Abstract

Experiments with the Whole Atmosphere Community Climate Model (WACCM) are used to understand the influence of the stratospheric tropical quasi-biennial oscillation (QBO) in the troposphere. The zonally symmetric circulation in thermal wind balance with the QBO affects high-frequency eddies throughout the extratropical troposphere. The influence of the QBO is strongest and most robust in the North Pacific near the jet exit region, in agreement with observations. Variability of the stratospheric polar vortex does not appear to explain the effect of the QBO in the troposphere in the model, although it does contribute to the response in the North Atlantic. Anomalies in tropical deep convection associated with the QBO appear to damp, rather than drive, the effect of the QBO in the extratropical troposphere. Rather, the crucial mechanism whereby the QBO modulates the extratropical troposphere appears to be the interaction of tropospheric transient waves with the axisymmetric circulation in thermal wind balance with the QBO. The response to QBO winds of realistic amplitude is stronger for perpetual February radiative conditions and sea surface temperatures than perpetual January conditions, consistent with the observed response in reanalysis data, in a coupled seasonal WACCM integration, and in dry model experiments described in Part I.

1. Introduction

Stratospheric anomalies have been shown to affect the wintertime tropospheric circulation. In particular, variability of the stratospheric polar vortex is linked with the northern annular mode (NAM) (Baldwin and Dunkerton 1999; Polvani and Kushner 2002; Limpasuvan et al. 2004). Even though the equatorial stratospheric quasi-biennial oscillation (QBO) is the dominant mode of interannual stratospheric variability in the tropics, its effect on the wintertime tropospheric circulation has been less thoroughly investigated.

Crooks and Gray (2005), Coughlin and Tung (2005), and Haigh et al. (2005) find that the anomalous QBO winds seemingly curve downward in a horseshoe-shaped pattern into the subtropical troposphere, with wind anomalies of opposite sign in the deep tropics and extratropics [see also Fig. 6 of Giorgetta et al. (1999), Fig. 6 of Garfinkel and Hartmann (2010), and Fig. 2 in Part I of this paper (Garfinkel and Hartmann 2011, hereafter Part I)]. The signal is especially strong during February and March in the North Pacific (NP). Figure 1 shows the difference in 300-hPa zonal winds between easterly QBO (EQBO) and westerly QBO (WQBO) composites in the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis1 and in a coupled Whole Atmosphere Community Climate Model (WACCM) integration. The QBO appears to have a robust influence in the troposphere in both, whereby lower stratospheric easterlies lead to a weaker subtropical tropospheric jet in the NP. How this signal is communicated downward through the tropopause, however, has not been fully established.

Fig. 1.

Zonal wind anomalies at 300 hPa associated with the QBO (EQBO − WQBO) during February and March in (a) the ECMWF reanalysis data and (b) a coupled seasonal WACCM run. See section 2 of Part I for a description of the two data sources. Months in which the wind anomalies at 70 hPa exceed 2 m s−1 are used to define composites of EQBO and WQBO. There are 45 WQBO months and 30 EQBO months in the reanalysis composites, and 101 WQBO months and 73 EQBO months in the WACCM composites. Contours are shown at ±0.5, ±1.5, ±3, ±5, and ±8 m s−1. Significant regions at 95% are shaded and negative contours are thick.

Fig. 1.

Zonal wind anomalies at 300 hPa associated with the QBO (EQBO − WQBO) during February and March in (a) the ECMWF reanalysis data and (b) a coupled seasonal WACCM run. See section 2 of Part I for a description of the two data sources. Months in which the wind anomalies at 70 hPa exceed 2 m s−1 are used to define composites of EQBO and WQBO. There are 45 WQBO months and 30 EQBO months in the reanalysis composites, and 101 WQBO months and 73 EQBO months in the WACCM composites. Contours are shown at ±0.5, ±1.5, ±3, ±5, and ±8 m s−1. Significant regions at 95% are shaded and negative contours are thick.

We consider three possible mechanisms. The first mechanism is that once the anomalous QBO winds influence the strength of the stratospheric polar vortex (Holton and Tan 1980; Baldwin et al. 2001; Coughlin and Tung 2001; Ruzmaikin et al. 2005; Marshall and Scaife 2009), it can then affect the troposphere just like any polar vortex anomaly. While this vortex mechanism is likely important for the response in the North Atlantic, Garfinkel and Hartmann (2010) find that the response to the QBO in the NP is qualitatively different from the influence of the polar vortex in the NP. Some additional mechanism must be present for a complete explanation of the effect of the QBO in the troposphere.

In the second and third mechanisms considered, the meridional circulation that maintains thermal wind balance with QBO equatorial stratospheric wind anomalies interacts with the troposphere. See section 4 of Part I and references therein for a full discussion of this meridional circulation. Briefly, this circulation includes 1) upwelling and a cold temperature anomaly near the equator and subsidence and warm temperature anomalies near 30° in the lowermost stratosphere, and 2) arching of QBO wind anomalies downward to the troposphere and poleward to near 20°. Even though this meridional circulation penetrates the tropopause only weakly, this circulation can affect tropospheric variability in (at least) two ways: either through tropical convection or through extratropical eddies.

If the meridional circulation of the QBO can modulate tropical deep convection (Gray et al. 1992; Giorgetta et al. 1999; Collimore et al. 2003), it can affect the Hadley circulation and the extratropics as well. Ho et al. (2009) suggest that such a change in tropical convection affects NP summertime tropical cyclone tracks. Demonstrating a connection between anomalies in convection and the QBO in winter in the observational record is difficult, however. Over the past three decades, when reliable satellite observations of outgoing longwave radiation (OLR) are available, westerlies in the lowermost equatorial stratosphere and El Niño have tended to coincide (Garfinkel and Hartmann 2007).2 The difference in OLR between composites of WQBO and EQBO events therefore resembles the difference between El Niño and La Niña (not shown). It is not clear whether the QBO would modulate wintertime tropical convection, and thermally driven subtropical jets, under fixed sea surface temperatures (SSTs). Furthermore, Part I showed that the QBO can influence the troposphere in a model that lacks realistic polar vortex variability and convection, implying that these are not the sole pathways through which the QBO can influence the troposphere.

Finally, the meridional circulation of the QBO can influence extratropical tropospheric eddies directly and thereby bias the troposphere toward its leading modes of variability. Part I showed that in the presence of tropospheric variability, eddies amplify the zonal wind anomalies that would exist in the absence of eddies. Even though the results of Part I are consistent with observations, the simplified model used there does not include convection or vortex variability. We therefore explore the tropospheric response to the QBO in a more realistic model configuration.

The primary tools used are perpetual January and February WACCM runs with fixed SSTs. WACCM runs with a neutral QBO stratospheric wind profile are compared to runs with an easterly QBO stratospheric wind profile. Because the SSTs are fixed, our model runs cleanly isolate the influence that the QBO may have on parameterized convection. Because our model also has a variable polar vortex, the influence of the QBO on the troposphere through its effect on the vortex is simulated as well. By understanding how the QBO affects the troposphere in these model runs, the mechanism(s) through which the QBO affects the troposphere in observations can be elucidated.

Section 2 introduces the model used and section 3 discusses some aspects of the control run. Section 4 discusses the response to imposing EQBO stratospheric winds and the role of high-frequency eddies for the amplifying the tropospheric response. Section 4a shows that convection and vortex variability do not explain the effect of the QBO in the extratropical troposphere, and section 4b shows that the model results are robust to model configuration. Section 5 presents conclusions.

2. The model

Perpetual January and February simulations are used to investigate the response to QBO stratospheric wind anomalies. WACCM version 3.1.9 (Marsh et al. 2007; Garcia et al. 2007; Richter et al. 2008) is run with fixed SSTs, land surface, and ice, perpetual 15 January or 5 February radiative forcing, the Zhang and McFarlane (1995) convection scheme as implemented in CAM3 (Collins et al. 2006), and with interactive chemistry turned off. The horizontal resolution is 4° latitude × 5° longitude, and the model has 66 vertical levels (a hybrid sigma vertical coordinate) extending into the thermosphere.

a. QBO relaxation

The QBO wind relaxation is as in Matthes et al. (2010; see references therein). Between model layers 0.0026 and 0.085 65 (approximately 2.6 and 85.7 hPa), the winds at the equator are linearly relaxed toward the specified profile with a 10-day time scale. At model level 0.1005 (approximately 100.5 hPa), the relaxation time scale at the equator is 20 days. Away from the equator, the linear relaxation time scale increases as days, where φ is latitude and τeq is the relaxation time scale at the equator. Winds evolve freely poleward of 22° latitude.

Figure 2 shows the QBO profiles relaxed toward. The default EQBO wind profile (solid line in Fig. 2) is similar to, although slightly stronger than, a typical EQBO wind profile, and is the same wind profile used in Garfinkel and Hartmann (2010) and Part I. Unlike real EQBO wind profiles, the profile relaxes to climatological easterlies in the upper stratosphere. Section 4b shows that sensitivity to the upper stratospheric wind anomalies is weak. To demonstrate the realism of the magnitude of the EQBO wind anomalies, asterisks in Fig. 2 denote the 5%–95% range of variability of equatorial zonal wind from May 1953 to April 2007; our EQBO profile is within the natural variability of the QBO. Huesmann and Hitchman (2001) and Collimore et al. (2003) define an event by a 1.5 m s−1 shear per 20 hPa between 50 and 70 hPa. The difference in the zonal wind relaxed toward between these two levels is 11.3 m s−1. Our lower stratospheric shear is stronger than a typical event in Huesmann and Hitchman (2001) but still within the natural variability [see Fig. 6 of Huesmann and Hitchman (2001)].

Fig. 2.

The QBO profiles relaxed toward in the WACCM runs. The 5%–95% range of variability of equatorial zonal wind from May 1953 to April 2007 is shown (stars).

Fig. 2.

The QBO profiles relaxed toward in the WACCM runs. The 5%–95% range of variability of equatorial zonal wind from May 1953 to April 2007 is shown (stars).

b. Methodology

The runs are initialized by first running WACCM with the full seasonal cycle for at least one year. Perpetual 15 January (or 5 February) conditions are then imposed, and equatorial stratospheric winds are relaxed to the climatological equatorial stratospheric winds from May 1953 to April 2007 (dashes in Fig. 2). The model integration is then continued for at least an additional 235 months, of which the first 10 months are discarded. These constitute our January and February control runs, denoted JCONT and FCONT on Table 1.

Table 1.

Different runs for understanding the tropospheric signal of the QBO. For the branch runs, the first number of “run length” denotes the number of ensemble members, each of which extends for four months. For all other runs, the run length denotes the number of months.

Different runs for understanding the tropospheric signal of the QBO. For the branch runs, the first number of “run length” denotes the number of ensemble members, each of which extends for four months. For all other runs, the run length denotes the number of months.
Different runs for understanding the tropospheric signal of the QBO. For the branch runs, the first number of “run length” denotes the number of ensemble members, each of which extends for four months. For all other runs, the run length denotes the number of months.

Two types of EQBO runs are explored. In the first type we 1) branch off the instantaneous atmospheric state at the beginning of each month of the control runs, 2) impose an EQBO wind profile, and then 3) integrate each ensemble member for an additional 120 days. Because the relaxation time scale for the QBO winds is no faster than 10 days, the atmosphere can smoothly adjust to the EQBO equatorial stratospheric profile. We thus generate a large ensemble of the transient NP response to EQBO winds. We focus on the transient response rather than the equilibrated steady-state response 1) because the transient response allows us to investigate the causality more cleanly, and 2) because the response to the QBO in observations will never reach equilibrium because of the seasonal cycle, implying that the transient response is more relevant to the observed response. Two ensembles were created. The first is in perpetual February with a 3 times stronger EQBO (hereafter 3×EQBO) profile, denoted FBRANCH (as in ensemble of branch runs in February) in Table 1. The second is in perpetual January with the default EQBO profile (JBRANCH in Table 1). We focus on the FBRANCH ensemble because our diagnostics are cleaner in this case. Only the segment of the control run common to the ensemble of branch runs is used.

The second type of runs are long quasi-steady equilibrium runs in which the EQBO relaxation is always present (see Table 1). The runs are used to explore sensitivity of our results to the EQBO profile chosen and to the choice of perpetual February as opposed to perpetual January. In the first run, the 3×EQBO profile is used. In the second, the default EQBO profile is used; in the third, the EQBO profile includes upper stratospheric westerlies (triangles in Fig. 2). In the fourth, lower stratospheric winds are relaxed toward a weak EQBO profile, while winds below model level σ = 0.0615 are relaxed toward the neutral QBO profile; the resulting QBO perturbation is weaker than many observed QBO events and localized to the lower stratosphere (pluses on Fig. 2). In the fifth, we attempt to mimic the internally generated QBO in some recent climate models (e.g., Giorgetta et al. 2006; Anstey et al. 2010; Kulyamin et al. 2009; Kawatani et al. 2010; Osprey et al. 2010; Hurwitz et al. 2011). In many of these models the QBO winds are too weak or do not extend strongly enough below 50 hPa. To mimic such a QBO, we relax to a weak sheared EQBO profile only above model level σ = 0.0615 (diamonds in Fig. 2).

Any robust differences between the control case and the EQBO case in the midlatitudes or in the troposphere are part of the response to the EQBO relaxation rather than a result of the relaxation itself. Because the QBO relaxation method here and in Part I is identical, the meridional circulation discussed at length in section 4 of Part I is assumed to exist in WACCM. This assumption will be verified in section 4. Significance is determined by a two-tailed Student t difference of means test.

c. Vorticity budget

Part I found that transient eddies are crucial for amplifying the tropospheric response to the QBO. It is therefore important to diagnose, in our model runs, the role of transient eddies in providing the vorticity fluxes that amplify the tropospheric response and extend it downward to the surface. Barnes and Hartmann (2010a, hereafter BH10) and Barnes and Hartmann (2010b) show that eddy feedbacks in zonally confined regions can be diagnosed quantitatively by using the vorticity budget:

 
formula

Here, f is the Coriolis parameter, ω is the pressure velocity, ζ is the relative vorticity, u is the horizontal component of the velocity, and is the forcing due to friction or the QBO relaxation. BH10 argue that when the vorticity forcing terms in upper levels project onto vorticity anomalies in lower levels, lower-level vorticity anomalies can be maintained against damping. The barotropization of the anomaly in the troposphere follows Held (1975) and Edmon et al. (1980). See BH10 for more details.

The vorticity budget is used here to diagnose the role of transient tropospheric eddies in the tropospheric response to the QBO. The terms in the vorticity budget are computed for the FCONT run and FBRANCH ensemble, with the mean state from the FCONT run used for both. High-frequency eddies are separated by a ninth-order Butterworth filter with a 7-day cutoff, as in Part I. The difference in the forcing terms between FCONT and the FBRANCH ensemble diagnoses the generation of zonally asymmetric tropospheric vorticity anomalies in response to EQBO winds.

3. Mean state and variability of the control runs

Before we analyze the response to QBO winds in our model, we first explore the fidelity of the mean state and variability of the control runs. Figure 3 compares the time-mean zonally averaged zonal wind and temperature of the JCONT runs to the January mean state in the ECMWF reanalysis. In the Northern Hemisphere (NH), jets below 25 km are in the correct location and have the correct magnitude in the JCONT (and FCONT) runs. In the Southern Hemisphere (SH), however, disagreements are larger. The SH stratosphere has easterlies everywhere in the reanalysis, but in the JCONT run a weak vortex is present near the pole. The polar lowermost SH stratosphere is too cold, and the SH tropospheric jet is too strong, in JCONT as compared to the reanalysis. Biases in FCONT are even larger than biases in JCONT. The lack of fidelity between the SH mean state of the WACCM runs and the reanalysis indicates that the results for the SH should be regarded with caution, particularly if the stratospheric summer easterlies are regarded as important. The direct downward effect in the tropics is likely not strongly affected by these biases, however.

Fig. 3.

Average zonal-mean zonal wind and temperature in the reanalysis in January and in the JCONT run. Contours shown at ±5, ±15, ±25, ±35, ±45, ±55, and ±65 m s−1 and every 10 K. Shading marks regions where the mean state during February (FCONT) differs from that during January (JCONT) by 1 K or 2 m s−1.

Fig. 3.

Average zonal-mean zonal wind and temperature in the reanalysis in January and in the JCONT run. Contours shown at ±5, ±15, ±25, ±35, ±45, ±55, and ±65 m s−1 and every 10 K. Shading marks regions where the mean state during February (FCONT) differs from that during January (JCONT) by 1 K or 2 m s−1.

Although the NH mean state of the WACCM is similar to observations, WACCM v3.1.9 has too little stratospheric polar vortex variability and too few sudden stratospheric warmings. The standard deviation of temperature area averaged poleward from 70°N and vertically from 70 to 150 hPa (i.e., lower stratospheric polar cap temperature) is 3.5 K in JCONT, while it is 4.2 K from 1958 to 2007 in the ECMWF reanalysis. A similar bias is present in FCONT. Such reduced variability might bias our conclusions as to whether the QBO can influence the troposphere via the vortex. To show that this is likely not the case, we compare the influence of the QBO on the lowermost stratosphere in the reanalysis to the WACCM runs. The difference in polar cap temperature between our equilibrium EQBO January runs and JCONT is 1.5 K, which is comparable to the difference between a reanalysis EQBO composite and climatology over these pressure levels during winter. Changes in EP flux are consistent with linear theory (Garfinkel et al. 2011, manuscript submitted to J. Atmos. Sci.). Only higher in the stratosphere is the influence of the QBO on polar cap temperatures greater in the reanalysis than in our WACCM runs. As the influence of polar vortex variability on the troposphere is communicated through the lower stratosphere, we expect that our WACCM runs capture reasonably well the ability of the QBO to influence the troposphere by first influencing the vortex.

Part I found that QBO winds incline a tropospheric jet toward one phase of its dominant mode of variability, but that the direction that the jet shifts depends on properties of the mean state and modes of variability of the jet. In particular, a jet whose dominant mode of variability is north–south shifting shifts poleward in response to EQBO winds, with the poleward shift being stronger for a jet at 30°N than for a jet at 40°N. A strong subtropical jet whose dominant mode of variability resembles pulsing of the jet responds weakly to QBO winds. We therefore explore the dominant modes of lower tropospheric variability of the JCONT and FCONT runs.3 An EOF analysis is performed for weighted daily zonal wind variability at 925 hPa over the North Pacific region (poleward from 20°N, 150°E–150°W), North Atlantic (poleward from 20°N, 60°W–0°), and the Southern Hemisphere poleward from 20°S. The autocorrelation of the first and second principal component is computed, and the portion of the autocorrelation function above 1/e is fit to a decaying exponential. The e-folding time scale of this decaying exponential and the variance explained by the EOF is listed in Table 2.

Table 2.

Characteristics of the first and second principal component of daily 925-hPa zonal wind variability in the North Atlantic (poleward from 20°N, 60°W–0°), North Pacific (poleward from 20°N, 150°E–150°W), and Southern Hemisphere (poleward from 20°S). Each cell contains, in order, the persistence time scale (days) for the first EOF, the percentage of variance explained by the first EOF, the persistence time scale (days) for the second EOF, and the percentage of variance explained by the second EOF. In JCONT and FCONT, the first EOF of SH and North Atlantic variability resembles a shifting jet.

Characteristics of the first and second principal component of daily 925-hPa zonal wind variability in the North Atlantic (poleward from 20°N, 60°W–0°), North Pacific (poleward from 20°N, 150°E–150°W), and Southern Hemisphere (poleward from 20°S). Each cell contains, in order, the persistence time scale (days) for the first EOF, the percentage of variance explained by the first EOF, the persistence time scale (days) for the second EOF, and the percentage of variance explained by the second EOF. In JCONT and FCONT, the first EOF of SH and North Atlantic variability resembles a shifting jet.
Characteristics of the first and second principal component of daily 925-hPa zonal wind variability in the North Atlantic (poleward from 20°N, 60°W–0°), North Pacific (poleward from 20°N, 150°E–150°W), and Southern Hemisphere (poleward from 20°S). Each cell contains, in order, the persistence time scale (days) for the first EOF, the percentage of variance explained by the first EOF, the persistence time scale (days) for the second EOF, and the percentage of variance explained by the second EOF. In JCONT and FCONT, the first EOF of SH and North Atlantic variability resembles a shifting jet.

The first EOF in the SH resembles the southern annular mode (SAM; not shown) and is well separated from the second EOF. The long time scale for the first EOF is consistent with the presence of eddy feedback for a shifting jet (Lorenz and Hartmann 2001; Barnes and Hartmann 2010b).4 The second SH EOF is not well separated from the third. In the Atlantic sector, the first and second EOFs are well separated from each other and from the third EOF, and indicate shifting [i.e., the North Atlantic Oscillation (NAO)] and pulsing of the jet respectively (not shown). We therefore expect a poleward jet shift via the SAM and NAO in response to EQBO winds if our dry model results are relevant to WACCM.

In the Pacific sector, the first EOF resembles a shifting jet for FCONT and hybrid pulsing/shifting for JCONT, as in Eichelberger and Hartmann (2007) (see Figs. 4a,c).5 The difference in lower tropospheric first EOFs suggests stronger transient eddy feedback in FCONT than in JCONT. We therefore might expect a slightly stronger response in the North Pacific to EQBO winds during February, if the intuition from Part I is relevant.

Fig. 4.

Time-average zonal wind (shading and stars) and first two EOFS of 925-hPa zonal wind (contours), in the (a),(b) JCONT and (c),(d) FCONT runs. Contours are shown at ±1, ±2, ±3, and ±4 m s−1 per standard deviation of the first EOF. Shading marks regions where the mean wind exceeds ±5 m s−1 at 925 hPa (the lower-level Pacific jet is less zonally asymmetric than the upper-level Pacific jet). Stars mark the jet maximum.

Fig. 4.

Time-average zonal wind (shading and stars) and first two EOFS of 925-hPa zonal wind (contours), in the (a),(b) JCONT and (c),(d) FCONT runs. Contours are shown at ±1, ±2, ±3, and ±4 m s−1 per standard deviation of the first EOF. Shading marks regions where the mean wind exceeds ±5 m s−1 at 925 hPa (the lower-level Pacific jet is less zonally asymmetric than the upper-level Pacific jet). Stars mark the jet maximum.

The jets examined in Part I were highly idealized, however, while the jets examined here are much more realistic. The first EOF (the annular mode) dominates the variability in a dry model unrealistically, while both pulsing and shifting of the jet are present and contribute to Pacific sector jet variability. In addition, the jets examined here have realistic zonal asymmetry, unlike the jets in Part I. Finally, the mean state of the FCONT and JCONT runs are much more similar than the mean state of the strong and weak subtropical jets from Part I (i.e., the subtropical NP jet maximum is only about 5 m s−1 greater in JCONT than FCONT, whereas the difference in Part I was about 30 m s−1). We therefore test, in the rest of this paper, whether and how EQBO winds influence realistic tropospheric jets.

4. Tropospheric response to EQBO winds

The response in the troposphere to EQBO tropical stratospheric winds is now explored. We first explore the response in an ensemble of runs in which 3×EQBO winds are switched on. We then assess the robustness in a series of equilibrium runs.

The meridional circulation associated with the QBO, as discussed in Part I, is manifested in the FBRANCH ensemble. In the first month after branching, zonally averaged NH temperature anomalies in the lowermost stratosphere resemble that shown in Part I (cf. Figs. 1 and 5 of Part I and Fig. 5b here). These temperature anomalies, and the associated upward and downward motion, are part of the circulation in thermal wind balance with the QBO. Zonal wind anomalies arch downward and poleward from the equatorial QBO anomaly, similar to the effect of the QBO in the absence of eddies (Fig. 5a). Anomalies in the lowermost stratosphere in WACCM and in Part I are nearly identical because the meridional circulation associated with the QBO in the absence of eddies is model independent. In the second month after branching, the lowermost stratospheric temperature anomalies are near their equilibrium values and the zonal wind horseshoe begins to extend into the troposphere (Figs. 5c,d). In ensuing months, the horseshoe amplifies in the troposphere. NH anomalies in the fourth month after branching are quantitatively similar in WACCM and in the dry model (cf. Fig. 9 of Part I and Fig. 5f here).

Fig. 5.

Zonally averaged zonal wind and temperature anomalies (a),(b) one, (c),(d) two, and (e),(f) four months after branching in the perpetual February ensemble with a 3×EQBO profile (FBRANCH-FCONT). Contours are shown at ±0.33, ±1, ±2, ±5, ±10, and ±20 m s−1 and ±0.25, ±0.75, ±1.5, ±3, and ±6 K. The plotting conventions from Part I are used to ease comparison. Significant regions at 95% are shaded and negative (i.e. easterly or cold) contours are thick.

Fig. 5.

Zonally averaged zonal wind and temperature anomalies (a),(b) one, (c),(d) two, and (e),(f) four months after branching in the perpetual February ensemble with a 3×EQBO profile (FBRANCH-FCONT). Contours are shown at ±0.33, ±1, ±2, ±5, ±10, and ±20 m s−1 and ±0.25, ±0.75, ±1.5, ±3, and ±6 K. The plotting conventions from Part I are used to ease comparison. Significant regions at 95% are shaded and negative (i.e. easterly or cold) contours are thick.

Figure 6 shows a map view of the response after branching with EQBO winds. Zonal wind anomalies at the 150-hPa level are significant everywhere in the 15°–30°N latitude band but are strongest in the Pacific near the jet exit region (Figs. 6a,c,e). The meridional circulation associated with the QBO is stronger in the NH.6 At the 300-hPa level (Figs. 6b,d,f), the zonal wind anomalies develop more slowly, although in the fourth month after branching the zonal wind anomaly in the North Pacific approaches 4 m s−1 (Figs. 6b,d,f). The westerly anomaly in the deep tropics, easterly anomaly in the subtropics, and westerly anomaly in the extratropics are strongest in the Pacific. The Atlantic and SH jets shift poleward as well, consistent with expectations. These changes in the troposphere extend to the surface (not shown).

Fig. 6.

Zonal wind anomalies after branching in the perpetual February ensemble with a 3×EQBO profile (FBRANCH-FCONT). Contours are shown at ±0.5, ±1.5, ±3, ±5, and ±8 m s−1. Significant regions at 95% are shaded and negative (easterly) contours are thick. The response in the third month after branching is omitted for brevity.

Fig. 6.

Zonal wind anomalies after branching in the perpetual February ensemble with a 3×EQBO profile (FBRANCH-FCONT). Contours are shown at ±0.5, ±1.5, ±3, ±5, and ±8 m s−1. Significant regions at 95% are shaded and negative (easterly) contours are thick. The response in the third month after branching is omitted for brevity.

We now use the vorticity budget [Eq. (1)] to diagnose the contribution of transient eddies to the growth of tropospheric zonal wind anomalies and the associated vorticity anomalies in response to the QBO. In particular, we want to show the important role of vorticity fluxes by transient eddies [second term on right-hand side of Eq. (1)] for amplifying the vorticity anomalies and for extending them toward the surface. Figure 7 shows the anomalous vorticity forcing by high-frequency eddies in the Pacific and Atlantic sectors four months after QBO winds are imposed in the FBRANCH ensemble. These eddy vorticity flux convergences strongly support the zonal wind anomalies and poleward-shifted jet shown in Figs. 5 and 6. High-frequency eddy forcing increases steadily after QBO onset, implying that eddy feedbacks are amplifying with the tropospheric response to the QBO. High-frequency eddies in the Atlantic and Pacific are independent of each other and are not well correlated in time, but they react similarly to the external QBO forcing. In the reanalysis data as well, the difference in the high-frequency eddy forcing between EQBO and WQBO reinforces the tropospheric NP zonal wind and vorticity anomalies and leads to a poleward-shifted jet (not shown, but similar to Fig. 7). Analysis of the stretching and wave terms in Eq. (1) indicates that they are less important than the transient eddy flux but do act to move vorticity anomalies downward and sustain them against surface drag (not shown), much as shown by BH10 for the NAO. Analysis of the vertical terms in Eq. (1) indicates that they are important in the lowermost stratosphere but not below 200 hPa. As in Part I, high-frequency eddies are vital for amplifying the QBO signal in the troposphere.

Fig. 7.

Contours denote anomalous upper-level (i.e., 300–150 hPa) high-frequency eddy term of the vorticity budget in the fourth month after branching in the FBRANCH ensemble (FBRANCH-FCONT). Contour interval is 1.5 × 10−12 s−2. Shading denotes mass weighted vorticity anomalies between 700 and 925 hPa that exceed 1.75 × 10 −6 s−1.

Fig. 7.

Contours denote anomalous upper-level (i.e., 300–150 hPa) high-frequency eddy term of the vorticity budget in the fourth month after branching in the FBRANCH ensemble (FBRANCH-FCONT). Contour interval is 1.5 × 10−12 s−2. Shading denotes mass weighted vorticity anomalies between 700 and 925 hPa that exceed 1.75 × 10 −6 s−1.

a. Influence of convection and polar vortex variability

One might posit that the influence of the QBO on the polar vortex or on convection is involved in the QBO response in the troposphere. Even though Part I found that high-frequency eddies modify the jets in the troposphere even in the absence of convection and polar vortex variability, it is conceivable that these pathways are important in the real atmosphere. We therefore investigate whether convection or polar vortex variability is important for the tropospheric response in FBRANCH.

We first discuss how convection changes in response to EQBO winds in the FBRANCH ensemble. We then seek to understand how convective variability and polar vortex variability influence the NP in the FCONT run. Finally, we project the changes in convection and in the polar vortex in the FBRANCH ensemble onto the effect that such an anomaly had in FCONT. We can thus deduce the role that convection and polar vortex variability may have had in the FBRANCH ensemble.

Static stability is decreased, upwelling enhanced, and tropopause height increased, in the equatorial upper troposphere as part of the meridional circulation in thermal wind balance with EQBO winds [see Fig. 5b herein, section 4 of Part I, and Collimore et al. (2003)]. These changes are expected to increase the height to which deep convection can rise in deep convecting regions of the tropics, although it is not clear whether the amount of convection should be increased. To demonstrate that deep convecting cloud-top heights are affected by the QBO, we show the change in OLR at the top of the atmosphere in the FBRANCH ensemble in Figs. 8a,c,e. EQBO leads to a significant decrease in OLR in regions where high clouds are present climatologically. The negative OLR anomalies peak near Indonesia and Africa at 12.7 W m−2 in the first month and 23 W m−2 in the fourth month after branching. Positive OLR anomalies are present in the subtropics and extratropics, consistent with the descending motion that is part of the meridional circulation of the QBO (Collimore et al. 2003). High cloud fraction is also enhanced fairly uniformly throughout the deep tropics and decreased in the subtropics (not shown). The height to which deep convection can rise is affected by the QBO in the FBRANCH ensemble; it is likely that the QBO can influence the height to which deep convection can rise in the atmosphere as well.

Fig. 8.

OLR at the top of the atmosphere and convective precipitation anomalies after branching for the 3×EQBO ensemble in perpetual February (FBRANCH-FCONT). Contours are shown at ±3, ±9, ±15, ±21, and ±27 W m−2 for OLR and ±0.1, ±0.3, ±0.5, ±0.7, and ±0.9 mm day−1 for convective precipitation. Significant regions at 95% are shaded and negative contours are thick.

Fig. 8.

OLR at the top of the atmosphere and convective precipitation anomalies after branching for the 3×EQBO ensemble in perpetual February (FBRANCH-FCONT). Contours are shown at ±3, ±9, ±15, ±21, and ±27 W m−2 for OLR and ±0.1, ±0.3, ±0.5, ±0.7, and ±0.9 mm day−1 for convective precipitation. Significant regions at 95% are shaded and negative contours are thick.

Changes in OLR are well correlated with changes in total convection, but decreased OLR does not prove that there is more convection. We therefore examine convective mass flux, heating from moist processes, and total convective precipitation, to determine whether the total amount of tropical convection is modulated by the QBO. Zonally averaged convective mass flux and heating from moist processes equatorward of 6° are increased significantly at the 95% level above 400 hPa but not in the lower troposphere. Even at upper levels, the changes are highly zonally asymmetric. The strongest enhancement in convection is in the intertropical convergence zone (ITCZ) in the Pacific (e.g., convective precipitation in Fig. 8). If the convection in the Pacific ITCZ is removed from the zonal average, the increase in the amount of tropical convection is no longer significant. Future work is necessary to understand why anomalies in the amount of convection are more regional than changes in OLR.

We now address whether anomalies in convection due to the QBO cause zonal wind anomalies in the extratropics. We do this by first quantifying (in a linear sense) the connection between OLR variability and zonal wind variability in the FCONT run, and then applying this linear connection to the observed OLR anomalies in the FBRANCH ensemble. The procedure is as follows: 1) The OLR anomalies for the FCONT run are decomposed into 100 EOFs. 2) Multiple regression is used to relate OLR EOFs to 300-hPa zonal wind anomalies in the FCONT run. 3) The EOFs are projected against the OLR anomaly in month 4 (Fig. 8e). 4) These projection coefficients are multiplied by the map that connects OLR EOFs to zonal wind anomalies in the FCONT run, thus generating the 300-hPa zonal wind associated (in a linear sense) with the OLR anomalies in Fig. 8.7 Figure 9a shows that a decrease in tropical OLR is typically associated with an intensification of the subtropical jet. An increase in tropical heating typically leads to stronger subtropical jet in dry models (e.g., section 6 of Part I) and in observations (e.g., in El Niño) as well. EQBO leads to a weakening of the subtropical jet in our model runs, contrary to what might be expected from the change in OLR. We therefore conclude that anomalies in convection due to the meridional circulation of the QBO are not responsible for the extratropical response. Rather, changes in convection might damp the extratropical response.

Fig. 9.

300-hPa zonal wind anomalies linearly associated with the OLR anomalies in month 4 of Fig. 8 and with the weakening of the polar vortex for the FBRANCH ensemble. Contours are shown at ±0.5, ±1.5, ±3, ±5, and ±8 m s−1. Regions with large anomalies are shaded.

Fig. 9.

300-hPa zonal wind anomalies linearly associated with the OLR anomalies in month 4 of Fig. 8 and with the weakening of the polar vortex for the FBRANCH ensemble. Contours are shown at ±0.5, ±1.5, ±3, ±5, and ±8 m s−1. Regions with large anomalies are shaded.

We now investigate the role of polar vortex variability in the tropospheric response to the QBO in FBRANCH. The polar cap temperature increases by about 1.5 K over the course of the 120-day FBRANCH simulation.8 Linear regression is used to estimate the zonal wind anomalies in the troposphere associated with such a temperature increase. A time series of the polar cap temperature anomaly in each month of the FCONT run is created, and the polar cap time series is then regressed against the 300-hPa zonal wind anomalies in the troposphere. Figure 9b shows the zonal wind anomalies associated with a 1.5-K increase in polar cap temperature. The modulation of the troposphere by the vortex does not project onto the NP response to the QBO, just as in Garfinkel and Hartmann (2010). The vortex pathway is especially strong in the Atlantic where a weakened vortex leads to an equatorward shift in the jet, but even there the influence of the vortex on jet position is overwhelmed in our model run. We therefore conclude that the QBO influences the NP mainly via its meridional circulation in thermal wind balance interacting with tropospheric high-frequency eddies, as in the dry model in Part I.

b. Robustness

The robustness of the FBRANCH results and the sensitivity to the choice of the date for the perpetual forcing are explored. We first discuss the JBRANCH ensemble before discussing the equilibrium EQBO cases.

As in the FBRANCH ensemble, zonal wind anomalies in the JBRANCH ensemble extend out of the tropical stratosphere in a horseshoe pattern. The zonal wind anomaly at 150 hPa is stronger in the Pacific sector of the Northern Hemisphere than in any other region (Fig. 10). The zonal wind anomaly at 300 hPa in the Pacific is qualitatively similar to, though much weaker than, the zonal wind anomaly in FBRANCH with 3×EQBO winds. The OLR anomalies are also qualitatively similar to those in the FBRANCH ensemble (not shown) but peak at only 4.5 W m−2 in the first month and 7 W m−2 in the third month after branching.

Fig. 10.

As in Fig. 6, but for the perpetual January ensemble with the default EQBO profile (i.e., JBRANCH-JCONT).

Fig. 10.

As in Fig. 6, but for the perpetual January ensemble with the default EQBO profile (i.e., JBRANCH-JCONT).

We now discuss the tropospheric response in the equilibrium EQBO runs (Figs. 11 and 12). See section 2b for a description of the model runs.

Fig. 11.

Difference in 300-hPa zonal wind between the equilibrium EQBO cases and the FCONT and JCONT control cases for the different EQBO wind profiles. Below each panel is the difference in polar cap temperature (area averaged poleward from 70°N from 70 to 150 hPa) between each EQBO case and its corresponding control case. Contour intervals are as in Fig. 6. Significant regions at 95% are shaded and negative (easterly) contours are thick.

Fig. 11.

Difference in 300-hPa zonal wind between the equilibrium EQBO cases and the FCONT and JCONT control cases for the different EQBO wind profiles. Below each panel is the difference in polar cap temperature (area averaged poleward from 70°N from 70 to 150 hPa) between each EQBO case and its corresponding control case. Contour intervals are as in Fig. 6. Significant regions at 95% are shaded and negative (easterly) contours are thick.

Fig. 12.

As in Fig. 11, but for OLR. Contour intervals are as in Fig. 8.

Fig. 12.

As in Fig. 11, but for OLR. Contour intervals are as in Fig. 8.

  • The subtropical jet in the NP in the perpetual January and February 3×EQBO equilibrium runs is significantly weaker than in JCONT and FCONT, as expected (Figs. 11a,b). The response is stronger during January than during February, contrary to expectations.9 The NP zonal wind anomaly in FBRANCH four months after branching is bigger than the NP zonal wind anomaly in the equilibrium 3×EQBO run (Fig. 11b vs Fig. 6f). Anomalies in OLR are near their equilibrium value by the fourth month after branching (Fig. 12b vs Fig. 8f) as well, suggesting that the tropospheric response to QBO winds equilibrates after four months.

  • The response in the NP to the default EQBO profile qualitatively resembles the response to the 3×EQBO profile (Figs. 11c,d and 12c,d). The zonal wind anomaly in the NP in JBRANCH reaches the amplitude of that in the equilibrium EQBO run three months after branching (Fig. 10 vs Fig. 11c).

  • Including upper stratospheric westerlies present in realistic QBO profiles weakens the NP zonal wind response (Figs. 11e,f) and the tropical OLR response (Figs. 12e,f), but the overall pattern is nearly identical. Sensitivity of the tropospheric response to upper stratospheric westerlies is weak. Such weak sensitivity might be expected from the nearly identical meridional circulations in the lower stratosphere between Figs. 6a and 6h of Part I.

  • The NP zonal wind response to weak EQBO winds forced only below the 0.061 hybrid σ level is comparable to the response when winds are relaxed over more of the stratosphere during perpetual February (Fig. 11h vs Fig. 11d) but not during perpetual January. The difference in NP zonal wind response between perpetual February and perpetual January is statistically significant. The OLR response is similar to that for the default EQBO profile (Fig. 12h vs Fig. 12d). The response is stronger during February than during January, consistent with expectations.

  • The response to weak EQBO winds forced only above the 0.061 hybrid σ level is significant only in the perpetual February case (Figs. 11i,j and 12i,j). The winds between 100 and 60 hPa are similar to the control case, so that the thermal wind balanced meridional circulation associated with the QBO is weaker in the lowermost stratosphere (not shown). The NP zonal wind response is significantly stronger during February than during January, consistent with observations and expectations. Changes in OLR in the deep convecting regions of the tropics are smaller than that for the other QBO profiles tested, suggesting that the direct modulation of tropical convection is weaker when QBO wind anomalies do not extend into the lowermost stratosphere (Figs. 12i,j). We therefore expect that a model whose internally generated QBO does not extend deeply enough into the lower stratosphere will miss some of the effect of the QBO in the troposphere.

Even though the response in these cases varies as the QBO winds relaxed toward vary, the NP responses, if present, are qualitatively similar. The similarity in response suggests that in all of these cases, the QBO affects the subtropical NP by influencing tropospheric eddies. Finally, we have examined the tropospheric response to realistic WQBO stratospheric winds and to 3×WQBO winds (see Fig. 3 of Part I for the profile used). The NP response to WQBO winds is nearly equal (though the magnitude of the anomalies is slightly weaker) and opposite to that for EQBO shown in Fig. 11; details are excluded for brevity.

The effect of the QBO in the troposphere in the Atlantic is only robust with 3×EQBO winds. The weakened response to realistic EQBO winds is consistent with Part I, where we found that only 3×EQBO winds affected a jet whose mean position is near 40° (J40 in section 5 of Part I). Furthermore, the effect of the QBO in the troposphere through the polar vortex is particularly strong in the Atlantic sector (e.g., Fig. 9b) and opposes the effect of the QBO on tropospheric eddies through its thermal wind balanced meridional circulation. Finally, the modeled North Atlantic response to WQBO winds is not robust (not shown). The net effect of realistic QBO winds in the North Atlantic during January and February is weak, just like in the reanalysis and coupled seasonal WACCM run in Fig. 1.

In the cases shown here, as well as in Fig. 6f, the QBO appears to influence the SAM near Australia and the Indian Ocean. The SH tropospheric jet is shifted poleward in response to EQBO winds, consistent with expectations. While this seems to be a robust feature in WACCM, it does not appear in the reanalysis (e.g., Fig. 1). As discussed in section 3, the SH mean state in the FCONT and JCONT runs is not particularly true to the observed mean state. Only a seasonal integration of WACCM can accurately model the SH. We are therefore not ready to attach any meaning to a comparison of these SH WACCM results to the reanalysis.

5. Conclusions

In this paper we have examined the tropospheric response to easterly QBO wind anomalies using an ensemble of experiments with the WACCM. The results support previous work done with a dry model in Part I.

In both the reanalysis record and a coupled seasonal WACCM run, winds are significantly weaker near and south of the climatological jet position in the subtropical Pacific during the easterly phase of the QBO relative to its westerly phase during February and March. To understand why the QBO may have this effect, the influence of the QBO on the troposphere during perpetual February and January WACCM runs with fixed SSTs is studied. As shown in Part I, QBO momentum anomalies require a meridional circulation to establish thermal wind balance. The zonal wind associated with this circulation arches down to the subtropical troposphere. High-frequency eddies can then interact with this circulation and extend it to the surface. The subtropical Pacific jet is weakened and contracted. The response is stronger during February than during January. All of these responses are similar to those found for a dry GCM in Part I and in reanalysis data. Cloud tops rise in response to the QBO, likely in response to the circulation that brings the easterly stratospheric wind anomalies into geostrophic balance. But these convective anomalies appear to work against the extratropical tropospheric circulation anomalies. Qualitative resemblance between the anomalies shown here and in the reanalysis (i.e., between Figs. 1 and 6) suggests that our model can capture the mechanisms through which the QBO influences the troposphere.

Garfinkel and Hartmann (2010) argue that the QBO influences El Niño teleconnections. The results here and in Part I affirm 1) that the QBO can alter the background state in the North Pacific experienced by an anomalous Rossby wave train as it propagates poleward away from anomalous SSTs and 2) that the “direct” effect in section 4 of Garfinkel and Hartmann (2010) is a real physical phenomenon.

The response to the QBO is strongest in the North Pacific during February and weaker in other basins and during January, consistent with Part I and observations. Our results highlight the sensitivity of the response to the mean state of the unperturbed climate. We therefore strongly suggest that when studying the tropospheric response to other external perturbations, attention should be paid to regional and intraseasonal differences.

The perpetual January and February model runs used here still have limitations in their ability to simulate the actual atmosphere. In our model runs, the stratospheric QBO winds do not propagate downward but rather remain fixed. Because the lowermost stratospheric meridional circulation has reached most of its strength by the second month, we expect that the stratospheric circulation for a downward propagating QBO will be nearly identical to that shown here and in Part I. The tropospheric anomalies do not reach their equilibrium value until the third or fourth month after branching; we therefore expect a slightly weaker tropospheric response for a downward propagating QBO than that shown here. Further analysis is needed to address whether beating of the QBO with the annual cycle or decadal variability of the effect of the QBO on the vortex might manifest in the troposphere (Ruzmaikin et al. 2005; Lu et al. 2008). The QBO in our model configuration is imposed by relaxing toward a specified tropical zonal wind profile. Attempts to incorporate an internally generated QBO into GCMs have met with increasing success. Many of these models still have trouble simulating the propagation of QBO anomalies down to the lowermost stratosphere, however. The current generation of models that internally generate a QBO may therefore have difficulty capturing the effect of the QBO in the troposphere. Should the internally generated QBO winds in future generations of models reach the lowermost stratosphere, it will be attractive to use such models to revisit the QBO’s effect on the troposphere. Finally, models with a more realistic sudden stratospheric warming frequency and with better resolved convection are necessary to confirm our results. Nevertheless, similarities among our perpetual winter WACCM and dry model runs, the fully coupled seasonal WACCM run, and the reanalysis suggest that the mechanism we are describing is real and robust.

Results presented here, in Dall’Amico et al. (2010), in Giorgetta et al. (1999), and in Marshall and Scaife (2009) suggest that seasonal forecasting of tropospheric variability could be improved since the QBO phase is predictable months in advance. In particular, the current generation of models should incorporate zonal mean zonal wind relaxation to predicted QBO-dependent anomalous winds.

Acknowledgments

This work was supported by the National Science Foundation under Grant AGS 0960497. We thank the developers at NCAR for the WACCM 3.1.9, our anonymous reviewers for their helpful comments, and Marc L. Michelsen for assisting our model runs.

REFERENCES

REFERENCES
Anstey
,
J. A.
,
T. G.
Shepherd
, and
J. F.
Scinocca
,
2010
:
Influence of the quasi-biennial oscillation on the extratropical winter stratosphere in an atmospheric general circulation model and in reanalysis data
.
J. Atmos. Sci.
,
67
,
1402
1419
.
Baldwin
,
M. P.
, and
T. J.
Dunkerton
,
1999
:
Propagation of the Arctic Oscillation from the stratosphere to the troposphere
.
J. Geophys. Res.
,
104
(
D24
),
30 937
30 946
.
Baldwin
,
M. P.
, and
Coauthors
,
2001
:
The Quasi-Biennial Oscillation
.
Rev. Geophys.
,
39
,
179
229
.
Barnes
,
E. A.
, and
D. L.
Hartmann
,
2010a
:
Dynamical feedbacks and the persistence of the NAO
.
J. Atmos. Sci.
,
67
,
851
865
.
Barnes
,
E. A.
, and
D. L.
Hartmann
,
2010b
:
Dynamical feedbacks of the southern annular mode in winter and summer
.
J. Atmos. Sci.
,
67
,
2320
2330
.
Barnes
,
E. A.
,
D. L.
Hartmann
,
D. M. W.
Frierson
, and
J.
Kidston
,
2010
:
Effect of latitude on the persistence of eddy-driven jets
.
Geophys. Res. Lett.
,
37
,
L11804
,
doi:10.1029/2010GL043199
.
Collimore
,
C. C.
,
D. W.
Martin
,
M. H.
Hitchman
,
A.
Huesmann
, and
D. E.
Waliser
,
2003
:
On the relationship between the QBO and tropical deep convection
.
J. Climate
,
16
,
2552
2568
.
Collins
,
W. D.
, and
Coauthors
,
2006
:
The Formulation and Atmospheric Simulation of the Community Atmosphere Model Version 3 (CAM3)
.
J. Climate
,
19
,
2144
2161
.
Coughlin
,
K. T.
, and
K.-K.
Tung
,
2001
:
QBO signal found at the extratropical surface through northern annular modes
.
Geophys. Res. Lett.
,
28
,
4563
4566
.
Coughlin
,
K. T.
, and
K.-K.
Tung
,
2005
:
Empirical mode decomposition of climate variability in the atmospheric
.
Hilbert–Huang Transform: Introduction and Application, N. Huang and S. Shen, Eds., World Scientific, 149–164
.
Crooks
,
S. A.
, and
L. J.
Gray
,
2005
:
Characterization of the 11-year solar signal using a multiple regression analysis of the ERA-40 dataset
.
J. Climate
,
18
,
996
1015
.
Dall’Amico
,
M.
,
P. A.
Stott
,
A. A.
Scaife
,
L. J.
Gray
,
K. H.
Rosenlof
, and
A. Y.
Karpechko
,
2010
:
Impact of stratospheric variability on tropospheric climate change
.
Climate Dyn.
,
34
,
399
417
,
doi:10.1007/s00382-009-0580-1
.
Edmon
,
H. J.
,
B. J.
Hoskins
, and
M. E.
McIntyre
,
1980
:
Eliassen–Palm cross sections for the troposphere
.
J. Atmos. Sci.
,
37
,
2600
2616
.
Eichelberger
,
S. J.
, and
D. L.
Hartmann
,
2007
:
Zonal jet structure and the leading mode of variability
.
J. Climate
,
20
,
5149
5163
.
Garcia
,
R. R.
,
D.
Marsh
,
D.
Kinnison
,
B.
Boville
, and
F.
Sassi
,
2007
:
Simulations of secular trends in the middle atmosphere, 1950–2003
.
J. Geophys. Res.
,
112
,
D09301
,
doi:10.1029/2006JD007485
.
Garfinkel
,
C. I.
, and
D. L.
Hartmann
,
2007
:
Effects of the El Niño–Southern Oscillation and the quasi-biennial oscillation on polar temperatures in the stratosphere
.
J. Geophys. Res.
,
112
,
D19112
,
doi:10.1029/2007JD008481
.
Garfinkel
,
C. I.
, and
D. L.
Hartmann
,
2010
:
The influence of the quasi-biennial oscillation on the North Pacific and El Niño teleconnections
.
J. Geophys. Res.
,
115
,
D20116
,
doi:10.1029/2010JD014181
.
Garfinkel
,
C. I.
, and
D. L.
Hartmann
,
2011
:
The influence of the quasi-biennial oscillation on the troposphere in winter in a hierarchy of models. Part I: Simplified dry GCMs
.
J. Atmos. Sci.
,
68
,
1273
1289
.
Gerber
,
E. P.
, and
L. M.
Polvani
,
2009
:
Stratosphere–troposphere coupling in a relatively simple AGCM: The importance of stratospheric variability
.
J. Climate
,
22
,
1920
1933
.
Giorgetta
,
M. A.
,
L.
Bengtsson
, and
K.
Arpe
,
1999
:
An investigation of QBO signals in the East Asian and Indian monsoon in GCM experiments
.
Climate Dyn.
,
15
,
435
450
,
doi:10.1007/s003820050292
.
Giorgetta
,
M. A.
,
E.
Manzini
,
E.
Roeckner
,
M.
Esch
, and
L.
Bengtsson
,
2006
:
Climatology and forcing of the quasi-biennial oscillation in the MAECHAM5 model
.
J. Climate
,
19
,
3882
3901
.
Gray
,
W. M.
,
J. D.
Sheaffer
, and
J. A.
Knaff
,
1992
:
Hypothesized mechanism for stratospheric QBO influence on ENSO variability
.
Geophys. Res. Lett.
,
19
,
107
110
.
Haigh
,
J. D.
,
M.
Blackburn
, and
R.
Day
,
2005
:
The response of tropospheric circulation to perturbations in lower-stratospheric temperature
.
J. Climate
,
18
,
3672
3685
.
Held
,
I. M.
,
1975
:
Momentum transport by quasi-geostrophic eddies
.
J. Atmos. Sci.
,
32
,
1494
1497
.
Ho
,
C.-H.
,
H.-S.
Kim
,
J.-H.
Jeong
, and
S.-W.
Son
,
2009
:
Influence of stratospheric quasi-biennial oscillation on tropical cyclone tracks in the western North Pacific
.
Geophys. Res. Lett.
,
36
,
L06702
,
doi:10.1029/2009GL037163
.
Holton
,
J. R.
, and
H. C.
Tan
,
1980
:
The influence of the equatorial quasi-biennial oscillation on the global circulation at 50 mb
.
J. Atmos. Sci.
,
37
,
2200
2208
.
Huesmann
,
A. S.
, and
M. H.
Hitchman
,
2001
:
The stratospheric quasi-biennial oscillation in the NCEP reanalyses: Climatological structures
.
J. Geophys. Res.
,
106
,
11 859
11 874
.
Hurwitz
,
M. M.
,
I.-S.
Song
,
L. D.
Oman
,
P. A.
Newman
,
A. M.
Molod
,
S. M.
Frith
, and
J. E.
Nielsen
,
2011
:
Response of the Antarctic stratosphere to warm pool El Niño events in the GEOS CCM
.
Atmos. Chem. Phys. Discuss.
,
11
,
9743
9767
,
doi:10.5194/acpd-11-9743-2011
.
Kawatani
,
Y.
,
S.
Watanabe
,
K.
Sato
,
T. J.
Dunkerton
,
S.
Miyahara
, and
M.
Takahashi
,
2010
:
The roles of equatorial trapped waves and internal inertia–gravity waves in driving the quasi-biennial oscillation. Part I: Zonal mean wave forcing
.
J. Atmos. Sci.
,
67
,
963
980
.
Kulyamin
,
D. V.
,
E. M.
Volodin
, and
V. P.
Dymnikov
,
2009
:
Simulation of the quasi-biennial oscillations of the zonal wind in the equatorial stratosphere. Part II: Atmospheric general circulation models
.
Izv. Atmos. Ocean. Phys.
,
45
,
37
54
,
doi:10.1134/S0001433809010046
.
Limpasuvan
,
V.
,
D. W. J.
Thompson
, and
D. L.
Hartmann
,
2004
:
The life cycle of the Northern Hemisphere sudden stratospheric warmings
.
J. Climate
,
17
,
2584
2596
.
Lorenz
,
D. J.
, and
D. L.
Hartmann
,
2001
:
Eddy–zonal flow feedback in the Southern Hemisphere
.
J. Atmos. Sci.
,
58
,
3312
3327
.
Lu
,
H.
,
M. P.
Baldwin
,
L. J.
Gray
, and
M. J.
Jarvis
,
2008
:
Decadal-scale changes in the effect of the QBO on the northern stratospheric polar vortex
.
J. Geophys. Res.
,
113
,
D10114
,
doi:10.1029/2007JD009647
.
Marsh
,
D.
,
R. R.
Garcia
,
D. E.
Kinnison
,
B. A.
Boville
,
F.
Sassi
,
S. C.
Solomon
, and
K.
Matthes
,
2007
:
Modeling the whole atmosphere response to solar cycle changes in radiative and geomagnetic forcing
.
J. Geophys. Res.
,
112
,
D23306
,
doi:10.1029/2006JD008306
.
Marshall
,
A. G.
, and
A. A.
Scaife
,
2009
:
Impact of the QBO on surface winter climate
.
J. Geophys. Res.
,
114
,
D18110
,
doi:10.1029/2009JD011737
.
Matthes
,
K.
,
D. R.
Marsh
,
R. R.
Garcia
,
D. E.
Kinnison
,
F.
Sassi
, and
S.
Walters
,
2010
:
Role of the QBO in modulating the influence of the 11-year solar cycle on the atmosphere using constant forcings
.
J. Geophys. Res.
,
115
,
D18110
,
doi:10.1029/2009JD013020
.
Osprey
,
S. M.
,
L. J.
Gray
,
S. C.
Hardiman
,
N.
Butchart
,
A. C.
Bushell
, and
T. J.
Hinton
,
2010
:
The climatology of the middle atmosphere in a vertically extended version of the Met Office’s climate model. Part II: Variability
.
J. Atmos. Sci.
,
67
,
3637
3651
.
Polvani
,
L. M.
, and
P. J.
Kushner
,
2002
:
Tropospheric response to stratospheric perturbations in a relatively simple general circulation model
.
Geophys. Res. Lett.
,
29
,
1114
,
doi:10.1029/2001GL014284
.
Richter
,
J. H.
,
F.
Sassi
,
R. R.
Garcia
,
K.
Matthes
, and
C. A.
Fischer
,
2008
:
Dynamics of the middle atmosphere as simulated by the Whole Atmosphere Community Climate Model, version 3 (WACCM3)
.
J. Geophys. Res.
,
113
,
D08101
,
doi:10.1029/2007JD009269
.
Ruzmaikin
,
A.
,
J.
Feynman
,
X.
Jiang
, and
Y. L.
Yung
,
2005
:
Extratropical signature of the quasi-biennial oscillation
.
J. Geophys. Res.
,
110
,
D11111
,
doi:10.1029/2004JD005382
.
Uppala
,
S. M.
, and
Coauthors
,
2005
:
The ERA-40 Re-Analysis
.
Quart. J. Roy. Meteor. Soc.
,
131
,
2961
3012
.
Zhang
,
G. J.
, and
N. A.
McFarlane
,
1995
:
Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model
.
Atmos.–Ocean
,
33
,
407
446
.

Footnotes

1

That is, the 40-yr ECMWF Re-Analysis (ERA-40; Uppala et al. 2005) plus five additional years of operational data; see Part I for more details.

2

In contrast, EQBO and El Niño tended to coincide from 1957 to the start of the satellite era. The average Niño-3.4 index for the composites used in Fig. 1a, which sample from 1957 to 2007, are within 0.15 of each other. The significant anomaly in Fig. 1a is not due to aliasing from El Niño.

3

Because lower tropospheric variability more cleanly isolates the influence of high-frequency eddies than the influence of the thermal component of the Hadley circulation, we focus on lower levels.

4

In the SH, the FCONT jet is farther poleward yet its jet shifts persist for longer than those of the JCONT jet. That a farther poleward jet would have stronger eddy feedback seemingly contradicts Barnes et al. (2010) where more poleward jets have weaker eddy feedbacks and smaller jet-shifting persistence time scales. One possible explanation is that the weak yet variable stratospheric polar vortex in FCONT couples to tropospheric variability while the even weaker vortex in FCONT does not (not shown). Coupling between the lowermost stratosphere and troposphere has been shown to increase the persistence time scale of tropospheric annular modes (Gerber and Polvani 2009). A thorough investigation is outside the scope of this work.

5

The first PC in the Pacific has a shorter time scale than the first PC in the other regions, consistent with the weakened eddy feedbacks in a pulsed jet.

6

The asymmetry between hemispheres is much weaker in Part I because the parameterizations for the troposphere are hemispherically symmetric.

7

A second method was also examined. The area averaged pattern correlation was taken between the OLR anomalies in Fig. 8 and the OLR anomalies in each month of the control run. The resultant time series is normalized and is then regressed against the zonal wind anomalies of the control run. The zonal wind pattern is qualitatively similar to that shown in Fig. 9a. Results are also qualitatively similar if the OLR pattern in month 1 or 2 after branching is used, and is not sensitive to halving or doubling the number of OLR EOFs retained. Results are also qualitatively similar if the convective precipitation anomalies are used.

8

Polar cap temperature is defined as the area-averaged temperature poleward from 70°N and vertically from 70 to 150 hPa. The increase is nearly 3 K from 20 to 50 hPa. Changes in Rossby wave propagation and EP flux convergence are broadly consistent with linear theory; a future paper by C. Garfinkel et al. will investigate the changes in Rossby wave propagation in more detail.

9

In the strong subtropical jet case in Part I, the jet shifts equatorward and not poleward in response to EQBO winds to maintain stability. Here, the jet is stable in all cases examined and shifts poleward. Nevertheless, a weaker shift is expected during January than during February because of weakened eddy feedback during January. A systematic analysis of the response to a wide range of EQBO amplitudes is beyond the scope of this study.