Abstract

In this paper, the modulations of the North Atlantic tripole (NAT) decadal variability from global warming are studied by conducting a series of coupled ocean–atmosphere experiments using the Fast Ocean Atmosphere Model (FOAM). The model reasonably captures the observed NAT decadal variability with a preferred time scale of about 11 years. With the aid of partial-blocking and partial-coupling experiments, it is found that the NAT decadal cycle can be attributed to oceanic planetary wave adjustment in the subtropical basin and ocean–atmosphere coupling over the North Atlantic.

In a doubled CO2 experiment, the spatial pattern of the NAT is preserved; however, the decadal cycle is significantly suppressed. This suppression appears to be associated with the acceleration of oceanic planetary waves due to an increase of buoyancy frequency in global warming. This shortens the time from a decadal to an interannual time scale for the first-mode baroclinic Rossby waves to cross the subtropical North Atlantic basin, the primary memory for the NAT decadal variability in the model. The modeling study also found that the global warming does not modulate the North Atlantic air–sea coupling significantly, but it may be model dependent.

1. Introduction

Observational study shows that the majority of the total heat flux of the earth system over the last 40 years has gone into the oceans (Levitus et al. 2001). This implies that the oceans as a major heat sink of the earth have been slowing down the warming trend substantially. The elevation of greenhouse gases concentration not only leads to the formation of distinct regional sea surface temperature (SST) warming patterns, but may also cause regime shifts of climate variability modes. So far, studies have been mainly focusing on the response of interannual variability modes to global warming, for instance, changes in patterns and amplitudes of ENSO [see a recent review by Collins et al. (2010)] and the Indian Ocean dipole (IOD; Saji et al. 1999) under global warming (e.g., Vecchi and Soden 2007; Abram et al. 2008; Ihara et al. 2008). In this paper, we will investigate the impacts of global warming on low frequency, in particular the decadal variations of SST over the North Atlantic.

SST anomalies over the North Atlantic exhibit two prominent and recurrent patterns: a tripole and a monopole pattern (e.g., Deser and Blackmon 1993). The tripole pattern [known as the North Atlantic tripole (NAT)] displays three distinct centers respectively located in the subpolar, western subtropical and eastern tropical North Atlantic (e.g., Marshall et al. 2001a; Visbeck et al. 2003; Wu and Liu 2005). While the monopole pattern is associated with the North Atlantic meridional overturning circulation and dominated by multidecadal variability, the NAT is closely related to the North Atlantic Oscillation (NAO), a fluctuation of sea level pressure (SLP) difference between the subtropical Atlantic and the Arctic (see reviews by Marshall et al. 2001b; Hurrell et al. 2003; Czaja et al. 2003).

The NAT displays distinct interannual and decadal variations (Deser and Blackmon 1993; Kushnir 1994). At interannual time scales, the NAO variability intrinsic to the atmosphere drives large-scale SST variability through changes of surface heat flux as well as Ekman flow (e.g., Bjerknes 1964; Battisti et al. 1995; Delworth 1996; Marshall et al. 2001a; Visbeck et al. 2003). In addition to interannual variability, both observations and climate models indicate that the NAT displays significant decadal variability with a time scale of 10–20 yr (e.g., Deser and Blackmon 1993; Molinari et al. 1997; Grötzner et al. 1998; Wu and Liu 2005; Fan and Schneider 2012).

Several mechanisms have been proposed for the formation of the decadal NAT. One school of thought has suggested that SST decadal variations may arise from atmospheric internal stochastic forcing in conjunction with the mean flow advection or intrinsic Rossby wave dynamics in the ocean (Frankignoul et al. 1997; Saravanan and McWilliams 1998; Wu and Liu 2003). The propagation of coherent SST fluctuations following the mean gyre circulation has been revealed in both observations (e.g., Sutton and Allen 1997) and coupled GCMs (e.g., Grötzner et al. 1998), which may manifest the effects of the mean advection and/or anomalous advection due to adjustment of the ocean to the anomalous wind stress forcing (e.g., Curry and McCartney 2001; Frankignoul et al. 2001; Marshall et al. 2001b; Visbeck et al. 2003). On the other hand, a mechanism similar to that put forward by Latif and Barnett (1994) for the North Pacific, which includes a delayed negative feedback of oceanic gyre adjustment and a positive feedback of ocean–atmosphere coupling, has been suggested for the decadal variability in the North Atlantic (Grötzner et al. 1998; Wu and Liu 2005), although this mechanism is not apparent in other coupled GCMs (e.g., Zorita and Frankignoul 1997; Frankignoul et al. 2001). In particular, a coupled modeling study by Wu and Liu (2005) explicitly demonstrates a critical role of coupling between the NAT and NAO in sustaining the decadal cycle of the NAT detected in their coupled model.

So far, modeling studies have examined the potential shifts of the NAO induced by rising of the greenhouse gases in the atmosphere. An analysis of the second Coupled Model Intercomparison Project (CMIP2) models appears to demonstrate an increasing trend in the NAO index, but the magnitude of the trend differs substantially from model to model and is generally weaker than the observed (e.g., Stephenson et al. 2006). Furthermore, none of these models were able to simulate sufficient decadal variability (Stephenson et al. 2006). In this study, we will focus on the response of the decadal NAT to carbon dioxide (CO2) increase using a coupled ocean–atmosphere general circulation model, and show that the global warming will suppress decadal variability of both the NAT and NAO. This suppression is likely associated with acceleration of oceanic planetary Rossby wave due to an increase of buoyancy frequency in global warming.

The paper is arranged as follows. Section 2 gives a brief introduction to the model and data used in this study. Sections 3 and 4 describe the model-simulated NAT in the control and doubled CO2 (2CO2) experiment. Section 5 demonstrates the mechanisms of the NAT and some possible mechanisms responsible for the change of NAT in global warming. The paper concludes with a summary and some further discussion.

2. Model and data

The model we used is the Fast Ocean Atmosphere Model (FOAM), a fully coupled ocean–atmosphere general circulation model (Jacob 1997). The atmospheric component of FOAM is a fully parallel version of the National Center for Atmospheric Research (NCAR) Community Climate Model version 2 (CCM2), in which the atmospheric physics are replaced by those of NCAR CCM3. The atmosphere model has a R15 resolution, but with a vertical resolution of 18 levels, which is normally set for T42 resolution. The ocean model was developed following the Geophysical Fluid Dynamics Laboratory (GFDL) Modular Ocean Model (MOM), with a resolution of 1.4° latitude × 2.8° longitude × 32 levels. Without flux adjustment, the fully coupled model has been integrated for over 2000 years without apparent climate shift.

FOAM reasonably captures features of the observed climatology (Jacob 1997). Over the North Atlantic, the mean total horizontal mass transport in the model is about 40 Sv (1 Sv ≡ 106 m3 s−1) for the subtropical gyre and 20 Sv for the subpolar gyre, somewhat weaker than the observed because of a coarse OGCM resolution. Similar to many CGCMs, the model fails to capture right locations of the warm pool and cold tongue in the tropical Atlantic, and also exhibits considerable errors in SST over the Gulf Stream region, due to a too diffusive western boundary current in the model. The model generates a stable meridional overturning circulation in the Atlantic, although the magnitude is slightly stronger than that of the observed (Wu et al. 2008). In spite of these model biases, FOAM reasonable captures tropical Atlantic variability (Wu et al. 2007), NAO, and North Atlantic decadal climate variability (Wu and Liu 2005). It should be noted that the version used in this study is basically the same as that used in Wu and Liu (2005), but with vertical resolution of the ocean model increased from 24 to 32 levels and some other changes in the ice and land model components.

The FOAM control experiment (Ctrl) has been integrated for over 2000 years with greenhouse gases concentrations fixed at 1990 level, which is about 355 ppm. In the doubled CO2 run (2CO2), the model is integrated for 800 years with CO2 fixed at 710 ppm. The last 400 years in both the Ctrl and 2CO2 experiment are used for analysis.

The monthly mean observational SST data used in this study is from the Hadley Centre Global Sea Ice and Sea Surface Temperature dataset (HadISST; Rayner et al. 2003). It consists of data from 1870 to 2010, with resolution of 1.0° latitude × 1.0° longitude. The monthly mean geopotential height data are from the National Centers for Environmental Prediction (NCEP)–NCAR reanalysis from 1948 to 2010.

We also use outputs of the GFDL Climate Model version 2.0 (CM2.0) preindustrial and A1B (720 ppm) stabilization experiments (2100–2300) to verify results derived from FOAM simulations. In this paper, we focus on cold season from October to March because the variability in both the ocean and atmosphere is stronger than other seasons. All the data have been detrended before analysis.

3. Model-simulated NAT

Although the model-simulated decadal variability of the NAT has been documented in Wu and Liu (2005), we will reexamine it first due to a change of vertical resolution of the oceanic component. To validate the model simulations, the observations are also analyzed for a comparison.

In the observation (HadISST), the NAT emerges as the second EOF mode of wintertime SST anomalies over the North Atlantic (north of 20°N), which accounts for 19% of total variances (Fig. 1a). The temporal behavior of the NAT demonstrates interannual to decadal variability (Fig. 1c). Enhanced power can be seen around the decadal time scale (∼10 yr), although the spectrum is not significant because of a short period of the records. Associated with the NAT mode, the variations of the atmospheric circulation reveal a dipole pattern, which is obtained by regressing the 500-mb geopotential height from the NCEP–NCAR reanalysis against the NAT index (defined as the principal component of the NAT mode) from1948 to 2010 (Fig. 1e). This pattern resembles the leading EOF mode of the wintertime geopotential height over the North Atlantic sector, namely the NAO (Fig. 2a).

Fig. 1.

(left) Observed and (right) model-simulated leading modes of North Atlantic (north of 20°N) winter (October–March) anomalies. (a),(b) NAT SST mode; (c),(d) power spectrum of NAT time series (principal component of the corresponding EOF mode); and (e),(f) regression of the 500-mb geopotential height anomalies against the principal component. Units for SST EOFs and Z500 regression are °C and gpm, respectively. Cpy denotes cycles per year.

Fig. 1.

(left) Observed and (right) model-simulated leading modes of North Atlantic (north of 20°N) winter (October–March) anomalies. (a),(b) NAT SST mode; (c),(d) power spectrum of NAT time series (principal component of the corresponding EOF mode); and (e),(f) regression of the 500-mb geopotential height anomalies against the principal component. Units for SST EOFs and Z500 regression are °C and gpm, respectively. Cpy denotes cycles per year.

Fig. 2.

The first EOF mode of (a) observed and (b) model-simulated 500-mb geopotential height anomalies in winter (October–March). (c) Power spectrum of the time series of the Z500 mb EOF1 in the model. Units for Z500 EOFs and frequency are gpm and cpy, respectively.

Fig. 2.

The first EOF mode of (a) observed and (b) model-simulated 500-mb geopotential height anomalies in winter (October–March). (c) Power spectrum of the time series of the Z500 mb EOF1 in the model. Units for Z500 EOFs and frequency are gpm and cpy, respectively.

The leading EOF mode of the winter SST anomalies over the North Atlantic in the FOAM control experiment features a tripole pattern, which accounts for 23% of total variances (Fig. 1b). The pattern is broadly similar to the observed, but differences are also notable (Fig. 1a). Compared to the observed pattern, the model-simulated NAT has weaker amplitudes in the subpolar region and an eastward displacement of the subtropical lobe. The weaker subpolar lobe is likely due to an unrealistically deeper mixed layer depth in that region. The simulated NAT displays a significant spectral peak around 11 yr (Fig. 1d), which appears to be much stronger than observations (Deser and Blackmon 1993; Molinari et al. 1997).

The regression of 500-mb geopotential height against the NAT index exhibits a similar pattern as the observed (Fig. 1f). In the model, a NAT with cooling in the subtropics and warming in the subpolar region and eastern subtropics is associated with a weaker-than-normal Icelandic low and subtropical high. Compared with the observed, the center of the Icelandic low somewhat shifts to the west, resembling the leading EOF mode of the 500-mb geopotential height in the model (Fig. 2b). This EOF mode, manifesting the NAO in the model, displays significant variations at a decadal time scale with a frequency also around 11 yr (Fig. 2c). This frequency coherence between the NAT and the NAO suggests an important role of ocean–atmosphere coupling in decadal NAT variations.

4. NAT in global warming

a. Mean state

With a doubled CO2 concentration, the mean SST response over the Atlantic shows a general warming pattern, but with remarkable regional differences (Fig. 3a). Over the North Atlantic basin, the SST increases with a maximum of 1.8°C over the Gulf Stream region and displays a remarkable cooling over the subpolar North Atlantic. The decrease of the warming toward the high-latitude ocean may be associated with a weakening of deep convection over the subpolar ocean in a warm climate. The SST pattern simulated here resembles that simulated in models for the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) report (e.g., Xie et al. 2010). The warm anomalies also penetrate into the subsurface ocean, with more substantial warming in the subsurface (Fig. 3b). The maximum warming appears around 600 m in the subpolar region, which extends to lower latitudes with a greater depth. In addition, most of the water becomes saltier, with the stronger salinity anomalies in the subsurface (Fig. 3c). The salinity change in the subpolar region shows opposite polarity with the freshening in the surface and salinification in the subsurface. The unique warming and salinification pattern in the North Atlantic is likely associated with a decrease of deep convection in the subpolar region due to surface freshening and advection of the meridional overturning circulation. Global warming also exerts significant impacts on winds. In the warm climate, there forms a westerly anomaly over 45°–50°N, implying a strengthening and northward shift of the mean westerlies (Fig. 3d). The enhancement of the westerly is associated with a positive NAO-like atmospheric response (not shown).

Fig. 3.

Anomalies (2CO2 − Ctrl) of (a) SST, depth-latitudinal section of (b) temperature and (c) salinity over the Atlantic (0°–60°N), and (d) wind. Units for temperature, salinity, and wind are °C, practical salinity unit (psu), and m s−1, respectively.

Fig. 3.

Anomalies (2CO2 − Ctrl) of (a) SST, depth-latitudinal section of (b) temperature and (c) salinity over the Atlantic (0°–60°N), and (d) wind. Units for temperature, salinity, and wind are °C, practical salinity unit (psu), and m s−1, respectively.

b. NAT

The first EOF mode of the North Atlantic winter SST anomalies in the 2CO2 experiment shows the NAT resembling that in the Ctrl experiment (Fig. 4a). The mode explains about 24% of the total variance, and exhibits a well-defined tripole structure. The lobe in the subpolar region now appears to be stronger than that in the control experiment because of a shallower mixed layer in a warm climate. The regression of 500-mb geopotential height against the NAT index also displays a NAO-like pattern similar to that in the control experiment (not shown), suggesting that NAT–NAO coupled pattern preserves in warm climate.

Fig. 4.

Leading EOF of winter (October–March) (a) SST and (c) Z500 anomalies in 2CO2 experiment. Power spectrum of (b) SST PC1 and (d) Z500 PC1. Units for SST and Z500 EOFs are °C and gpm, respectively.

Fig. 4.

Leading EOF of winter (October–March) (a) SST and (c) Z500 anomalies in 2CO2 experiment. Power spectrum of (b) SST PC1 and (d) Z500 PC1. Units for SST and Z500 EOFs are °C and gpm, respectively.

The most striking changes of the NAT–NAO mode occur in their temporal evolution. In the 2CO2 experiment, the NAT does not show a preferred decadal time scale, in a sharp contrast to the control run where a significant decadal peak is identified (Fig. 4b vs 1d). Similar to the NAT, the NAO in the 2CO2 run displays a pattern similar to that in Ctrl experiment (Fig. 4c), but with decadal variability suppressed (Fig. 4d vs 2c). There is a tendency for the preferred time scale to shift from decadal to interannual band for both the NAT and the NAO in a warm climate, although the statistical significance is not so strong. For example, there are weak spectral peaks around 8 and 4 yr for both the NAO and NAT mode (Figs. 4b,d).

In the following, we will investigate dynamic processes responsible for the suppression of the decadal variability in warm climate.

5. Mechanisms

a. Role of ocean dynamics in decadal NAT

To demonstrate how decadal SST anomalies are generated over the North Atlantic, a heat budget analysis is conducted for the upper-100-m ocean in the control simulation. Various terms in the heat balance include surface heat flux, anomalous advection, mean advection, and dissipation (diffusion and convection). All of these terms are averaged over the upper 100 m and are then regressed against the NAT index.

The heat budget analysis reveals that the NAT is primarily associated with the anomalous meridional advection, surface heat flux, and dissipation (Fig. 5). Specifically, in the subpolar Atlantic, warm (cold) anomalies are created predominantly by the anomalous meridional advection and surface heat flux (Fig. 5a), in response to the anomalous easterlies (westerlies) (not shown), which reduce (enhance) oceanic latent heat loss and bring warm (cold) water from lower (higher) latitudes to sustain the warm (cold) SST anomalies through anomalous Ekman advection. The vertical mixing and diffusion play a damping role (Fig. 5a). It is interesting to see that SST in the subpolar region reaches the maximum with about 1-yr lag, seemingly indicating a passive response of the subpolar SST.

Fig. 5.

Lagged regression of different heat budget terms averaged over the (a) subpolar (45°–60°N, 30°–80°W), (b) western subtropical (25°–45°N, 30°–80°W), and (c) eastern subtropical(5°–25°N, 10°–50°W) poles of the NAT against the NAT index. Thick red curve with stars and thin black curve with asterisks, thin red and blue curves, thin green curve with plus signs and thin pink curve, thick green curve with circles, thick blue curve with diamonds, and thick pink curve with squares demonstrate anomalous and mean meridional advection, anomalous and mean zonal advection, anomalous and mean vertical advection, SST, diffusion, and heat flux, respectively. All data are bandpassed to retain variability between 6 and 20 yr. Units are W m−2 for heat budget term and °C for SST. Uppercase and lowercase letters represent climatology and anomaly, respectively.

Fig. 5.

Lagged regression of different heat budget terms averaged over the (a) subpolar (45°–60°N, 30°–80°W), (b) western subtropical (25°–45°N, 30°–80°W), and (c) eastern subtropical(5°–25°N, 10°–50°W) poles of the NAT against the NAT index. Thick red curve with stars and thin black curve with asterisks, thin red and blue curves, thin green curve with plus signs and thin pink curve, thick green curve with circles, thick blue curve with diamonds, and thick pink curve with squares demonstrate anomalous and mean meridional advection, anomalous and mean zonal advection, anomalous and mean vertical advection, SST, diffusion, and heat flux, respectively. All data are bandpassed to retain variability between 6 and 20 yr. Units are W m−2 for heat budget term and °C for SST. Uppercase and lowercase letters represent climatology and anomaly, respectively.

In the western subtropical Atlantic (Fig. 5b), warm (cold) SST anomalies are primarily created by anomalous meridional advection and are damped by vertical diffusion and convection. The former tends to be caused by the weakening (strengthening) of the subtropical gyre in response to the interior anomalous positive (negative) wind stress curl. The mean advection plays a less significant role in SST. In the eastern subtropical Atlantic (Fig. 5c), warm (cold) anomalies are created by the surface heat flux and anomalous upwelling (downwelling). The former is associated with the anomalous southeast (northeast) trades, which reduce (enhance) latent heat loss, while the latter tends to be associated with the negative (positive) wind stress curl as a result of the equatorward diminishing of the anomalous northeast (southwest) trades. The SST anomalies are further substantiated by vertical diffusion and mixing. Overall, the decadal SST anomalies are primarily associated with anomalous meridional advection, which is presumably induced by wind-forced Rossby wave adjustment. In the following, we will further demonstrate the role of Rossby wave adjustment in the decadal NAT.

To illustrate the decadal cycle of the NAT in the Ctrl experiment, we calculated the lagged regressions of the SST, upper-250-m heat content (HC), and wind stress curl anomalies against the NAT index (Fig. 6). To focus on decadal time scale, all variables are bandpassed to retain variability between 6 and 20 yr.

Fig. 6.

Lagged regression of (left) SST, (middle) wind stress curl, and (right) upper-250 m heat content against the NAT index in the Ctrl experiment. All data are bandpassed to retain variability between 6 and 20 yr. The contour interval is 0.01°C, 2 × 10−8 N m−3, and 1 m °C−1 for SST, wind stress curl, and heat content, respectively.

Fig. 6.

Lagged regression of (left) SST, (middle) wind stress curl, and (right) upper-250 m heat content against the NAT index in the Ctrl experiment. All data are bandpassed to retain variability between 6 and 20 yr. The contour interval is 0.01°C, 2 × 10−8 N m−3, and 1 m °C−1 for SST, wind stress curl, and heat content, respectively.

At the NAT mature phase (Fig. 6a3), the surface wind stress curl anomaly has a negative polarity in both subpolar and tropical North Atlantic and a positive one in the subtropics (Fig. 6b3), leading to a shoaling of thermocline in the subtropics and a deepening in both flanks (Fig. 6c3). The cold HC or thermocline anomalies in the subtropics move toward the west, accompanied by poleward propagation along the North African coast and then westward propagation of the north tropical Atlantic warm thermocline anomalies (Fig. 6c4). After reaching the western boundary, the cold anomalies move equatorward along the American coast following the coastal Kelvin wave (Fig. 6c5) and then eastward following the equatorial Kelvin wave (Fig. 6c6). Meanwhile, the cold subsurface anomalies upwell to the surface in the tropical North Atlantic where mean upwelling prevails, to generate cold anomalies (Figs. 6a5–6). The cold thermocline anomalies also subsequently propagate poleward along the North African coast following the coastal Kelvin and subsequently disperse into the interior through Rossby wave (Figs. 6c6, 6c1–c2). The wave dynamic processes displayed here reflect a delayed adjustment of the subtropical gyre forced by the wind stress curl dipole at the mature phase (Fig. 6b3). The delayed adjustment time scale is primarily associated with the westward propagation of the first baroclinic Rossby wave within 30°–40°N (Figs. 6c1–6). A westward propagation can be also seen for both SST (Figs. 6a1–6) and wind stress curl (Figs. 6b1–7) anomalies, implying a coupled interaction between the ocean and atmosphere in sustaining the NAT decadal cycle. A westward propagation of HC anomalies can be also found in the subpolar region. The Rossby–Kelvin wave mechanism identified here is essentially similar to what was found in Wu and Liu (2005), highlighting the important role of oceanic wave dynamics in setting the preferred time scale of the decadal NAT. Here the preferred time scale is likely determined by the cross-basin time scale of the baroclinic Rossby wave in the subtropics.

To further explore the role of subtropical Rossby wave in the decadal NAT, a partial-blocking (PB; Wu and Liu 2003) experiment is performed. This experiment is configured the same as the Ctrl, except that the temperature and salinity are fixed to the model climatology from 100 m to the ocean floor in the region (5°–40°N, 35°–45°W) at each time step, so that temperature and salinity anomalies are damped within this blocking “wall” (Fig. 7). This type of idealized setting can presumably cut off the westward propagation of the Rossby wave from the east of the subtropical basin. However, the wind-forced Rossby wave adjustment and geostrophic current remains in the west of the blocking wall. Also, there is a possibility that the blocking wall may create temperature and salinity fronts and radiate Rossby waves, although we have added buffer zones at each side of the blocking wall to minimize the temperature and salinity jumps. The PB experiment has been run for 500 years, with the last 400 years used for analysis. A similar experiment has been performed by Zhong and Liu (2009) using CCSM3.

Fig. 7.

Schematic figure of the partial-blocking experiment. The shaded strip denotes the sponge wall (5°–40°N, 35°–45°W) in which temperature and salinity are fixed to the model climatology from 100 m to the ocean floor in the region.

Fig. 7.

Schematic figure of the partial-blocking experiment. The shaded strip denotes the sponge wall (5°–40°N, 35°–45°W) in which temperature and salinity are fixed to the model climatology from 100 m to the ocean floor in the region.

The first EOF mode of the winter North Atlantic SST also features a tripole pattern, consistent with the Ctrl experiment, suggesting the wave dynamics is not necessary to maintain the NAT pattern (Fig. 8a). Notable deviations can be found in the magnitudes where both the eastern and the western subtropical anomalies are reduced while the subpolar anomaly is somewhat strengthened. Yet, the most striking change is the frequency of the NAT. As shown in Fig. 1, the NAT in the Ctrl experiment displays a rather strong decadal variability, while the NAT in the PB experiment shows no dominant cycle at decadal time scale (Fig. 8b). This readily suggests the critical role of subtropical Rossby wave dynamics in establishing the decadal cycle of the NAT. The time for the Rossby wave to cross the subtropical North Atlantic basin is about 5–6 yr, equivalent to half of the decadal cycle. This is also consistent with Münnich et al. (1998) that the cycle of decadal oscillation can be roughly diagnosed as twice the time for the Rossby wave to cross the basin.

Fig. 8.

As in Fig. 4, but for the PB experiment.

Fig. 8.

As in Fig. 4, but for the PB experiment.

The important role of subtropical Rossby waves in determining the decadal time scale can be also reflected in the atmosphere. In the PB experiment, the pattern of the NAO remains similar to that in control experiment (Fig. 8c). However, the distinct 11-yr cycle of the NAO seen in the Ctrl experiment disappears (Fig. 8d), indicating an oceanic control of the NAO decadal variability.

The subtropical Rossby wave adjustment provides the necessary memory for the NAT decadal time scale. Next, we will assess the role of the ocean–atmosphere coupling in the decadal cycle of the NAT, which has been shown to be a necessary ingredient in maintaining the decadal NAT from previous modeling studies (Grötzner et al. 1998; Wu and Liu 2005).To explore the role of ocean–atmosphere coupling in the decadal NAT, a partial coupling (PC; Wu and Liu 2003) experiment is performed. In this experiment, the North Atlantic Ocean (20°–70°N, 100°W–0°) is decoupled with the atmosphere by prescribing the model climatological SST to force the atmosphere, while the ocean is fully coupled with the atmosphere in other regions. In the PC experiment, variability in the ocean and atmosphere is generated predominantly by the atmospheric internal stochastic forcing.

Without the ocean–atmosphere coupling, both the NAT and the NAO remain robust and display similar pattern as those in the Ctrl experiment (Figs. 9a,c). This indicates that the ocean–atmosphere coupling does not play a crucial role in generating the NAT, which is consistent with previous studies (e.g., Battisti et al. 1995; Delworth 1996; Wu and Liu 2005). Compared with the Ctrl pattern, the magnitudes of the NAT in PC are enhanced over the northern center by 33% and decreased over both the eastern and the western centers by 14% and 25%, respectively.

Fig. 9.

As in Fig. 4, but for the PC experiment.

Fig. 9.

As in Fig. 4, but for the PC experiment.

The most prominent change in the PC experiment, however, occurs on the decadal evolution of both the NAT and the NAO. Without the ocean–atmosphere coupling, the decadal cycle disappears for both NAT and NAO (Figs. 9b,d), in a sharp contrast to that in Ctrl. These vanished decadal oscillations suggest the significant role of the ocean–atmosphere coupling in sustaining the North Atlantic decadal variability. It is noted that although the 11-yr decadal cycle is suppressed in the PC experiment, the multidecadal variability becomes more significant (Fig. 9b), implying an oceanic origin presumably associated with Atlantic meridional overturning circulation. Here we mainly focus on the NAT decadal variability.

In summary, the mechanism of the NAT decadal cycle is closely associated with both oceanic dynamics and ocean–atmosphere coupling. Therefore, it is natural to wonder whether the suppression of the NAT decadal variability in the warm climate is associated with modulation of Rossby wave dynamics and air–sea coupling from global warming.

b. Modulation of Rossby wave dynamics by global warming

One important oceanic change in global warming is the stratification. In a warm climate, the surface ocean warms faster than the subsurface, leading to a strengthening of the oceanic stratification. However, this is different in the North Atlantic. In the 2CO2 experiment, the subsurface warms more than the surface because of the advanced meridional overturning circulation (AMOC) weakening (Fig. 3b). Yet, the salinity in the subsurface (Fig. 3c) also increases because of the AMOC weakening, which leads to an increase of the stratification and thus the buoyancy frequency, most notably within the thermocline (Fig. 10). This should stabilize the ocean.

Fig. 10.

Vertical profile of Brunt–Väisälä frequency averaged over the North Atlantic Ocean for Ctrl, 2CO2 experiments, and their differences. Unit for Brunt–Väisälä frequency is s−1.

Fig. 10.

Vertical profile of Brunt–Väisälä frequency averaged over the North Atlantic Ocean for Ctrl, 2CO2 experiments, and their differences. Unit for Brunt–Väisälä frequency is s−1.

Without consideration of the mean flow, the Rossby wave speed can be expressed as (e.g., Chelton et al. 1998)

 
formula

where f is the Coriolis parameter, β = ∂f/∂y, λm is the Rossby radius of deformation, N(z) is the buoyancy frequency, and m is the order of the baroclinic Rossby wave. The formula indicates that the speed of Rossby wave is determined by latitude and buoyancy frequency (i.e., the Rossby wave is faster in low latitudes and with strong stratification).

Here we will focus on the first-mode baroclinic Rossby wave, which is related to the NAT decadal variability. Figure 11 features the speed of the first-mode baroclinic Rossby wave in the Ctrl and 2CO2 experiment, respectively. The wave speed in the Ctrl is consistent with Killworth et al. (1997) that the speed decreases toward high latitudes.

Fig. 11.

Latitudinal distribution of zonal-averaged first baroclinic Rossby wave speed over the North Atlantic in the FOAM Ctrl and 2CO2 experiment. The changing rate of the wave speed is defined as (2CO2 − Ctrl)/Ctrl. Unit for Rossby wave speed is m s−1.

Fig. 11.

Latitudinal distribution of zonal-averaged first baroclinic Rossby wave speed over the North Atlantic in the FOAM Ctrl and 2CO2 experiment. The changing rate of the wave speed is defined as (2CO2 − Ctrl)/Ctrl. Unit for Rossby wave speed is m s−1.

In 2CO2 experiment, the first-mode baroclinic Rossby wave is accelerated at all latitudes with more pronounced acceleration in lower latitudes (Fig. 11). However, the changing ratio relative to the Ctrl remains similar, with a magnitude of about 20%–30% in the subtropical basin. Given the important role of the subtropical Rossby wave in setting up the time scale of the NAT decadal variability, the acceleration of the Rossby wave in the subtropics by global warming, therefore, leads to a shortage of decadal memory in the subtropical ocean. An increase of 30% in the first-mode baroclinic Rossby wave speed may lead to a reduction of the cycle from 11 to about 8 yr, consistent with the frequency shift of the NAT spectrum (Fig. 4).

c. Modulation of NAO–NAT coupling by global warming

In addition to the modulation of oceanic planetary Rossby waves, the ocean–atmosphere coupling over the North Atlantic may also be modulated in a warm climate, which has been shown to be another critical ingredient of the NAT decadal variability. Studies have suggested that the ocean-to-atmosphere feedback in the tropics is weakened because of an increase of static stability of the troposphere from global warming (e.g., Zheng et al. 2010). However, in the midlatitudes, it remains elusive because of the strong nonlinear interaction between synoptic eddies, the stationary wave, and the jet stream (Kushnir et al. 2002).

To investigate potential modulation of air–sea coupling by global warming, we use the maximum covariance analysis (MCA), the same method used in the study of Czaja and Frankignoul (2002). This technique is based on the singular value decomposition (SVD) of the temporal covariance matrix between SST and an atmospheric variable (e.g., 500-mb geopotential height at a given lag). Both variables are weighted by the square root of the cosine of latitude and normalized by the corresponding mean (domain averaged) seasonal cycle of standard deviation. The MCA is useful in statistically distinguishing between cause and effect. If the ocean responds passively to atmospheric stochastic forcing, there should be no covariance when SST leads by more than the atmospheric persistence time; and if the SST can influence the atmosphere, their cross covariance does not vanish when the SST leads.

The domain for the MCA analysis covers the North Atlantic north of 20°N. In the model, the squared covariance (SC) between monthly SST and 500-mb geopotential height anomalies indicates a dominant forcing of the atmosphere in all seasons (Fig. 12a), with peak when atmosphere leads by 1 month. This suggests that the NAT is mainly a response to the variability of NAO. The results here are consistent with Czaja and Frankignoul (2002), except they used 3-month mean data. However, significant SC is also found in November–January with Z500 anomalies correlated with SST of 3–7 months ahead. It is also noted that Z500 anomalies in January also correlates with SST anomalies in previous winter, being indicative of reemergence of SST anomalies (Cassou et al. 2007). The result in the model is broadly similar to that in observations, although the season of ocean-to-atmosphere feedback is shifted and the strength is also slightly weaker (Fig. 12b). Also the impacts of SST reemergence on the atmospheric circulation are not present.

Fig. 12.

Seasonal variance of squared covariance of the first MCA mode of midlatitude (20°–70°N) SST and Z500 anomalies in the (a) NCEP–NCAR reanalysis, (b) FOAM Ctrl experiment, and (c) FOAM 2CO2experiment. The shaded area indicates where the covariance is statistically significant at the 80%, 85%, and 95% level (light to dark shading, respectively). The x axis denotes the months of Z500, and the y axis indicates SST lags Z500 in months (positive when Z500 leads and negative when SST leads).

Fig. 12.

Seasonal variance of squared covariance of the first MCA mode of midlatitude (20°–70°N) SST and Z500 anomalies in the (a) NCEP–NCAR reanalysis, (b) FOAM Ctrl experiment, and (c) FOAM 2CO2experiment. The shaded area indicates where the covariance is statistically significant at the 80%, 85%, and 95% level (light to dark shading, respectively). The x axis denotes the months of Z500, and the y axis indicates SST lags Z500 in months (positive when Z500 leads and negative when SST leads).

The air–sea coupling appears to be not significantly modulated by global warming (Fig. 12c vs 12b). The signal denoting the ocean-to-atmosphere feedback remains robust in November and December in spite of a decrease of the leads of SST by about 1 month. In February, the ocean-to-atmosphere feedback appears to be weakened substantially. This weakening may help suppressing the coupled NAT–NAO variability. Overall, given the insignificant modulation of air–sea coupling from global warming in the North Atlantic, it is therefore concluded that the suppression of the NAT decadal variability in 2CO2 may be mainly associated with acceleration of oceanic planetary waves from the enhanced stratification by global warming.

6. Summary and discussion

In this paper, the modulations of the NAT decadal variability from global warming are studied by conducting a series of coupled ocean–atmosphere experiments. The model reasonably captures the observed North Atlantic decadal variability with a preferred time scale of about 11 yr. With the aid of partial-blocking and partial-coupling experiments, it is found that the NAT decadal cycle may be attributed to oceanic planetary wave adjustment in the subtropical basin and ocean–atmosphere coupling.

In 2CO2 experiment, while the spatial pattern of the NAT preserves, the decadal cycle is significantly suppressed. This suppression is likely associated with the acceleration of oceanic planetary wave due to an increase of buoyancy frequency in global warming. This shortens the cross-basin time for the first-mode baroclinic Rossby waves, the primary memory for the NAT decadal variability in the model. It is also found that the global warming does not modulate the North Atlantic NAT–NAO coupling significantly.

This acceleration, however, can be also found in GFDL CM2.0 model (Fig. 13). To verify that, we contrast the speed of the first-mode baroclinic Rossby wave in the preindustrial and A1B stabilization experiment. Consistent with FOAM, the acceleration of the first-mode baroclinic Rossby waves in A1B is robust, with a magnitude of about 30% increase in the subtropical basin. This is even more significant than that in FOAM 2CO2 experiment. The difference in the magnitude is because the CO2 concentration in A1B stabilization experiment (∼720 ppm) is about 3 times of that in the preindustrial experiment (∼278 ppm).

Fig. 13.

Latitudinal distribution of zonal-averaged first-mode baroclinic Rossby wave speed over the North Atlantic in the GFDL CM2.0 preindustrial and the A1B stabilization experiment.

Fig. 13.

Latitudinal distribution of zonal-averaged first-mode baroclinic Rossby wave speed over the North Atlantic in the GFDL CM2.0 preindustrial and the A1B stabilization experiment.

Our modeling studies suggest that the global warming will not significantly modulate the air–sea coupling over the North Atlantic. Given the highly nonlinear interaction of eddies, mean flow, and stationary waves in the midlatitude, this conclusion may be model dependent. Here we further analyze the modulation of midlatitude air–sea coupling from global warming in GFDL CM2.0 using the MCA method. In the preindustrial run, the model displays much more significant oceanic influences on the atmosphere from November to May than the observation and FOAM model (Fig. 14a). In the A1B experiment, these influences however, virtually disappear (Fig. 14b). Although the detailed mechanisms responsible for the weakening of the air–sea coupling over the North Atlantic in global warming are complicated and beyond the scope of this paper, it is conceivable that the acceleration of oceanic planetary wave speed and the weakening of air–sea coupling from global warming should collaboratively suppress the NAT decadal variability in a warm climate. Indeed, the 11-yr decadal cycle of the NAT can be also identified in GFDL CM2.0 preindustrial run (not shown). Although this 11-yr cycle in A1B stabilization period (2100–2300) is weakened, we may not be able to attribute to the impacts of global warming with confidence because of a short integration of the experiment. A further model intercomparison will be helpful to validate the modeling results here.

Fig. 14.

As in Fig. 12, but for GFDL CM2.0 (a) preindustrial and (b) A1B stabilization experiment.

Fig. 14.

As in Fig. 12, but for GFDL CM2.0 (a) preindustrial and (b) A1B stabilization experiment.

It should be noted that although we highlight the roles of the baroclinic Rossby wave acceleration in suppressing NAT decadal variability in warm climate, the decadal NAT may be affected by other processes including ocean gyres, thermohaline circulation, as well as air–sea coupling (e.g., Marshall et al. 2001a).

In our study, the time scale of the thermohaline circulation is much slower than and well separated from the NAT; therefore, its contribution to the NAT decadal variations may not be significant. In reality, the thermohaline circulation may be coupled with the wind-driven circulation. Recent studies based on climate model simulations used for the IPCC AR4 assessment indicate an increasing trend of NAO variability in a warm climate (e.g., Stephenson et al. 2006), which is also found in our model simulation (not shown). The amplification of the NAO wind stress may increase the efficiencies of the gyre heat transport to change the NAT temperature anomalies and reduce the damping of SST through an intensification of the surface Ekman advection, favoring a more unstable NAT (Marshall et al. 2001a). It is therefore conceivable that the global warming may suppress the decadal but enhance interannual variations of the NAT. Another potential effect of the NAO is the potential shift of its action center in global warming (e.g., Hu and Wu 2004), which can affect the adjustment time scale of gyre circulation by westward propagation of Rossby waves (Jin 1997; Neelin and Weng 1999). In our model, the center of the wind stress forcing in the subtropics associated with NAO is located slightly to the west in the doubled CO2 simulation relative to the control simulation (not shown), which may also shorten the time for Rossby waves to reach the western boundary. Nevertheless, current climate models capture the increasing trend of the NAO in warm climate; however, the simulated amplitude of the trend is generally weaker and model dependent. This may underestimate the climatic shift of the NAT in global warming. As for the air–sea coupling, it remains an open question whether the global warming will enhance the coupling between the NAO and the NAT, because the coupling effect may depend on the ocean mixed layer depth, the atmospheric stationary wave, and storm tracks, etc.

One should also keep in mind that our model has weakness in simulating the subpolar pole of the NAT in the control experiment, which may be associated with the unrealistically deep mixed layer in the model. Also, the sharp 11-yr peak in the power spectrum of atmospheric variability does not occur in the previous observational studies. Therefore, there is a potential model-dependent issue for the proposed Rossby wave mechanism in the modulation of the NAT decadal variability from global warming. A multimodel intercomparison study will help with addressing these issues.

Acknowledgments

This work is supported by China National Natural Science Foundation (NSFC) Distinguished Young Investigator Project (40788002), NSFC Innovation Team Project (40921004). Discussions with Drs. Edwin Schneider, Ping Chang, and R. Saravanan were helpful. We appreciate comments from three anonymous reviewers and the editor, Dr. Michael Alexander, which improved the paper substantially.

REFERENCES

REFERENCES
Abram
,
N. J.
,
M. K.
Gagan
,
J. E.
Cole
,
W. S.
Hantoro
, and
M.
Mudelsee
,
2008
:
Recent intensification of tropical climate variability in the Indian Ocean
.
Nat. Geosci.
,
1
,
849
853
,
doi:10.1038/ngeo357
.
Battisti
,
D. S.
,
U. S.
Bhatt
, and
M. A.
Alexander
,
1995
:
A modeling study of the interannual variability in the wintertime North Atlantic Ocean
.
J. Climate
,
8
,
3067
3083
.
Bjerknes
,
J.
,
1964
:
Atlantic air–sea interaction
.
Advances in Geophysics, Vol. 10, Academic Press, 1–82
.
Cassou
,
C.
,
C.
Deser
, and
M. A.
Alexander
,
2007
:
Investigating the impact of reemerging sea surface temperature anomalies on the winter atmospheric circulation over the North Atlantic
.
J. Climate
,
20
,
3510
3526
.
Chelton
,
D. B.
,
R. A.
deSzoeke
,
M. G.
Schlax
,
K.
El Naggar
, and
N.
Siwertz
,
1998
:
Geographical variability of the first baroclinic Rossby radius of deformation
.
J. Phys. Oceanogr.
,
28
,
433
460
.
Collins
,
M.
, and
Coauthors
,
2010
:
The impact of global warming on the tropical Pacific Ocean and El Niño
.
Nat. Geosci.
,
3
,
391
397
.
Curry
,
R. G.
, and
M. S.
McCartney
,
2001
:
Ocean gyre circulation changes associated with the North Atlantic Oscillation
.
J. Phys. Oceanogr.
,
31
,
3374
3400
.
Czaja
,
A.
, and
C.
Frankignoul
,
2002
:
Observed impact of Atlantic SST anomalies on the North Atlantic Oscillation
.
J. Climate
,
15
,
606
623
.
Czaja
,
A.
,
A. W.
Robertson
, and
T.
Huck
,
2003
:
The role of Atlantic ocean–atmosphere coupling in affecting North Atlantic Oscillation variability
.
The North Atlantic Oscillation: Climatic Significance and Environment Impact, Geophys. Monogr.,
Vol. 134, Amer. Geophys. Union, 147–172
.
Delworth
,
T.
,
1996
:
North Atlantic interannual variability in a coupled ocean–atmosphere model
.
J. Climate
,
9
,
2356
2375
.
Deser
,
C.
, and
M. L.
Blackmon
,
1993
:
Surface climate variations over the North Atlantic Ocean during winter: 1900–93
.
J. Climate
,
6
,
1743
1753
.
Fan
,
M.
, and
E.
Schneider
,
2012
:
Observed decadal North Atlantic tripole SST variability. Part I: Weather noise forcing and coupled response
.
J. Atmos. Sci.
,
69
,
35
50
.
Frankignoul
,
C.
,
P.
Muller
, and
E.
Zorita
,
1997
:
A simple model of the decadal response of the ocean to stochastic wind forcing
.
J. Phys. Oceanogr.
,
27
,
1533
1546
.
Frankignoul
,
C.
,
E.
Kestenare
,
N.
Sennéchael
,
G.
de Coëtlogon
, and
F.
D’Andréa
,
2001
:
On decadal-scale ocean–atmosphere interactions in the extended ECHAM1/LSG climate simulation
.
Climate Dyn.
,
16
,
333
354
.
Grötzner
,
A.
,
M.
Latif
, and
T. P.
Barnett
,
1998
:
A decadal climate cycle in the North Atlantic Ocean as simulated by the ECHO coupled GCM
.
J. Climate
,
11
,
831
847
.
Hu
,
Z.-Z.
, and
Z.
Wu
,
2004
:
The intensification and shift of the annual North Atlantic Oscillation in a global warming scenario simulation
,
Tellus
,
56A
,
112
124
.
Hurrell
,
J. W.
,
Y.
Kushnir
,
G.
Ottersen
, and
M.
Visbeck
,
2003
:
An overview of the North Atlantic Oscillation
.
The North Atlantic Oscillation: Climatic Significance and Environmental Impact, Geophys. Monogr., Vol. 134, Amer. Geophys. Union, 1–35
.
Ihara
,
C.
,
Y.
Kushnir
, and
M. A.
Cane
,
2008
:
Warming trend of the Indian Ocean SST and Indian Ocean Dipole from 1880 to 2004
.
J. Climate
,
21
,
2035
2046
.
Jacob
,
R. L.
,
1997
:
Low frequency variability in a simulated atmosphere ocean system
.
Ph.D. thesis, University of Wisconsin—Madison, 155 pp
.
Jin
,
F.-F.
,
1997
:
A theory of interdecadal climate variability of the North Pacific ocean–atmosphere system
.
J. Climate
,
10
,
1821
1835
.
Killworth
,
P. D.
,
D. B.
Chelton
, and
R. A.
DeSzoeke
,
1997
:
The speed of observed and theoretical long extratropical planetary waves
.
J. Phys. Oceanogr.
,
27
,
1946
1966
.
Kushnir
,
Y.
,
1994
:
Interdecadal variations in North Atlantic sea surface temperature and associated atmospheric conditions
.
J. Climate
,
7
,
142
157
.
Kushnir
,
Y.
,
W. A.
Robinson
,
I.
Blade
,
N. M. J.
Hall
,
S.
Peng
, and
R.
Sutton
,
2002
:
Atmospheric GCM response to extratropical SST anomalies: Synthesis and evaluation
.
J. Climate
,
15
,
2233
2256
.
Latif
,
M.
, and
T. P.
Barnett
,
1994
:
Causes of decadal climate variability in the North Pacific/North Atlantic sector
.
Science
,
266
,
634
637
.
Levitus
,
S.
,
J. I.
Antonov
,
J.
Wang
,
T. L.
Delworth
,
K. W.
Dixon
, and
A. J.
Broccoli
,
2001
:
Anthropogenic warming of Earth’s climate system
.
Science
,
292
,
267
270
.
Marshall
,
J.
,
H.
Johnson
, and
J.
Goodman
,
2001a
:
A study of the interaction of the North Atlantic Oscillation with the ocean circulation
.
J. Climate
,
14
,
1399
1421
.
Marshall
,
J.
, and
Coauthors
,
2001b
:
North Atlantic climate variability: Phenomena, impacts and mechanisms
.
Int. J. Climatol.
,
21
,
1863
1898
.
Molinari
,
R. L.
,
D. A.
Mayer
,
J. F.
Festa
, and
H. F.
Bezdek
,
1997
:
Multiyear variability in the near-surface temperature structure of the midlatitude western North Atlantic Ocean
.
J. Geophys. Res.
,
102
,
3267
3278
.
Münnich
,
M.
,
M.
Latif
,
S.
Venzke
, and
E.
Maier-Reimer
,
1998
:
Decadal oscillations in a simple coupled model
.
J. Climate
,
11
,
3309
3319
.
Neelin
,
J. D.
, and
W.
Weng
,
1999
:
Analytical prototypes for ocean–atmosphere interaction at midlatitudes. Part I: Coupled feedbacks as a sea surface temperature dependent stochastic process
.
J. Climate
,
12
,
697
721
.
Rayner
,
N. A.
,
D. E.
Parker
,
E. B.
Horton
,
C. K.
Folland
,
L. V.
Alexander
,
D. P.
Rowell
,
E. C.
Kent
, and
A.
Kaplan
,
2003
:
Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century
.
J. Geophys. Res.
,
108
,
4407
,
doi:10.1029/2002JD002670
.
Saji
,
N. H.
,
B. N.
Goswami
,
P. N.
Vinayachndran
, and
T.
Yamagata
,
1999
:
A dipole mode in the tropical Indian Ocean
.
Nature
,
401
,
360
363
.
Saravanan
,
R.
, and
J. C.
McWilliams
,
1998
:
Advective ocean–atmosphere interaction: An analytical stochastic model with implications for decadal variability
.
J. Climate
,
11
,
165
188
.
Stephenson
,
D. B.
,
V.
Pavan
,
M.
Collins
,
M. M.
Junge
,
R.
Quadreli
, and
Participating CMIP2 Modeling Group
,
2006
:
North Atlantic Oscillation response to transient greenhouse gas forcing and the impact on European winter climate: A CMIP2 multi-model assessment
.
Climate Dyn.
,
27
,
401
420
.
Sutton
,
R. T.
, and
M. R.
Allen
,
1997
:
Decadal predictability of North Atlantic sea surface temperature and climate
.
Nature
,
388
,
563
567
.
Vecchi
,
G. A.
, and
B. J.
Soden
,
2007
:
Global warming and the weakening of the tropical circulation
.
J. Climate
,
20
,
4316
4340
.
Visbeck
,
M.
,
E.
Chassignet
,
R.
Curry
,
T.
Delworth
,
B.
Dickson
, and
G.
Rahmann
,
2003
:
The ocean’s response to NAO variability
.
The North Atlantic Oscillation: Climatic Significance and Environment Impact, Geophys. Monogr., Vol. 134, Amer. Geophys. Union,
113
146
.
Wu
,
L.
, and
Z.
Liu
,
2003
:
Decadal variability in the North Pacific: The eastern North Pacific mode
.
J. Climate
,
16
,
3111
3131
.
Wu
,
L.
, and
Z.
Liu
,
2005
:
North Atlantic decadal variability: Air–sea coupling, oceanic memory, and potential Northern Hemisphere resonance
.
J. Climate
,
18
,
331
349
.
Wu
,
L.
,
F.
He
,
Z.
Liu
, and
C.
Li
,
2007
:
Atmospheric teleconnections of tropical Atlantic variability: Interhemispheric, tropical–extratropical, and cross-basin interactions
.
J. Climate
,
20
,
856
870
.
Wu
,
L.
,
C.
Li
,
C.
Yang
, and
S.
Xie
,
2008
:
Global teleconnections in response to a shutdown of the Atlantic Meridional Overturning Circulation
.
J. Climate
,
21
,
3002
3019
.
Xie
,
S.
,
C.
Deser
,
G. A.
Vecchi
,
J.
Ma
,
H.
Teng
, and
A. T.
Wittenberg
,
2010
:
Global warming pattern formation: Sea surface temperature and rainfall
.
J. Climate
,
23
,
966
986
.
Zheng
,
X.-T.
,
S.-P.
Xie
,
G.
Vecchi
,
Q.
Liu
, and
J.
Hafner
,
2010
:
Indian Ocean dipole response to global warming: Analysis of ocean–atmospheric feedbacks in a coupled model
.
J. Climate
,
23
,
1240
1253
.
Zhong
,
Y.
, and
Z.
Liu
,
2009
:
On the mechanism of Pacific multidecadal climate variability in CCSM3: The role of the subpolar North Pacific Ocean
.
J. Phys. Oceanogr.
,
39
,
2052
2076
.
Zorita
,
E.
, and
C.
Frankignoul
,
1997
:
Modes of North Atlantic decadal variability in the ECHAM1/LSG coupled atmosphere–ocean general circulation model
.
J. Climate
,
10
,
183
200
.