Atmosphere–Ocean Coupled Variability in the South Atlantic

2904 JOURNAL OF CLIMATE VOLUME 10

Atmosphere±Ocean Coupled Variability in the South Atlantic

S. A. VENEGAS,L.A.MYSAK, AND D. N. STRAUB Centre for Climate and Global Change Research and Department of Atmospheric and Oceanic Sciences, McGill University, Montreal, Quebec, Canada (Manuscript received 9 July 1996, in ®nal form 9 April 1997)

ABSTRACT The climate variability of the South Atlantic region is determined from 40 yr (1953±92) of Comprehensive Ocean±Atmosphere Data Set monthly sea surface temperature (SST) and sea level pressure (SLP) data using the empirical orthogonal function (EOF) and the singular value decomposition (SVD) analysis methods. The EOF method is applied to each ®eld separately, whereas the SVD method is applied to both ®elds simultaneously. The signi®cance of the atmosphere±ocean interaction is revealed by a strong resemblance between individual (EOF) and coupled (SVD) modes of SST and SLP. The three leading modes of coupled variability on interannual and interdecadal timescales are discussed in some detail. The ®rst coupled mode, which accounts for 63% of the total square covariance, represents a 14±16-yr period oscillation in the strength of the subtropical anticyclone, accompanied by ¯uctuations of a north±south dipole structure in the SST. The atmosphere±ocean coupling is strongest during the southern summer. The second coupled mode (20% of the total square covariance) is characterized by east±west shifts of the anticyclone center, in association with 6±7-yr period ¯uctuations of SST off the coast of Africa. The coupling depicted by this mode is weaker than that found in the ®rst and third modes. The third coupled mode (6% of the total square covariance) is characterized by north±south displacements of the anticyclone, accompanied by SST ¯uctuations over a latitudinal band in the central South Atlantic. These oscillations occur on a relatively short interannual timescale (ϳ4 yr). As with the ®rst mode, the atmosphere±ocean coupling is strongest during the southern summer. This mode is found to be temporally and spatially correlated with the El NinÄo±Southern Oscillation phenomenon. The statistical robustness of the results is tested by using a Monte Carlo approach, which indicates that the presented results are highly signi®cant.

1. Introduction sure (SLP) ¯uctuations and their coupled variability During the past few decades there has been consid- have also been widely investigated by analyzing a va- erable effort devoted to obtaining a better understanding riety of datasets particularly over the North Atlantic and of natural climate variability on interannual to inter- the North Paci®c Oceans (Deser and Blackmon 1993; decadal timescales. To determine the mechanisms gov- Kushnir 1994; Latif and Barnett 1996; Mann and Park erning these climatic variations, it is essential to char- 1996) and by performing experiments with atmospheric, acterize the large-scale interactions between the ocean oceanic, or coupled atmosphere±ocean models (Palmer and the overlying atmosphere. A number of recent stud- and Sun 1985; Kushnir and Lau 1992; Delworth et al. ies have sought to identify such interactions and their 1993; Zebiak 1993; Peng et al. 1995; Halliwell 1996). associated timescales, by analyzing both observational Thermodynamic and wind-induced mechanisms such records and the results of climate model simulations. as air±sea heat exchange, entrainment, vertical mixing, The nature of the global air and sea surface temper- and wind stress forcing, have been frequently proposed ature variability on interannual to interdecadal time- as ways in which the atmosphere forces upper-ocean scales has been documented in both data studies (Fol- temperature anomalies (Frankignoul 1985; Wallace et land et al. 1984; Ghil and Vautard 1991; Houghton and al. 1990; Cayan 1992a; Cayan 1992b; Miller et al. 1994; Tourre 1992; Mann and Park 1993; Mann and Park Battisti et al. 1995). Other studies, however, have sug- 1994) and model simulations (Manabe and Stouffer gested that the ocean may drive atmospheric circulation 1994; Mehta and Delworth 1995). The relationships be- anomalies through large-scale changes in the thermo- tween sea surface temperature (SST) and sea level pres- haline circulation on interdecadal or longer timescales timescales (Levitus 1989; Ghil and Vautard 1991; Stocker and Mysak 1992). A few investigators have also proposed positive feedbacks between SST and the at- Corresponding author address: Dr. Lawrence A. Mysak, Centre for Climate and Global Change Research, McGill University, 805 mospheric circulation on interdecadal timescales (Deser Sherbrooke Street West, Montreal, PQ H3A 2K6, Canada. and Blackmon 1993; Latif and Barnett 1994). E-mail: [email protected] A common feature of these studies is that the major

᭧ 1997 American Meteorological Society

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2905 modes of large-scale variability in the Northern Hemi- sphere are characterized by interannual to interdecadal timescales. It has been generally accepted that interannual variability is primarily driven by the atmosphere, while interdecadal or longer-scale variability is mainly associated with changes in the ocean (Deser and Black- mon 1993; Kushnir 1994). However, recent work by Halliwell (1996) and results from this paper suggest that atmospheric forcing of the ocean on interdecadal timescales may be more important than previously realized. As a complement to the Northern Hemisphere focus of the above studies, this paper (which is partly sum- marized in Venegas et al. 1996) investigates the natural climate variability in one region of the Southern Hemi- sphere, namely, the South Atlantic. The ®rst goal is to identify the principal modes of behavior of the SST and the overlying atmospheric circulation, in order to pro- FIG. 1. Percentage of months in which data was available at each grid point over the 480-month period analyzed. vide insight into the variability of the South Atlantic coupled atmosphere±ocean system on interannual to interdecadal timescales. A secondary goal is to determine pling based on the SVD analysis are discussed in section whether the South Atlantic modes of variability are con- 4. A summary and the conclusions are given in section 5. nected with any of the well-known tropical and/or Northern Hemisphere climate oscillations (e.g., El NinÄo±Southern Oscillation, North Atlantic Oscillation, 2. Data and methods Paci®c±North American pattern). The suggestion of a. Data such a connection between hemispheres on interdecadal The data employed in this study include South At- timescales was proposed in Mysak and Power (1992). lantic sea SST, sea level pressure (SLP), and vector wind These connections were further explored by Mann and extracted from the Comprehensive Ocean±Atmosphere Park (1993, 1994). Data Set (COADS) (Woodruff et al. 1987). These data Empirical orthogonal function (EOF) and singular are available as monthly means over a grid of 2Њ lat ϫ value decomposition (SVD) analyses are used to de- 2Њ long, and span the period 1854±1992. The 40-yr scribe both the independent and coupled variability of period analyzed in this paper extends from 1953±92, SST and SLP in the South Atlantic. While the EOF during which time the data coverage over the South analysis method has been widely applied in geophysics Atlantic basin is fairly good. The area of interest of our research (Davis 1976; Wallace et al. 1990; Deser and study is the South Atlantic Ocean, from the equator to Blackmon 1993; En®eld and Mayer 1997), the SVD 50ЊS, and from 70ЊW (or South American coast) to 20ЊE analysis method has become more commonly used only (or South African coast). The region south of 40ЊS be- in the last decade (Lanzante 1984; Fang and Wallace tween 40ЊW and 20ЊE is excluded from the analysis, 1994; Peng and Fyfe 1996), although it was ®rst pro- however, due to poor data coverage (see Fig. 1). posed in a meteorological context by Prohaska (1976). The SST, SLP, and wind climatologies are calculated The interest in trying to model the type of South for each calendar month at each grid point by averaging Atlantic variability described in this paper has substan- the data over the 40-yr period. Monthly anomalies are tially increased since the creation of two recent large- then de®ned as deviations from this mean annual cycle. scale climate programs: CLIVAR (Climate Variability To reduce the noise inherent in the monthly observa- and Predictability Program) and ACCP [The Atlantic tions, the anomalies are temporally smoothed by per- Climate Change Program, described in Molinari et al. forming three-month running means (3MRM). At a giv- (1994)]. It is hoped that this observational study will en grid point, 3MRMs are computed by averaging offer new insights into the mechanisms and long-range anomalies of three consecutive months and by assigning linkages of interannual to interdecadal climate vari- that mean value to the central month. If two or more ability. These insights will also help the modeling stud- neighboring monthly anomalies are missing, the cor- ies associated with these programs. responding 3MRM is not computed and the month is The paper is organized as follows. A brief description given a missing value ¯ag. of the datasets and methods, together with an introduc- For a successful application of the EOF and SVD tion to the main South Atlantic oceanic and atmospheric analysis methods described in the next subsection, the features are given in section 2. Results from the inde- spatial gaps in the SST and SLP data are ®lled using pendent SST and SLP EOF analyses follow in section an inverse distance interpolation method (Thiebaux and 3. The results pertaining to the atmosphere±ocean cou- Pedder 1987). As a ®nal step before applying the sta-

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2906 JOURNAL OF CLIMATE VOLUME 10 tistical methods, the data from the original 2Њ by 2Њ grid are averaged onto a coarser 4Њ lat ϫ 10Њ long grid to reduce the size of the datasets used in our analysis. b. Methods Detailed discussions on EOF and SVD analyses can be found in Bretherton et al. (1992), Wallace et al. (1992), von Storch and Navarra (1995), and Newman and Sardeshmukh (1995). A brief description of both methods follows here. EOF analysis is a statistical technique based on fun- damental matrix operations. It has proven to be helpful in objectively identifying the principal (spatially uncorrelated) modes of variability of a given ®eld. The analysis applies to a space- and time-dependent ®eld with zero temporal mean. The covariance matrix of the FIG. 2. Schematic representation of the South Atlantic upper-level ®eld is constructed and diagonalized, resulting in a set ocean circulation (from Peterson and Stamma 1991). of eigenvalues and corresponding eigenvectors. Each eigenvector (EOF) can be regarded as a spatial pattern (a map). To see how a given spatial pattern ``evolves'' to be complete. When there are gaps in the data, and in time, the eigenvector is projected onto the original these are taken into account in the construction of the ®eld to obtain a time series (the expansion coef®cient). covariance matrix, the resulting modes are no longer Just as the EOFs are orthogonal in space, the associated strictly orthogonal. This problem is overcome here by time series are orthogonal in time. The fraction of the ®lling the gaps as described previously. The differences total ®eld variance explained by a given EOF is pro- in the results using the original and the interpolated portional to its associated eigenvalue. Together, an ei- datasets are, however, minimal. The analyses presented genvalue with its corresponding EOF and expansion co- here are based on the interpolated data. ef®cient de®ne a mode of variability. The leading mode Correlation maps are used extensively in this study. (related to the largest eigenvalue) explains the largest These consist of gridpoint correlations between a time fraction of the total variance, the second mode explains series (e.g., the expansion coef®cient of an EOF mode, the largest fraction of the remaining variance, and so an index such as SOI, etc.) and gridpoint anomalies of on. a given ®eld. Thus, correlation maps indicate how well SVD analysis can be thought of as a generalization the gridpoint anomalies of the ®eld can be predicted to rectangular matrices of the diagonalization of a square from the knowledge of the time series. In addition, they symmetric matrix (i.e., EOF analysis). It is usually ap- provide a measure of the percentage of the variance plied to two data ®elds together in order to identify pairs explained locally by each mode, represented by the of coupled spatial patterns, which explain as much as square of the correlations (Houghton and Tourre 1992). possible the covariance between the two variables. The The spectral analyses and coherences performed in SVD of the cross-covariance matrix yields two spatially this study are based on the multitaper method (Thomson uncorrelated sets of singular vectors (analogous to the 1982), and the con®dence levels are obtained relative eigenvectors, but one for each variable) and a set of to an estimated red noise background (Mann and Lees singular values associated with each pair of vectors 1996). Signi®cance levels for all the correlations are (analogous to the eigenvalues). Each pair of singular estimated using the method described in Sciremammano vectors describe a fraction of the square covariance (SC) (1979). This method accounts for the month-to-month between the two variables. The ®rst pair describes the autocorrelation inherent in the time series to compute largest fraction of the SC and each succeeding pair de- the actual number of degrees of freedom before the scribes a maximum fraction of the SC that is unex- estimation of signi®cance levels. The 95% signi®cance plained by the previous pairs. The square covariance level is given in square brackets after each correlation fraction (SCF) accounted for by the kth pair of singular coef®cient. A Monte Carlo approach is used to deter- vectors is proportional to the square of the kth singular mine the signi®cance of the SST and SLP decomposi- value. The kth expansion coef®cient for each variable tions in modes of variability. A detailed description of is computed by projecting the respective kth singular the method is given in section 4d. vector onto each original data ®eld. The correlation value r between the kth expansion coef®cients of the two c. Mean South Atlantic conditions variables indicates how strongly related the coupled patterns are. Figure 2 depicts the ocean circulation features of the Both EOF and SVD analysis methods assume the data South Atlantic. The upper-level circulation is dominated

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2907

compare the two sets of EOFs with the set of SVD modes obtained for the coupled atmosphere±ocean system in section 4. Only a brief discussion of the EOF (independent) modes of variability is given here, since this study concentrates on the SVD (coupled) modes.

a. Oceanic variability An EOF analysis is performed on the unnormalized (i.e., not divided by the standard deviation) monthly SST anomalies over the South Atlantic Ocean for the period 1953±92. The three leading EOF modes together account for 47% of the total monthly SST variance. Individually, they explain 30%, 11%, and 6% of the variance. According to North's rule of thumb (North et al. 1982), the difference between our third and fourth eigenvalues is comparable to the magnitude of their sampling errors, which means that the error in the EOFs is comparable to the size of the EOFs themselves. Hence, these two modes are not well separated and the decomposition between them may not be stable. Nev- ertheless, we keep the third mode since it will prove to be of interest in connection with searching for links to other well-known climatic ¯uctuations. The spatial patterns associated with the ®rst three SST FIG. 3. (a) SST (in ЊC) climatology for 1953±92. Isotherm interval is 2ЊC. (b) SLP (in hPa minus 1000 hPa) and vector wind climatol- modes are depicted in Fig. 4 as homogeneous correlation ogies for 1953±92. Isobar interval is 3 hPa. maps. They are, respectively, labeled E1(SST), E2(SST), and E3(SST). Figure 5 shows the temporal variability of each EOF, represented by the expansion by the South Equatorial Current, the Subtropical Gyre, coef®cients e1(SST), e2(SST), and e3(SST), respec- and the Antarctic Circumpolar Current (ACC). The main tively. components of the Subtropical Gyre are the warm and E1(SST) (Fig. 4a) exhibits a monopole pattern ex- saline Brazil Current in the west and the cool and fresh tending over the entire domain. Recalling that the square Benguela Current in the east, joined together by the of the correlation values represents the local variance westward ¯owing South Atlantic Current. The cold and explained (Houghton and Tourre 1992), this mode ac- fresh Malvinas (or Falkland) Current originates as a counts for up to 64% of the variance in the region of branch of the ACC and meets the Brazil Current at the largest loadings, namely, off the coast of southern Af- con¯uence zone (ϳ36ЊS). rica. The fraction of local variance explained decreases The mean annual ®elds of SST, SLP, and vector wind toward the south and west. Similar temporal and spatial are computed as the 40-yr (1953±92) climatologies from structures of the ®rst EOF mode of SST were found by the COADS monthly means, and are shown in Fig. 3. Houghton and Tourre (1992) and by En®eld and Mayer The SST ®eld exhibits a generally weak southeast± (1997) in the common domain (0Њ to 20Њ±30ЊS) of their northwest gradient north of ϳ35ЊS. The isotherms ex- EOF analyses of the tropical Atlantic Ocean (30ЊNto tend zonally and are more closely spaced south of this 30ЊS). latitude. The SLP and wind ®elds are dominated by the E2(SST) (Fig. 4b) displays an out-of-phase relation- South Atlantic subtropical anticyclone whose center is ship between temperature anomalies north and south of located near 30ЊS, 5ЊW. It is surrounded by southeasterly about 25Њ±30ЊS. This mode explains up to 20%±40% trades to the north and westerlies to the south. The fol- of the variance near its centers of action. lowing analyses of the nonseasonal variability in this E3(SST) (Fig. 4c) exhibits three latitudinal bands of study describe departures from these mean patterns. centers of action with alternating signs. A 20Њ-wide band centered around 25ЊS is surrounded by two bands of the 3. The variability of SST and SLP: EOF analysis opposite polarity north of 15ЊS and south of 35ЊS. The temperature ¯uctuations in the center of action near the The purpose of this section is to analyze indepen- coast of South Africa account for up to 20% of the total dently the variability of the oceanic surface temperatures ®eld variance. and the atmospheric sea level pressure during the 40-yr The time series e1(SST) and e2(SST) (Figs. 5a,b) are period using the EOF method. By doing this, we can characterized by a mixture of interannual and interde-

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2908 JOURNAL OF CLIMATE VOLUME 10

FIG. 5. Time series of expansion coef®cients e1(SST), e2(SST), and s3(SST) of the three EOF modes of SST. The time series are smoothed using a 13-month running mean (13MRM), and their amplitudes are normalized by the standard deviation.

concerning the separation between the third and fourth modes. Once again, we keep the third mode since it is of particular interest to our results. Figure 6 shows the three spatial patterns E1(SLP), E2(SLP), and E3(SLP), as homogeneous correlation maps. Figure 7 shows their associated expansion coef®cients e1(SLP), e2(SLP), and e3(SLP). E1(SLP) (Fig. 6a) has uniform polarity over the entire domain. The center of action of the mode is located near the center of the subtropical anticyclone (see Fig. 3b). Hence, this mode describes the strengthening and weak- FIG. 4. Spatial patterns E1(SST), E2(SST), and S3(SST) of the ening of the anticyclone. E2(SLP) (Fig. 6b) has a dipole three EOF modes of SST. The spatial patterns are presented as ho- structure, which is related to east±west changes in the mogeneous correlation maps; i.e., the contours are scaled such that the value at each grid point is the correlation coef®cient between the location of the center of the subtropical anticyclone. time series of expansion coef®cients of Fig. 5 and the SST anomaly Another dipole structure can be seen in E3(SLP) (Fig. at that grid point. Contour interval is 0.1. Negative contours are 6c). In this case, changes in the position of the anti- dashed. Zero line is thicker. Correlations higher than (in absolute cyclone occur in the north±south direction. value) (a) Ϯ0.25, (b) Ϯ0.21, and (c) Ϯ0.16 are signi®cant at the 95% The time series e1(SLP) (Fig. 7a) is dominated by level. low-frequency (decadal-scale) oscillations, while e2(SLP) and e3(SLP) (Fig. 7b,c) exhibit predominantly cadal ¯uctuations, while e3(SST) (Fig. 5c) oscillates on interannual ¯uctuations. mainly interannual timescales. c. Relating oceanic and atmospheric behaviors b. Atmospheric variability To look for possible links between oceanic and at- A similar EOF analysis is performed on the unnor- mospheric ¯uctuations, the correlation coef®cients be- malized monthly SLP anomalies over the same domain tween the time series of the three discussed SST and and time period as for SST. The ®rst three EOF modes SLP modes are calculated and displayed in Table 1. account for 59% of the total monthly SLP variance, Correlations with the time series of the Southern Os- while they individually explain fractions of 35%, 16%, cillation index (SOI) are also included to test for pos- and 8%. The same considerations as for SST apply here sible links with the El NinÄo±Southern Oscillation

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2909

FIG. 7. Time series of expansion coef®cients e1(SLP), e2(SLP), and s3(SLP) of the three EOF modes of SLP. The time series are smoothed using a 13MRM, and their amplitudes are normalized by the standard deviation.

(ENSO) phenomenon. The SOI is de®ned as the normalized sea level pressure difference between Tahiti and Darwin. The table shows two signi®cant correlations (at the 95% level) between SST and SLP modes and two signi®cant correlations between ENSO and the third modes of SST and SLP. The time series e1(SLP) and e2(SST) are signi®cantly negatively correlated (Ϫ0.34), which is also evident from a visual comparison of the series in Figs. 7a and 5b. The fact that the ®rst mode of the atmospheric variability is related to the second mode of the ocean vari- FIG. 6. Spatial patterns E1(SLP), E2(SLP), and S3(SLP) of the three EOF modes of SLP. The spatial patterns are presented as homogeneous ability suggests that the two are linked via atmosphere- correlation maps; i.e., the contours are scaled such that the value at to-ocean forcing, since we expect a strong signal to force each grid point is the correlation coef®cient between the time series a weaker one, and not the other way around. Lagged of expansion coef®cients of Fig. 7 and the SLP anomaly at that grid correlations between these two expansion coef®cients point. Contour interval is 0.1. Negative contours are dashed. Zero line (Fig. 8a) indicate that the two modes are best correlated is thicker. Correlations higher than (in absolute value) (a) Ϯ0.18, (b) Ϯ0.16, and (c) Ϯ0.13 are signi®cant at the 95% level. (Ϫ0.36) when the atmosphere leads the ocean temperature by 1±4 months, consistent with the idea that atmosphere-to-ocean forcing links these two modes of variability. The series e3(SLP) and e3(SST) are also signi®cantly

TABLE 1. Correlation coef®cients between the three ®rst EOF modes of SST and SLP and between these modes and the SOI. Numbers in brackets are the 95% signi®cance levels calculated for each correlation (Sciremammano 1979). Highly signi®cant correlations are in bold face. e1(SLP) e2(SLP) e3(SLP) SOI e1(SST) Ϫ0.06 [0.25] 0.15 [0.16] Ϫ0.09 [0.19] Ϫ0.13 [0.24] e2(SST) Ϫ0.34 [0.23] 0.17 [0.16] Ϫ0.18 [0.18] 0.18 [0.21] e3(SST) Ϫ0.11 [0.15] Ϫ0.07 [0.14] 0.24 [0.17] Ϫ0.35 [0.20] SOI Ϫ0.05 [0.17] 0.00 [0.16] Ϫ0.35 [0.20]

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2910 JOURNAL OF CLIMATE VOLUME 10

FIG. 8. (a) Lagged correlations between e1(SLP) and e2(SST). SLP leads SST for positive lags. (b) Lagged correlations between e3(SLP) and e3(SST). SLP leads SST for positive lags. Dashed lines show the 95% signi®cance levels. Lags are in months. The lagged correlations are based on the original time series (not the ones smoothed by a 13MRM). correlated (0.24). Lagged cross correlations between the value decomposition (SVD) is performed on the cross- two modes (Fig. 8b) exhibit two signi®cant correlations. covariance matrix between SST and SLP. In contrast to One of these shows the atmosphere leading the ocean the individual EOF analysis performed in the previous by 2 months (0.30), suggesting again the presence of section, the SVD analysis on the two ®elds together will an atmosphere-to-ocean forcing. The other (at ϳϪ21 identify only those modes of behavior in which the SST months) is probably a re¯ection of the ϳ3.5-yr peri- and SLP variations are strongly coupled. odicity that exists in the two time series. That is, the The three leading SVD modes of the coupled SST two series are out-of-phase at a lag equal to half the and SLP variations account for 89% of the total square period of oscillation (ϳ21 months). covariance (TSC). Each mode independently accounts The third modes of both SST and SLP are found to for 63%, 20%, and 6% of the TSC, respectively. We be signi®cantly correlated with the SOI (Ϫ0.35). This label the spatial patterns as Sk, and the expansion co- suggests that ENSO has a discernible signal in the South ef®cients as sk, k ϭ 1, 3. Table 2 displays the square Atlantic, even though these modes explain only a small covariance fractions (SCF) explained by each mode and percentage of the variability in their respective ®elds the correlation coef®cient (r) between the expansion co- (6% and 8% for SST and SLP, respectively). Figure 9 ef®cients of both variables [sk(SST) and sk(SLP)], as shows lagged correlations between both third mode time indicators of the strength of the coupling. The ®rst mode series and the SOI. The highest correlations with ENSO exhibits a strong and highly signi®cant coupling be- are found when the variable (SST or SLP) leads the tween SST and SLP. By contrast, the coupling coef®- SOI time series by 1±3 months. Signi®cant correlations cient associated with the second mode is smaller and at lag ϳ21 months presumably have the same origin as closer to its signi®cance level than those of the other those in Fig. 8b. two, which suggests a relatively weaker coupling. Even though the third SVD mode explains the least variance, 4. The variability of the coupled ®elds: SVD its corresponding coupling coef®cient is higher than that analysis of the second mode. This reinforces our decision to include it in our discussion even though its associated To better assess and con®rm the relationships between singular value is not well separated from the fourth one SST and SLP variations obtained in section 3, a singular (following North et al. 1982). Seasonal means were computed from the monthly SST and SLP data and the SVD analysis was repeated to test the sensitivity of the results to the change of seasons. The southern winter is de®ned as the 6-month period from May to October, and the summer as the

TABLE 2. Square covariance fraction (SCF) and coupling correlation coef®cient between the expansion coef®cients of both variables (r) corresponding to the three leading SVD modes. In brackets are the FIG. 9. (a) Lagged correlations between e3(SST) and SOI. SST 95% signi®cance levels for the correlations. leads SOI for positive lags. (b) Lagged correlations between e3(SLP) and SOI. SLP leads SOI for positive lags. Dashed lines show the Mode SCF r 95% signi®cance levels. Lags are in months. The lagged correlations 1 63% 0.46 [0.29] are based on the original time series (not the ones smoothed by a 2 20% 0.30 [0.20] 13MRM). The SOI monthly time series is smoothed by a 3MRM to 3 6% 0.41 [0.21] be consistent with the other data.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2911

TABLE 3. Square covariance fraction (SCF) and coupling correlation coef®cient (r) for the three SVD modes using 6-month winters and summers. In brackets are the 95% signi®cance levels. Winter Summer Mode SCF r SCF r 1 59% 0.59 [0.40] 68% 0.74 [0.56] 2 18% 0.42 [0.42] 19% 0.52 [0.43] 3 10% 0.58 [0.44] 5% 0.68 [0.47]

FIG. 11. Same as Fig. 10 but for the second SVD mode. Correlations higher than (in absolute value) Ϯ0.25 (SST) and Ϯ0.15 (SLP) are signi®cant at the 95% level.

6-month period from November to April. Table 3 shows the same parameters as Table 2 but after the seasonal breakdown of the data. The coupling of the three modes is most signi®cant during the southern summer, which is a somewhat sur- prising result. It is likely that the seasonality existent in the data coverage may account in part for the stronger coupling in summer. However, this argument should ap- FIG. 10. Spatial patterns and expansion coef®cient of the ®rst SVD ply to studies on both hemispheres, and the majority of coupled mode of SST and SLP. Spatial patterns are homogeneous the Northern Hemisphere studies ®nd the strongest at- correlation maps. Contour interval is 0.1. Negative contours are mosphere±ocean couplings during winter. Our interpre- dashed. Zero line is thicker. Correlations higher than (in absolute value) Ϯ0.20 (SST) and Ϯ0.19 (SLP) are signi®cant at the 95% level. tation is that the coupling in the Southern Hemisphere Time series are smoothed by a 13MRM. Amplitudes are normalized does not necessarily depend on the strong air±sea con- by the standard deviation. trasts expected to occur during winter, since these are

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2912 JOURNAL OF CLIMATE VOLUME 10

probably not as strong as in the Northern Hemisphere, mainly due to the different distribution of land and ocean coverage of the Southern Hemisphere. Instead, we argue that the coupling is strongest during summer because of possible links with major climatic oscillations observed in the Northern Hemisphere, which are strongest during the northern winter (NAO, PNA, etc.). Therefore, the signals in both hemispheres seem to show their strongest amplitudes simultaneously. This idea will be explored further in the next subsections. A difference should be made, however, when analyzing the second mode in more detail. Although the simultaneous coupling is only signi®cant in summer, a lagged correlation analysis of the seasonal coupling coef®cients shows that the coupling is similarly signi®cant in winter when SST leads SLP by one winter (0.51, [0.43], not shown). Hence, in contrast to the ®rst and third modes, the second mode shows a similar degree of coupling in both seasons, although the forcing seems to act with different time lags in winter and summer. The coupled spatial patterns and expansion coef®- cients corresponding to each variable and each mode are displayed in Figs. 10, 11, and 12. Lagged cross correlations between the SST and SLP time series associated with each mode are shown in Fig. 13. The three modes of coupled SST and SLP variability are discussed separately in the following sections.

a. First mode Figure 10 depicts the spatial patterns and the expansion coef®cient of the ®rst SVD mode of the coupled ®elds. The SLP time series and pattern closely resemble those obtained for the ®rst EOF mode of SLP (Fig. 6a). On the other hand, the SST time series and pattern for the ®rst SVD mode resemble those of the second EOF mode of SST (Fig. 4b), with the signs reversed. Hence this mode is related to the correlation found between E1(SLP) and E2(SST) (Table 1). The correlation be- FIG. 12. Same as Fig. 10 but for the third SVD mode. Correlations higher than (in absolute value) Ϯ0.22 (SST) and Ϯ0.16 (SLP) are tween s1(SLP) and e1(SLP) is 0.93 [0.22], while the signi®cant at the 95% level. correlation between s1(SST) and e2(SST) is Ϫ0.63 [0.31]. Thus, the EOF mode of SLP alone provides more

FIG. 13. Lagged correlations of the coupling coef®cient r corresponding to each SVD mode. SLP leads SST for positive lags. Dashed lines show the 95% signi®cance levels. Lags are in months. The lagged correlations are based on the original time series (not the ones smoothed by a 13MRM).

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2913

FIG. 14. Correlation maps between s1(SLP) and SST gridpoint anomalies: (a) for SLP leading SST by two months and (b) for SLP lagging SST by two months. information on the coupling between the two ®elds than the subtropical anticyclone, accompanied by a slight the EOF mode of SST alone. displacement of the center toward the northeast in the A lagged correlation analysis between s1(SST) and weak anticyclone situations. The difference in the cen- s1(SLP) (Fig. 13a) indicates that the coupling is strong- tral pressure between both extrema reaches 6 hPa. est when SLP leads SST by 1±2 months (0.51 [0 29]), Spectral analyses of both s1(SST) and s1(SLP) (see corroborating what we inferred from Fig. 8a. Therefore, Fig. 4 in Venegas et al. 1996) exhibit highly signi®cant the ®rst SVD mode seems to depict an atmosphere-to- peaks at low frequencies (14±16 yr), indicating that in- ocean forcing, in which the ocean response to the at- terdecadal timescales dominate this mode. A compari- mospheric changes appears with an intraseasonal time son with the results obtained by Mann and Park (1996) lag. According to Wallace and Jiang (1987), a signature suggests the possibility of a relation between s1(SLP) of this kind of forcing is the marked asymmetry with and their interdecadal joint mode in Northern Hemi- respect to the zero lag exhibited in Fig. 13a. The sig- sphere surface temperature and sea level pressure. Fig- ni®cance of this asymmetry can be illustrated by con- ure 16 shows the South Atlantic mode s1(SLP) leading trasting the ϩ2- and Ϫ2-month lag correlation maps the Mann and Park interdecadal reference signal by ap- between s1(SLP) and the gridpoint SST anomalies proximately 4±5 yr. Even though the correlation be- (shown in Fig. 14). In agreement with what we infer tween the two time series shown in Fig. 16 is not highly from Fig. 13a, the SST pattern obtained when SLP leads signi®cant (0.89 [0.84] using a 5-yr running mean to SST by two months (Fig. 14a) exhibits much larger smooth the data), partly due to the shortness of our time amplitudes than its counterpart pattern (Fig. 14b) ob- series, we speculate that there may exist a connection. tained when SLP lags SST by two months. This contrast Furthermore, Mann and Park's analyses reveal that the clearly indicates that a maximum in the amplitude of Northern Hemisphere SLP anomalies associated with the SST signal appears as a response to a prior maximum the interdecadal signal are pronounced only during the in the amplitude of the SLP signal. However, the op- northern winter. This is simultaneous with the strongest posite is not true, supporting the hypothesis of an at- expression of the South Atlantic signal during the south- mosphere-to-ocean forcing. ern summer (Table 3). An independent analysis of global To illustrate the behavior of this mode of variability surface temperature variability (Mann and Park 1994) in terms of atmospheric con®gurations, Fig. 15 shows shows a consistent global-scale temperature signal in composites of the pressure distribution of the months phase with s1(SST), whose associated spatial pattern corresponding to the ®ve highest and the ®ve lowest exhibits large SST variability near 35ЊS off the coast of values of s1(SLP). These composites describe the at- South America, consistent with S1(SST). mospheric conditions in the two extreme situations of The time series s1(SLP) is found to be highly cor- the modal ¯uctuations. It is clear that this mode depicts related (0.55 [0.17]) with the SLP variations at Gough an oscillation between strengthening and weakening of Island (40.4ЊS, 9.9ЊW) (as depicted in Fig. 17). This

FIG. 15. SLP composites corresponding to extreme positive and negative values of s1(SLP). Isobars are in hPa minus 1000 hPa. Contour interval is 3 hPa.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2914 JOURNAL OF CLIMATE VOLUME 10

FIG. 17. Time series of monthly SLP anomalies at Gough Island FIG. 16. Comparison of the South Atlantic ®rst coupled mode with (thin line), located at 40.4ЊS, 9.9ЊW, and s1(SLP) (thick line). Both the North Paci®c projection of the interdecadal joint mode of Mann time series are smoothed by a 13MRM, and the amplitudes are nor- and Park (1996). malized by the standard deviation. A roughly 15-yr period oscillation is evident in both series. island is located just east of the center of action of S1(SLP) (Fig. 10c), therefore its SLP time series pro- a positive feedback between atmosphere and ocean. vides a useful index for describing the ®rst mode of Such an SST anomaly induces changes in the atmo- variability in the South Atlantic. Since these observa- spheric circulation by modifying the north±south SST tions are independent from those of COADS, the cor- gradient and these atmospheric changes in turn reinforce relation with s1(SLP) lends support to the existence of the existent SST anomaly by air±sea heat exchange ¯uc- such oscillations in the strength of the subtropical an- tuations. The resemblance between the SST and SLP ticyclone. patterns of Fig. 10 and their North Paci®c SST and SLP To investigate possible air±sea forcing mechanisms patterns and the similar (interdecadal) timescales sug- related to the ®rst SVD mode, the wind anomalies as- gest that such a feedback may also be present in the sociated with this mode are presented in Fig. 18 together South Atlantic. However, further work is needed to con- with S1(SST) (same as Fig. 10a). The correlations are ®rm this hypothesis. computed with the wind leading the SST by 1 month, since this lag was found to depict the strongest inter- b. Second mode action between atmosphere and ocean (Fig. 13a). The ®rst EOF mode of the wind anomalies over the South Figure 11 shows the components of the second SVD Atlantic (not shown) displays a distribution of wind mode of the coupled ®elds. The SST time series and anomalies similar to that observed in Fig. 18, indicating pattern closely resemble those of the ®rst EOF mode of that such a wind pattern is a preferred mode of vari- SST, with signs reversed (Fig. 4a). There is also some ability in the atmosphere. similarity between the SLP time series and pattern and The observed relationship between the wind and SST those of the second EOF mode of SLP, again with signs anomalies is mainly local. Stronger-than-normal trade reversed (Fig. 6b). The correlation coef®cient between winds (strong anticyclone episode) coincide with cool ocean temperatures in the northern part of the basin, while weaker-than-normal westerlies coincide to warm ocean temperatures in a band between 30ЊS and 40ЊS, lying west of 0Њ long. Also, southerly (northerly) winds are associated with negative (positive) SST anomalies. The local nature of the wind-SST relationship suggests that the anomalies in the ocean temperature are driven by wind-induced mechanisms (Wallace et al. 1990; De- ser and Blackmon 1993; Kushnir 1994). These results also suggest that the atmospheric forcing of the ocean temperatures on interdecadal timescales may be more important than previously thought. Recent work by Hal- liwell (1996) dealing with the generation of SST anomalies in the North Atlantic supports this hypothesis. FIG. 18. Wind and SST anomalies correlated with s1(SST). The u- and v-components of the wind anomalies are correlated with s1(SST) Latif and Barnett (1994) have recently suggested that at each grid point, and ``vector correlations'' are obtained by plotting SLP and SST anomalies in the North Paci®c distributed the components in vectorial form. The SST pattern is the same as in similarly to those in Fig. 10 could be in association with Fig. 10 (top). The wind anomalies lead the SST anomalies by 1 month.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2915

FIG. 19. Same as Fig. 15 but using s2(SLP). s2(SST) and e1(SST) is Ϫ0.91 [0.37], while that be- The central pressure remains at the same value during tween s2(SLP) and e2(SLP) is Ϫ0.65 [0.14]. Thus, two the whole cycle. facts seem to suggest that this mode of variability is Repeating the procedure used in section 4a to further associated with an ocean-to-atmosphere forcing: (i) the investigate the atmosphere±ocean relationship, Fig. 20 correlation betwen the two SST time series is stronger displays superimposed wind and SST correlations with than that between the two SLP time series, and (ii) this s2(SST). Both correlations are simultaneous here, since coupled mode is associated with the ®rst mode of in- the coupling in this mode was found to be strongest at dependent (EOF) SST variability. Further support for zero lag. To facilitate the interpretation, s2(SST) has this hypothesis comes from the lagged coupling cor- been multiplied by Ϫ1. Stronger-than-normal trade relation (Fig. 13b), which shows a signi®cant simulta- winds are observed in the northeastern part of the basin neous correlation between s2(SST) and s2(SLP) (0.30 over warmer-than-normal ocean temperatures. Thus, the [0.20]). The zero-lag correlation and the symmetry with SST-wind relationship cannot be explained by local respect to the zero lag depicted by this ®gure is sug- wind-induced processes, since such processes would im- gestive of an essentially immediate adjustment of the ply a negative SST anomaly being forced by anoma- atmosphere to changes in the ocean (Wallace and Jiang lously strong winds (Wallace et al. 1990; Deser and 1987). Blackmon 1993; Kushnir 1994). Additionally, a lagged The SLP composites in Fig. 19 are based on the ®ve highest and lowest values of s2(SLP), and illustrate the correlation between wind and SST indices computed at behavior of the atmospheric component in this mode. 0Њ,10ЊS [the center of action of S2(SST)] shows no An east±west displacement of the anticyclone center is signi®cant correlation for wind leading SST up to 10 clearly seen in these composites. A pressure trough ap- months. This lack of evidence for wind driving the SST pears (disappears) off the Argentine coast and the cir- anomalies also supports the ocean-to-atmosphere forc- culation becomes more meridional (zonal) when the an- ing hypothesis. ticyclone is at its easternmost (westernmost) position, The spectra of both s2(SST) and s2(SLP) (see Fig. 4 which occurs at the highest (lowest) values of s2(SLP). in Venegas et al. 1996) exhibit a signi®cant peak at 6± 7 yr. A spectral coherence (or cospectrum) analysis of s2(SLP) (only southern summer data) and the winter NAO index (as de®ned in Rogers 1984) exhibits a peak at the ϳ5-yr period, signi®cant at the 95% level (not shown). This period is consistent with the peak in the NAO spectrum at ϳ7 yr found by Rogers (1984). On the other hand, a temporal lagged correlation between summer s2(SLP) and the NAO index suggests that the North Atlantic signal leads the South Atlantic one by ϳ3 yr (Fig. 21). Even though this correlation is mar- ginally signi®cant at the 95% level, the evidence given by such correlations in the frequency and time domains suggest the possibility of a relationship between these North and South Atlantic modes of variability. Further

FIG. 20. Same as Fig. 18 but for the second mode. SST and Wind work is needed to understand the nature of this con- correlations are simultaneous in this case. nection.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2916 JOURNAL OF CLIMATE VOLUME 10

related to this mode (Fig. 25). A westerly wind anomaly associated with the strong SLP north±south gradient around 25ЊS [see S3(SLP) in Fig. 12] produces a weakening of the climatological easterly winds (see Fig. 3b). The weaker-than-normal winds result in smaller-than- normal ocean-to-atmosphere heat ¯uxes, which produces a zonally oriented positive SST anomaly in a band around 25ЊS. Furthermore, a lagged correlation between SST and zonal wind indices computed at 25ЊW, 2 5ЊS [the center of action of S3(SST)] shows a signi®cant correlation for zonal wind leading SST by 2 months (0.15 [0.11]), which reveals the local nature of the wind forcing. The behavior of the atmospheric ®eld in this mode FIG. 21. Lagged correlations between the winter NAO index (De- is illustrated by the SLP composites shown in Fig. 22. cember±March) and summer s2(SLP) (December±March). SLP leads A comparison of the two composites shows that a north- NAO for positive lags. Dashed lines show the 95% signi®cance levels. Lags are in years. ward (southward) displacement of the high pressure center and stronger (weaker)-than-normal trade winds near the equator correspond to high (low) values of s3(SLP). c. Third mode The north (south) shift in the location of the subtropical Figure 12 shows the components of the third SVD anticyclone leads (by 1±2 months) the warming (cool- mode. The SLP and SST time series and patterns bare ing) of the ocean over the latitudinal band centered some resemblance to the respective third EOF modes around 25Њ±30ЊS and extending from coast to coast (Fig. (Figs. 4c and 6c), which were previously found to be 12a). correlated with one another (Table 1). The correlation Based on Table 1, we anticipate a relationship be- coef®cient between s3(SLP) and e3(SLP) is 0.55 [0.15] tween this mode and ENSO. This is con®rmed in Fig. and that between s3(SST) and e3(SST) is 0.40 [0.18]. 23, which shows lagged correlations of s3(SLP) and Both correlations are signi®cant but much smaller than s3(SST) with the SOI. The highest correlations (Ϫ0.37, the ones obtained in the ®rst two modes. Also in contrast [0.20]) correspond to a 1-month lag, both SST and SLP to the previous modes, neither the SLP nor the SST leading SOI. Since the signi®cant correlations are neg- coupled patterns appear clearly from the independent ative, and negative values of SOI roughly correspond EOF analyses. to the mature phase of ENSO, this result suggests that Figure 13c shows the lagged coupling correlations of a discernible warming in the central South Atlantic oc- this mode, with the highest values (ϳ0.42 [0.21]) occurs ϳ1 month before the maximum warming in the curing when SLP leads SST by 1±2 months. There is equatorial Paci®c. also a marked asymmetry with respect to zero lag in The spatial distributions of the SST and SLP anom- this ®gure, with signi®cant correlations for SLP leading alies in the South Atlantic during ENSO episodes are SST by up to 10 months. This suggests that this mode depicted in Figs. 24a,b by correlation maps between SOI may be associated with an atmosphere-to-ocean forcing, and the grid point SST and SLP anomalies. The main as we discussed in section 4a (Wallace and Jiang 1987). centers of action displayed by these two maps are anal- Further support for this hypothesis comes from the com- ogous to those of S3(SST) and S3(SLP) (Fig. 12), with parison of the SST pattern with the wind anomalies the signs reversed. This lends support to the suggestion

FIG. 22. Same as Fig. 15 but using s3(SLP).

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2917

FIG. 23. (a) Lagged correlations between s3(SST) and the SOI. SST leads SOI for positive lags. (b) Lagged correlations between s3(SLP) and SOI. SLP leads SOI for positive lags. Dashed lines show the 95% signi®cance levels. Lags are in months. The lagged correlations are based on the original time series (not the ones smoothed by a 13MRM). The SOI monthly time series is smoothed by a 3MRM to be consistent with the other data. that this mode represents the ENSO signal in the South by-product of the fact that ENSO is the common de- Atlantic. Similar ENSO-related SST and SLP spatial nominator. The second one is a broad signi®cant peak patterns are shown in Fig. 16.8 of Peixoto and Oort between 2.3 and 2.7 yr, which suggests a strong con- (1992). nection between this South Atlantic mode and the PNA The large southeasterly wind anomalies in the trades on the quasibiennial timescale. Lagged temporal cor- observed in the wind pattern associated with this mode relations of the PNA index with s3(SLP) and s3(SST) (Fig. 25) are consistent with the ®ndings of En®eld and (Fig. 26) corroborate this connection, suggesting that Mayer (1997), who propose that the ENSO signal in the the positive phase of the PNA (low pressure over eastern tropical Atlantic consists of a northward migration of Paci®c, high over North America, low over western the intertropical convergence zone (ITCZ) accompanied North Atlantic) occurs ϳ6±7 months after the northward by strong southeasterly trades. shift of the South Atlantic subtropical anticyclone and The power spectra of s3(SST) and s3(SLP) (see Fig. the warming in the central South Atlantic. 4 in Venegas et al. 1996) exhibit signi®cant peaks at A positive trend is also detectable in both s3(SST) ϳ4 yr, which coincides with the typical ENSO period and 3(SLP) (Fig. 12). This trend during the 40-yr period of oscillation. The spectral coherence between SOI and 1953±92 is in good agreement with the 90±100 yr os- s3(SLP) con®rms their correlation in the frequency do- cillation found by Mann and Park (1994, Fig. 4). They main, showing a highly signi®cant peak at 4±5 yr (not suggest that a strong cooling trend over the northern shown) North Atlantic from ϳ1940 to the present is associated The spectral coherence between s3(SLP) and the PNA with a warming trend (of a much weaker amplitude) in index (as de®ned in Wallace and Gutzler 1981) has two the South Atlantic. signi®cant peaks (not shown). The ®rst one, at ϳ4yr, corresponds to the ENSO timescales and is consistent d. Statistical signi®cance test with the correlation between the positive phase of the PNA pattern and the equatorial Paci®c warmings (Horel To asses the statistical robustness of the results ob- and Wallace 1981). Hence, this connection is likely a tained from the SVD analysis, we perform here a sig-

FIG. 24. Correlation maps between the SOI and (a) gridpoint anomalies and (b) gridpoint SLP anomalies.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2918 JOURNAL OF CLIMATE VOLUME 10

FIG. 25. Same as Fig. 18 but for the third mode. ni®cance test using a Monte Carlo approach. The focus is on the SCs (the total square covariance, as well as FIG. 27. Total SC and SC accounted for by the ®rst three SVD that corresponding to each mode) rather than the SCFs modes from the observed SST and SLP datasets (asterisks) and from (square covariance fractions) or the r coupling coef®- the 100 scrambled datasets (dots). All the SCs are normalized by cients (Wallace et al. 1992). The SC is a direct measure dividing by the number of grid points of each variable (NM ϭ 117 of the coupling between the two ®elds, while the SCF ϫ 117). and r are only meaningful when they are associated with a signi®cant SC. We thus test the signi®cance of the total SC between the SST and SLP anomalies (the sum autocorrelation than the SST data, in order to minimize of the squares of all the singular values) and the SC the increase of degrees of freedom inherent in the scram- associated with the three ®rst SVD modes (the square bling procedure. of the corresponding singular values). We then perform an SVD analysis on the scrambled The procedure is similar to that described in Wallace datasets, and repeat the same procedure 100 times. An et al. (1992) and Peng and Fyfe (1996). We create a SC value from the original datasets will be statistically ``randomized'' SLP dataset by scrambling the SLP maps signi®cant at the 95% level if it is not exceeded by more of the 40 yr in the time domain, in order to break the than ®ve values of the corresponding SC from the chronological order of SST relative to SLP. The scram- scrambled datasets. The results of the 100 scrambled bling is done on the years and not on the months; that SVD analyses are shown in Fig. 27, together with the is, the 12 SST maps of a given year will be paired with results of the original SVD based on the observations. 12 SLP maps of another (randomly chosen) year, but The total SC and the SC accounted for by the three keeping the order of the months inside the year. In this modes in the original run are plotted with an asterisk, way, we keep the month-to-month autocorrelation in- while the corresponding values of the 100 randomized herent in the time series and do not deteriorate the in- runs are marked with small dots. The observed SC, SC1, traseasonal variability, which would lower the signi®- and SC2 are found to be highly signi®cant (at more than cance levels and make our results appear more signif- 99% level). Only two values of the scrambled runs ex- icant than they really are. We choose to scramble only ceeded the observed SC3, indicating that it is signi®cant the SLP dataset since it has a smaller month-to-month at the 98% con®dence level.

FIG. 26. (a) Lagged correlations between s3(SST) and the PNA index. SST leads PNA for positive lags. (b) Lagged correlations between s3(SLP) and the PNA index. SLP leads PNA for positive lags. Dashed lines show the 95% signi®cance levels. Lags are in months.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC NOVEMBER 1997 VENEGAS ET AL. 2919

In conclusion, the results discussed in this work are ¯uctuations over a latitudinal band in the central South based on a statistically signi®cant decomposition of Atlantic Ocean. Relatively high-frequency interannual modes of variability of the coupled SST and SLP ®elds. timescales (ϳ4 yr) dominate the SST and SLP oscillations. Evidence is presented that suggests an atmospheric forcing of the ocean in this mode. However, the 5. Discussion and conclusions interaction between the ocean and atmosphere cannot The three leading EOF modes of variability of the be described by local wind-related mechanisms. The individual SST and SLP ®elds in the South Atlantic spatial patterns and temporal behavior of the SST and region have been identi®ed and analyzed. The leading SLP anomalies are found to be strongly associated with modes of oceanic variability account for 47% of the ENSO. The ENSO-related signal in the South Atlantic total SST variance. A slightly larger percentage (59%) leads the Paci®c warmings by 1±2 months. A relation- of the total SLP variance is accounted for by the leading ship with the Northern Hemisphere PNA atmospheric modes of atmospheric variability. It is interesting to note pattern is also suggested. A warming trend identi®ed in that both decompositions leave a relatively large fraction this mode may be associated with a global century-scale of the total variance unexplained. oscillation. The principal features that characterize the individual The processes involved in the air±sea interactions de- modes of SST and SLP behavior are found again in the picted by these coupled modes of variability and the SVD of the coupled ®elds. The main patterns of vari- nature of the links between Southern and Northern ability of both variables independently provide consid- Hemisphere ¯uctuations constitute the main subjects of erable information on the coupling, but only one of the investigation for further work. two variables dominates each of the two ®rst coupled modes. While the ®rst coupled mode keeps the char- Acknowledgments. We thank HalldoÂr BjoÈrnsson for acteristics of the main SLP (independent) pattern, the his input and assistance with the EOF and SVD analyses second coupled mode repeats the behavior of the main and Michael Mann for valuable discussions and help SST (independent) mode, suggesting the relative im- with the spectral analyses. Special thanks go to Yo- portance of each component of the system in each dis- chanan Kushnir for kindly reviewing the ®rst draft of cussed coupled mode. the manuscript. We also thank Shiling Peng, John Fyfe, The ®rst coupled mode of variability between the David En®eld, and the two anonymous reviewers for ocean temperature and the atmospheric pressure can be their help and suggestions. This work was supported by described as a strengthening and weakening of the sub- research grants awarded to LAM and DNS from the tropical anticyclone, which seems to force ¯uctuations Canadian Natural Sciences and Engineering Research in a north±south dipole structure in the ocean temper- Council and Fonds FCAR (Quebec.) ature by wind-related processes. The atmospheric forcing of the SST changes is detectable in the ocean with a lag of 1±2 months. The observed ¯uctuations are dom- REFERENCES inated by interdecadal timescales (14±16-yr period). The atmosphere-ocean coupling is found to be strongest Battisti, D. S., U. S. Bhatt, and M. A. Alexander, 1995: A modeling study of the interannual variability in the wintertime North At- during the southern summer. A possible relationship be- lantic Ocean. J. Climate, 8, 3067±3083. tween this South Atlantic mode and interdecadal oscil- Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercom- lations of Northern Hemisphere SST and SLP is also parison of methods for ®nding coupled patterns in climate data. found. The possibility of an atmosphere±ocean feedback J. Climate, 5, 541±560. in the coupling depicted by this mode is suggested. Cayan, D. R., 1992a: Latent and sensible heat ¯ux anomalies over the northern oceans: The connection to monthly atmospheric The second mode of coupled atmosphere±ocean vari- circulation. J. Climate, 5, 354±369. ability in the South Atlantic is characterized by east± , 1992b: Latent and sensible heat ¯ux anomalies over the northwest displacements of the subtropical anticyclone center, ern oceans: Driving the sea surface temperature. J. Phys. Ocean- accompanied by large SST ¯uctuations over a broad ogr., 22, 859±881. Davis, R. E., 1976: Predictability of sea surface temperature and sea region off the coast of Africa, which occur on a 6±7-yr level pressure anomalies over the North Paci®c Ocean. J. Phys. timescale. This mode seems to depict an immediate re- Oceanogr., 6, 249±266. sponse of the atmospheric circulation to changes in the Delworth, T., S. Manabe, and R. J. Stouffer, 1993: Interdecadal vari- ocean, although the coupling between the SST and SLP ations of the thermohaline circulation in a coupled ocean±at- ¯uctuations is rather weak. A signi®cant frequency-do- mosphere model. J. Climate, 6, 1993±2001. Deser, C., and M. L. Blackmon, 1993: Surface climate variations over main correlation with the NAO index indicates that sim- the North Atlantic Ocean during winter: 1900±1989. J. Climate, ilar timescales characterize the ¯uctuations associated 6, 1743±1753. with these South and North Atlantic modes of vari- En®eld, D. B., and D. A. Mayer, 1997: Tropical Atlantic SST variability. ability and its relation to El NinÄo±Southern Oscillation. J. Geo- phys. Res., 102(C1), 929±945. The third South Atlantic mode is characterized by Fang, Z., and J. M. Wallace, 1994: Arctic sea ice variability on a north±south displacements of the subtropical anticy- timescale of weeks and its relation to atmospheric forcing. J. clone, accompanied (with a lag of 1±2 months) by SST Climate, 7, 1897±1913.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC 2920 JOURNAL OF CLIMATE VOLUME 10

Folland, C. K., D. E. Parker, and F. E. Kates, 1984: Worldwide marine Arctic and Greenland±Iceland sea and their relation to an inter- temperature ¯uctuations 1856±1981. Nature, 310, 670±673. decadal climate cycle. Climatol. Bull., 26(3), 147±176. Frankignoul, C., 1985: Sea surface temperature anomalies, planetary Newman, M., and P. D. Sardeshmukh, 1995: A caveat concerning waves, and air-sea feedback in the middle latitudes. Rev. Geo- singular value decomposition. J. Climate, 8, 352±360. phys., 23(4), 357±390. North, G. R., T. L. Bell, R. F. Cahalan, and F. J. Moeng, 1982: Sam- Ghil, M., and R. Vautard, 1991: Interdecadal oscillations and the pling error in the estimation of empirical orthogonal functions. warming trend in global temperature time series. Nature, 350, Mon. Wea. Rev., 110, 699±706. 324±327. Palmer, T. N., and Z. Sun, 1985: A modelling and observational study Halliwell, G. R., 1996: Atmospheric circulation anomalies driving of the relationship between sea surface temperature in the north- decadal/interdecadal SST anomalies in the North Atlantic. ACCP west Atlantic and the atmospheric general circulation. Quart. J. Notes, III(1), 8±11. Roy. Meteor. Soc., 111, 947±975. Horel, J. D., and J. M. Wallace, 1981: Planetary scale atmospheric Peixoto, J. P., and A. H. Oort, 1992: Physics of Climate. American phenomena associated with the interannual variability of sea Institute of Physics, 520 pp. surface temperature in the equatorial Paci®c. Mon. Wea. Rev., Peng, S., and J. Fyfe, 1996: The coupled patterns between sea level 109, 813±829. pressure and sea surface temperature in the midlatitude North Houghton, R., and Y. Tourre, 1992: Characteristics of low-frequency Atlantic. J. Climate, 9, 1824±1839. sea surface temperature ¯uctuations in the tropical Atlantic. J. , L. A. Mysak, H. Ritchie, J. Derome, and B. Dougas, 1995: On Climate, 5, 765±771. the difference between early and middle winter atmospheric re- Kushnir, Y., 1994: Interdecadal variations in North Atlantic sea sur- sponses to sea surface temperature anomalies in the northwest face temperature and associated atmospheric conditions. J. Cli- Atlantic. J. Climate, 8, 137±157. mate, 7, 141±157. Peterson, R. G., and L. Stramma, 1991: Upper-level circulation in , and N. C. Lau, 1992: The general circulation model response the South Atlantic Ocean. Progress in Oceanography, Vol. 26, to a North Paci®c SST anomaly: Dependence on time scale and Pergamon Press, 1±73. pattern polarity. J. Climate, 5, 271±283. Prohaska, J. T., 1976: A technique for analyzing the linear relation- Lanzante, J. R., 1984: A rotated eigenanalysis of the correlation be- ships between two meteorological ®elds. Mon. Wea. Rev., 104, tween 700 mb heights and sea surface temperature in the Paci®c 1345±1353. and Atlantic. Mon. Wea. Rev., 112, 2270±2280. Rogers, J. C., 1984: The association between the North Atlantic Os- Latif, M., and T. P. Barnett, 1994: Causes of decadal climate vari- cillation and the Southern Oscillation in the Northern Hemi- ability over the North Paci®c and North America. Science, 266, sphere. Mon. Wea. Rev., 112, 1999±2015. 634±637. Sciremammano, F., 1979: A suggestion for the presentation of cor- , and , 1996: Decadal climate variability over the North relations and their signi®cance levels. J. Phys. Oceanogr., 9, Paci®c and North America: Dynamics and predictability. J. Cli- 1273±1276. mate, 9, 2407±2423. Stocker, T. F., and L. A. Mysak, 1992: Climatic ¯uctuations on the Levitus, S., 1989: Interpentadal variability of temperature and salinity century time scale: A review of high-resolution proxy data and at intermediate depths of the North Atlantic Ocean: 1970±1974 possible mechanisms. Climate Change, 20, 227±250. versus 1955±1959. J. Geophys. Res., 94, 6091±6131. Thiebaux, H. J., and M. A. Pedder, 1987: Spatial Objective Analysis Manabe, S., and R. J. Stouffer, 1994: Multiple-century response of with Applications in Atmospheric Science. Academic Press, 299 a coupled ocean±atmosphere model to an increase of atmospheric pp. carbon dioxide. J. Climate, 7, 5±23. Thomson, D. J., 1982: Spectrum estimation and harmonic analysis. Mann, M. E., and J. Park, 1993: Spatial correlations of interdecadal IEEE Proc., 70, 1055±1096. variation in global surface temperatures. Geophys. Res. Lett., 20, Venegas, S. A., L. A. Mysak, and D. N. Straub, 1996: Evidence for 1055±1058. interannual and interdecadal climate variability in the South At- , and , 1994: Global-scale modes of surface temperature lantic. Geophys. Res. Lett., 23(19), 2673±2676. variability on interannual to century timescales. J. Geophys. von Storch, H., and A. Navarra, Eds., 1995: Analyses of Climate Res., 99, 25 819±25 833. VariabilityÐApplications of Statistical Techinques. Springer, , and J. Lees, 1996: Robust estimation of background noise and 334 pp. signal detection in climatic time series. Climate Change, 33, Wallace, J. M., and D. S. Gutzler, 1981: Teleconnections in the geo- 409±445. potential height ®eld during the Northern Hemisphere winter. , and J. Park, 1996: Joint spatiotemporal modes of surface tem- Mon. Wea. Rev., 109, 784±812. perature and sea level pressure variability in the Northern Hemi- , and Q. R. Jiang, 1987: On the observed structure of the inter- sphere during the last century. J. Climate, 9, 2137±2162. annual variability of the atmosphere±ocean climate system. At- Mehta, V. M., and T. Delworth, 1995: Decadal variability of the mospheric and Oceanic Variability, H. Cattle, Ed., Royal Me- tropical Atlantic Ocean surface temperature in shipboard mea- teorological Society, 17±43. surements and in a global ocean±atmosphere model. J. Climate, , C. Smith, and Q. R. Jiang, 1990: Spatial patterns of atmosphere± 8, 172±190. ocean interaction in the northern winter. J. Climate, 3, 990±998. Miller, A. J., D. R. Cayan, T. P. Barnett, N. E. Graham, and J. M. , , and C. S. Bretherton, 1992: Singular value decomposition Oberhuber, 1994: Interdecadal variability of the Paci®c Ocean: of wintertime sea surface temperature and 500-mb height anom- Model response to observed heat ¯ux and wind stress anomalies. alies. J. Climate, 5, 561±576. Climate Dyn., 9, 287±302. Woodruff, S. D., R. J. Slutz, R. L. Jenne, and P. M. Steurer, 1987: Molinari, R. L., D. Battisti, K. Bryan, and J. Walsh, 1994: The At- A Comprehensive Ocean±Atmosphere Data Set. Bull. Amer. Me- lantic Climate Change Program. Bull. Amer. Meteor. Soc., 75, teor. Soc., 68, 521±527. 1191±1199. Zebiak, S. E., 1993: Air±sea interaction in the equatorial Atlantic Mysak, L. A., and S. B. Power, 1992: Sea-ice anomalies in the western region. J. Climate, 6, 1567±1586.

Unauthenticated | Downloaded 10/02/21 10:40 AM UTC