seems not to be consistent with previous finding that ENSO variability is stronger during cold than during the warm AMO phase. In order to clarify the apparent controversial results, one must consider several important aspects. Most of the previous findings are based on the ENSO oceanic component represented by the Nino-3 SST index or by the eastern equatorial Pacific SST anomalous pattern (Dong et al., 2006, Dong and Sutton, 2007, Timmermann et al., 2007). So, the east–west SST gradient in the tropical Pacific is not taken into account. Furthermore, the higher ENSO variability during the AMO cold phase does not necessarily imply that both El Niño and La Niña events are strong. In other words, the higher ENSO variability might be dictated by one or by both ENSO events.

In order to illustrate the relationship of the ENSO variability and the AMO phases, the Niño-3 SST and the AMO indices are illustrated in Figure 9. The 121-month running variances and the variances multiplied by −1 are displayed together with the Niño-3 SST index in Figure 9. The low (high) variances of the Niño-3 SST index are particularly conspicuous during the warm (cold) AMO phase during the 1929–1958 period (1901–1925 and 1962–1994 periods). It is noticeable that the ENSO variability is more strongly defined by the magnitudes of the positive values of the Niño-3 SST index. The magnitudes of the positive values of the Niño-3 SST index representing the El Niño events of the cold AMO phase (1901–1925 and 1962–1994), in general, are larger than those of the warm AMO phase (1929–1958). This result supports our findings that, in general, the El Niño events in the cold AMO phase are stronger than those in the warm AMO phase. On the other hand, the magnitudes of the negative values of the Niño-3 SST index representing La Niña events do not show remarkable variability between the AMO phases, except at the end of the 1929–1958 period, when slightly larger magnitudes of the negative values of the Niño-3 SST index are noted. In this case, the east–west SST anomaly gradient in the tropical Pacific should be more crucial in defining the La Niña strength.

In order to further illustrate the ENSO intensities in terms of the east–west SST anomaly gradient in the equatorial Pacific, composites of the SST anomalies averaged between 4°N and 4°S are obtained in the zonal band between 120°E and 80°W from January (−1) to October (+1) of the El Niño and La Niña years (Figures 10 and 11). In the cold AMO phase, an El Niño is preceded by a La Niña and vice-versa. This confirms the higher ENSO variability in the cold rather than in the warm phase of the AMO, as previously documented (Dong and Sutton, 2002, Dong et al., 2006, Dong and Sutton, 2007, Sutton and Hodson, 2007, Timmermann et al., 2007). Another interesting aspect is that the El Niño- (La Niña-) related east–west SST gradient in the equatorial Pacific in the cold (warm) AMO phase is stronger than that in the warm (cold) AMO phase. Furthermore, the longitudinal band of the maximum El Niño-related warming of 1.2 standard deviations extends further west in the cold phase AMO (160–90°W) than in the warm phase AMO (100–90°W). For the La Niña, the maximum cooling has a similar longitudinal extension in both AMO phases, however, it is stronger in the warm (−1.2 SD) than in the cold (−0.9 SD) AMO phase. Thus, the ENSO-related east–west SST gradient in the equatorial Pacific varies with the AMO phases. This result has an important implication for modelling studies.

image

Figure 9. Niño-3 SST and AMO indices (units in °C). The light (dark) shade indicates positive (negative) values. The continuous black line with positive (negative) values displayed with Niño-3 SST index represents the plot of the 121-month running variances (variances multiplied by −1). The vertical dotted lines correspond to the 1927, 1960 and 1996 years.

Download figure to PowerPoint

image

Figure 10. Composites of the SST anomalies averaged between 4°N and 4°S in the Pacific longitudes in the 120°E–80°W band from January (−1) to October (+1) of the El Niño years in the: (a) cold AMO phase; (b) warm AMO phase. Contour interval is 0.3 SD. Continuous (dashed) lines are positive (negative). The zero line is the continuous thicker contour. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

image

Figure 11. Composites of the SST anomalies averaged between 4°N and 4°S in the Pacific longitudes in the 120°E–80°W band from January (−1) to October (+1) of the La Niña years in the: (a) cold AMO phase; (b) warm AMO phase. Continuous (dashed) lines are positive (negative). The zero line is the continuous thicker contour. Dark (light) shaded areas encompass significant positive (negative) values at the 90% confidence level using the Student's t-test.

Download figure to PowerPoint

Moreover, the strong El Niño (La Niña) events in the cold (warm) AMO phase are due to the presence of a negative (positive) inter-Pacific-Atlantic SST mode. The negative (positive) SST anomalies in the equatorial Atlantic reinforce the El Niño (La Niña) in the tropical Pacific through an anomalous Atlantic Walker circulation. The dynamics of the mechanism connecting the climate variability of the tropical Pacific and Atlantic basins was previously discussed by many authors (Wang, 2006, Kucharski et al., 2007, Kucharski et al., 2008, Kucharski et al., 2009, Wang et al., 2009, Losada et al., 2010, Kayano et al., 2011, among others). In contrast, the weak El Niño (La Niña) events in the warm (cold) AMO phase are due to the absence of a negative (positive) inter-Pacific-Atlantic SST mode.

These results provide evidence that the AMO-related SST anomaly pattern in the North Atlantic creates a strong climate background. If a La Niña (El Niño) event is developed in the tropical Pacific under the warm (cold) AMO conditions, a positive (negative) inter-Pacific-Atlantic SST mode is established in July (0) to January (+1) and the negative (positive) SST anomalies in the TNA overwhelm the AMO-related positive (negative) SST anomalies in this Atlantic sector from March (+1) to May (+1). The negative (positive) SST anomalies in the TNA from March (+1) to May (+1) are the remote influence of a strong La Niña (El Niño). According to Enfield and Mayer (1997), an El Niño (a La Niña) yields a warming (cooling) in the TNA during the late austral summer and autumn months due to the wind/evaporation/SST feedback mechanism (Chang et al., 1997). On the other hand, if an El Niño (La Niña) event is formed in the tropical Pacific under warm (cold) AMO phase, the inter-Pacific-Atlantic SST mode does not develop, and the same sign anomalies in the eastern tropical Pacific and TNA are the outstanding features from July (0) to May (+1). These results show that the canonical ENSO-related SST anomalous pattern connecting the tropical Pacific and the TNA (Enfield and Mayer, 1997) are not necessarily related to strong ENSO events. Rather, they can be formed due to the pre-existing SST pattern in the TNA associated with the AMO.

The analyses here clearly indicate that the strength of the ENSO extremes is closely connected to the AMO. The ENSO extremes are strengthened through the atmospheric bridge between the equatorial Atlantic and the eastern tropical Pacific associated with the inter-Pacific-Atlantic SST mode. In consequence, the ENSO-related precipitation anomalies over South America are more intense and with less horizontal structure when the ENSO extremes in the equatorial eastern Pacific are reinforced by equatorial Atlantic SST variability mode of opposite sign as that of the equatorial eastern Pacific. To the knowledge of the authors, the results presented here have not been discussed before and have important implications for regional climate monitoring, as well as for modelling studies.

Acknowledgements

  1. Top of page
  2. ABSTRACT
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Results
  6. 4. Conclusion and summary
  7. Acknowledgements
  8. References

The first author was partially supported by the Conselho Nacional de Desenvolvimento Científico e Tecnológico of Brazil, and the second author was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior of Brazil. This article is derived from the doctoral thesis of the second author. The authors thank the two anonymous reviewers for their helpful comments and suggestions.

References

  1. Top of page
  2. ABSTRACT
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Results
  6. 4. Conclusion and summary
  7. Acknowledgements
  8. References