An analysis of the relationship between zonal wind in the equatorial stratosphere and zonal wind, temperature, and Eliassen and Palm (E-P) flux in the Northern Hemisphere extratropical winter was performed using the CCSR/NIES chemistry-climate model (CCM) and the Japanese 25 year Reanalysis (JRA-25) data. The tropical zonal wind of the CCM was forced by observations, including the observational equatorial quasi-biennial oscillation (QBO). The influence of the QBO on the Northern Hemisphere winter was estimated by a composite analysis with statistical significance. The analyses suggest that the difference in latitude of the critical line (zero value line of zonal wind) in the low latitudes around 10 hPa as well as 50 hPa between the easterly and westerly phases is related to the wave propagation and circulation over the whole depth of the stratosphere. The circulation anomaly is further related to the temperature anomaly at the Northern Hemisphere midlatitudes and the zonal wind anomaly at the high latitudes. These results suggest a mechanism through which 10 hPa QBO could influence the polar vortex, while the mechanism of 50 hPa QBO influence is unclear but the possibility is not ruled out.
 The relationship between the equatorial quasi-biennial oscillation (QBO) and extratropical circulation in the Northern Hemisphere winter is known as the Holton-Tan effect (hereafter HT effect), and the importance of the effects of planetary waves on the HT effect has been discussed [e.g., Holton and Tan, 1980, 1982]. The HT effect is a phenomenon in which the westerly wind in the Northern Hemisphere high latitudes tends to be stronger in the westerly phase of the QBO while weaker in the easterly phase, which is defined by the equatorial zonal wind at 50 hPa. The HT effect has been simulated by three-dimensional (3-D) mechanistic models [e.g., O'Sullivan and Young, 1992; O'Sullivan and Dunkerton, 1994] and by 3-D general circulation models (GCMs) that generated the QBO internally [Niwano and Takahashi, 1998]. The HT effect has also been reproduced in GCMs where the QBO is imposed [Hamilton, 1998] and in GCMs where the QBO is produced through parameterized gravity waves [Marshall and Scaife, 2009], indicating that the HT effect is not too sensitive to the method of generating the QBO.
 The effects of planetary waves on the HT effect are explained as follows [e.g., Holton and Tan, 1980, 1982]: The phase of the QBO near 50 hPa defines the phase of the QBO, and the QBO is related to the latitude of the zero wind line of the zonal wind, which functions as a critical line for stationary planetary wave propagation. As a result, stationary planetary waves propagating from the extratropics toward the equator can reach the equatorial lower stratosphere in the westerly phase of the QBO, while, in the easterly phase of the QBO, planetary waves are restricted to the extratropics. This affects the propagation of planetary waves in the high latitudes. This explanation assumes that the Eliassen and Palm (E-P) flux anomaly in the westerly phase to that in the easterly phase indicates an equatorward direction in the tropical lower stratosphere.
 The upper boundary of the reanalysis data has been extended to the upper stratosphere above 10 hPa since the early 1990s [e.g., Swinbank and O'Neill, 1994], and the upper and middle stratosphere have been included for statistical analyses of the QBO effects in several studies. For example, using the 40 year European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) data, Anstey and Shepherd  and Lu et al.  demonstrated the importance of zonal wind in the tropical upper stratosphere to the polar vortex. Gray et al.  made two experiments using a 3-D mechanistic model. In both experiments, radiosonde data of equatorial zonal wind were assimilated in the model. One experiment assimilated the data in the lower stratosphere over 16–32 km, and the other experiment, in the entire stratosphere over 16–58 km. A realistic HT effect was reproduced only when the observation data were assimilated in the entire stratosphere. Pascoe et al.  conducted several 3-D GCM experiments where the QBO and semi annual oscillations (SAO) were imposed differently for the vertical extent and suggested that the QBO in the lower equatorial stratosphere influences early winter polar variability, whereas the QBO and SAO in the upper equatorial stratosphere together influence midwinter polar variability, and hence, the upper stratosphere wind is an important factor to simulate the realistic polar variability. Naoe and Shibata  analyzed the ERA-40 data and the output of the Meteorological Research Institute (MRI) chemistry-climate model (CCM) that generates the QBO internally. They showed that the equatorward E-P flux anomalies in the westerly phase of the QBO were evident around equatorial 50 hPa but not over the midlatitude lower stratosphere; rather, poleward E-P flux anomalies were seen there. They concluded that the QBO influence on the midlatitudes and high latitudes did not come from the lower equatorial stratosphere and suggested the importance of the zonal wind in the equatorial upper stratosphere for the circulation in the Northern Hemisphere winter. Anstey et al.  also investigated the QBO effect on the zonal mean zonal wind in the winter extratropical stratosphere from a viewpoint of the whole depth of the stratosphere using the ERA-40 data and the output of the Canadian Middle Atmosphere Model (CMAM) that generated the QBO internally. They analyzed the QBO effect on a monthly time scale. They suggest that the timing of QBO phase transitions in relation to the seasonal cycle causes the difference in the QBO effects on polar vortex intensity.
 For this study, using the Center for Climate System Research/National Institute for Environmental Studies (CCSR/NIES) CCM, which covers the region from the surface to about 80 km, we performed numerical experiments in which radiosonde data of equatorial zonal wind were assimilated in the model in order to simulate a realistic QBO. Then, we analyzed the anomalies of the equatorial zonal winds, the zonal winds in the Northern Hemisphere extratropical winter, the temperature, and the E-P flux and its divergence between the westerly and easterly phases of the QBO with statistical significance and discussed their relationships. Furthermore, we also conducted analyses of the Japanese 25 year Reanalysis (JRA-25) data for comparison with the CCM results.
2. Experimental Setup and Method for Analysis
 The model used in this study is the CCSR/NIES CCM. The model has a T42 horizontal resolution (approximately 2.8° × 2.8° latitude-longitude resolution) and 34 vertical layers. The CCM is the atmosphere model and not coupled to the ocean (the sea surface temperature is prescribed). The upper boundary of the model is located at about 80 km. The model includes modules for dynamics, stratospheric chemistry, solar radiation transfer, and infrared radiation transfer. It calculates the ozone transport in the atmosphere, ozone production by ultraviolet radiation, and radiation and temperature effects by simulated ozone (see Nagashima et al.  and Akiyoshi et al.  for details).
 The model experiment is the REFB1 experiment of the Chemistry Climate Model Validation-2 (hereafter CCMVal2-REFB1 experiment), which includes the interannual variabilities of the 11 year solar cycle, the QBO, Sea Surface Temperature (SST), volcanic effects, greenhouse gas (GHG) concentrations, and ozone-depleting substance (ODS) concentrations, such as halogen gas concentrations [Morgenstern et al., 2010]. The SST and sea ice evolutions were prescribed by the HadISST1 data, which was provided by the UK Met Office Hadley Centre [Rayner et al., 2003]. The interannual variations of the GHGs and the ODSs were given by the Intergovernmental Panel on Climate Change  SRES A1B scenario and WMO-adjusted scenario A1 [World Meteorological Organization, 2007], respectively. Details of including the 11 year solar cycle effect into the CCM are described by Yamashita et al. .
where the relaxation time of nudging (τ) is 5 days and y and z are the latitude (degree) and altitude (m), respectively. Considering the latitude and altitude ranges of the QBO, it is assumed that yh = 10, z0 = 30000, and zh = 8000. The observational wind data for nudging are the zonal mean data of zonal winds observed by rawinsonde at Canton Island (2.46°S, 171.43°W), Gan, Maldives (0.41°S, 73.09°E), and Singapore (1.22°N, 103.55°E). The data for these observational sites were merged into a single time series and provided by the Chemistry Climate Model Validation (CCMVal), which was used in this study. It should be noted that the data above 10 hPa were processed assuming a constant vertical propagation velocity of the QBO. The merged data set was interpolated to the pressure levels and the time step of the CCM.
 The CCMVal2-REFB1 experiment was performed for the period 1960–2006 following a 10 year spin-up run. Three ensemble runs were performed with different initial conditions for 1 January 1951 in the spin-up run, and the outputs of the three ensemble members were analyzed. The output has 138 Northern Hemisphere winters (December–January–February) from 46 years × 3 ensemble members. The westerly and easterly phases of the QBO were defined as the direction of the zonal mean zonal wind at 50 hPa averaged over 10°S–10°N. By this definition, 80 years were in the westerly phase, and 58 years were in the easterly phase in the Northern Hemisphere winter. In order to examine the effect of the QBO, quantities such as the zonal mean zonal wind, zonal mean temperature, and mean meridional circulation in the easterly phase were subtracted from the westerly phase.
 CCSR/NIES CCM has unrealistically large ozone responses in the tropical lower stratosphere to the 11 year solar cycle [e.g., see Eyring et al., 2010, Figure 8.11c; Austin et al., 2008, Figure 1]. This is because the solar term included erroneously large volcanic effects occurring at a 9 year interval (Mount El Chichón in Mexico in 1982 and Mount Pinatubo in the Philippines in 1991) due to the cold bias of the CCM in the tropical upper stratosphere/lower stratosphere [Yamashita et al., 2010]. In order to exclude the effects of the large volcanic effects and interannual variability of Sea Surface Temperature (SST), we also performed a CCM experiment without these effects (hereafter CNTL experiment).
 For comparisons with the observation, JRA-25/Japan Meteorological Agency (JMA) Climate Data Assimilation System (JCDAS) reanalysis data were also analyzed. Data for 1979–2006 are analyzed and compared with the CCM outputs because the JRA-25 data are not available before 1979. For the JRA-25 data, 14 years are in the westerly phase, and 13 years are in the easterly phase in the Northern Hemisphere winter. It must be noted that the observation data used for producing JRA-25 are almost the same as those for ERA-40 [Onogi et al., 2007]. The QBO in the ERA-40 data was validated by rocketsonde observations in Baldwin and Gray  and are very realistic up to 2–3 hPa. Figure 1a represents the zonal-mean zonal wind over the equator (10°S–10°N) of JRA-25 data. The amplitude and phase of QBO in the CCM oscillates in almost the same way as that of JRA-25 with a slight westerly bias (Figure 1b). Note that the amplitude and phase of the QBO are almost the same below 5 hPa for the JRA-25 and ERA-40 (not shown), because each reanalysis was assimilated to the same observations, although the channels of satellite sensors used to assimilate the data were different.
Figure 2a shows the meridional-vertical distribution of the zonal wind anomaly calculated by the subtraction of the easterly phase composite from the westerly phase composite (W–E) in the Northern Hemisphere winter (December–January–February) using the JRA-25 data. The QBO phases were defined by the 50 hPa wind. In the equatorial regions, a westerly anomaly of about 15 m s−1 is evident around 50 hPa, and an easterly anomaly of about −20 m s−1 is observed around 10 hPa. In the westerly phase composite, the critical line is located in the Northern Hemisphere around 10 hPa and 20°N (Figure 3a). In the northern high latitudes, a westerly anomaly of about 8 m s−1 is evident around 10 hPa with 99% statistical significance (Figure 2a).
 The result of the CCMVal2-REFB1 experiment indicates that the magnitude and distribution of zonal wind anomalies are comparable to those of the JRA-25. In the westerly phase composite, the critical line is also located in the Northern Hemisphere around 10 hPa and 25°N (Figure 3c). The westerly anomaly in the northern high latitudes around 10 hPa is also simulated with a maximum value of about 6 m s−1 with 99% statistical significance (Figure 2b), although the maximum value is slightly smaller than that in the JRA-25.
 The E-P flux divergence anomaly calculated from the JRA-25 and the CCMVal2-REFB1 is depicted by contours in Figures 4a and 4c, respectively, with the statistical significance. The vectors indicate the magnitudes and directions of the E-P flux anomaly. First, we examine the results around 50 hPa, where HT effect has been discussed in the literature, as mentioned earlier. As previous studies have shown, a pattern that was expected by the HT effect is found; because the critical line shifts southward in the westerly phase, equatorward anomalies of the E-P flux are evident around 50–100 hPa in the Northern Hemisphere low latitudes in Figures 4a and 4c. Accordingly, the E-P flux divergence anomalies are evident around 50–100 hPa, 30°N. The E-P flux divergence anomalies are also seen at lower levels around 150 hPa, 10°N, with the upward motion anomaly at 100–200 hPa, 10–20°N and the downward motion anomaly at 100–200 hPa, 0–10°N (see also Figures 4b and 4d). Around 40°N, a poleward anomaly of the meridional component of the E-P flux exists around 50 hPa, as Naoe and Shibata  reported in their analysis of MRI CCM.
 Next, expanding the height region to the upper stratosphere, we investigate the wave propagation, the divergence of the E-P flux, and the meridional circulation associated with the divergences in terms of the QBO phase. During the westerly phase at 50 hPa in the tropics, the zonal wind around 10 hPa is easterly. Hence, the critical line is located in the Northern Hemisphere around 10 hPa. In this situation, around 10 hPa, planetary waves propagating from the midlatitude troposphere in the Northern Hemisphere toward the low-latitude upper stratosphere are restricted to the extratropics. In contrast, in the easterly phase at 50 hPa, the zonal wind around 10 hPa is westerly, and then the critical line is located in the Southern Hemisphere (Figures 3b and 3d). In that case, stationary planetary waves can reach the equator around 10 hPa. Therefore, the E-P flux anomaly in the W – E is poleward around the equatorial 10 hPa.
 Accordingly, a convergence anomaly of the E-P flux occurs around 30°N, 10 hPa, which is expected from the latitude shift of the critical line. The divergence anomaly below that level around 30°N, 50 hPa arises for the same reason, that is, the divergence anomaly represents the decrease in E-P flux convergence that results from wave activity being able to propagate farther equatorward when equatorial winds are westerly at 50 hPa. These explanations expect a quadruple pattern in the anomaly: anomalous divergence over convergence at the equator and anomalous convergence over divergence at 30°N, as seen in Figure 4. These anomalies have 99% statistical significance. The magnitude of the convergence anomaly is about −0.5 m s−1d−1 in the JRA-25 and about −0.4 m s−1d−1 in the CCMVal2-REFB1 experiment, showing consistency between the observation and the CCM experiment. The convergence anomaly of the E-P flux is consistent with the poleward anomaly of the residual mean meridional circulation, shown as vectors in Figures 4b and 4d. Then, an upward motion anomaly in the north of the wave forcing and a downward motion anomaly in the south occur. The anomaly in the vertical component of the residual circulation has 95% statistical significance in the JRA-25 and 99% in the CCMVal2-REFB1 experiment (not shown). Accordingly, a warm anomaly around 30–60°N and 30 hPa is evident with 99% statistical significance in Figures 4b and 4d because the downward anomaly causes adiabatic heating.
 The warm anomaly corresponds to a strong meridional gradient of temperature, which is consistent with the westerly wind anomaly around 60°N shown in Figure 2 by the thermal wind relationship. The westerly wind anomaly around the 60°N implies that the polar jet is strengthened. The strengthening anomaly of the polar jet could suppress the propagation of planetary waves from the troposphere to the stratosphere, which corresponds to the downward and divergence anomalies of the E-P flux in the Arctic stratosphere (Figures 4a and 4c). The E-P flux divergence anomaly is consistent with the circulation anomaly from the north polar region to the midlatitude around 10–50 hPa (Figures 4b and 4d). The responses in the midlatitude and polar regions are larger in the winter hemisphere (Northern Hemisphere) than in the summer hemisphere (Southern Hemisphere), which is consistent with previous studies [Jones et al., 1998; Kinnersley, 1999], suggesting that those responses are associated with not only the QBO-induced mean meridional circulation [e.g, Plumb and Bell, 1982] but also, more importantly, the anomaly in wave activity due to the shift of the critical line.
 These results suggest that the difference in latitude of the critical line around 10 hPa associated with the QBO phase is related to the zonal wind anomaly in the midlatitudes and high latitudes through planetary wave propagation. The results further suggest that the zonal wind anomaly in the northern high latitudes could maintain the anomaly through the wave mean flow interaction at the midlatitudes and high latitudes. A schematic diagram for this explanation is shown in Figure 5.
4.1. Results of the CNTL Experiment
 All the results reported in section 3 are from the CCMVal2-REFB1 experiment that includes the effect of volcanic eruptions and SST variations. The CNTL experiment excluding these effects also indicates very similar results for the anomalies of the zonal wind, E-P flux, the E-P flux divergence, the residual mean circulation, and the temperature to those in Figures 2 and 4, with insignificant differences in magnitude. Thus we conclude that the unrealistically large ozone response to the volcanic eruptions in the tropical lower stratosphere in the CCMVal2-REFB1 experiment does not significantly affect the results and interpretations made in section 3 but, rather, that the anomalies are mainly caused by the QBO.
4.2. Contribution of Ozone Variability to the Responses in the Extratropics
 CCM calculates an ozone budget through the transport and chemical production/destruction, and the results in CCM may reflect some interaction processes between ozone chemistry, radiation and dynamics. In order to validate the contribution of ozone variability to the extratropical anomalies associated with QBO, the anomaly in ozone volume mixing ratio is analyzed (Figure 6a). It is obvious that the statistically significant negative anomaly in the north polar stratosphere below 10 hPa is related to the upward motion anomaly shown in Figure 4d, which suggests the negative anomaly in ozone transport into the polar vortex during the westerly phase of the QBO. As shown in Figure 6b, however, the heating effect of ozone is negligibly small because of the little sunlight in the polar night during winter. The analysis of the heating budget indicates that the adiabatic cooling anomaly (Figure 6d) induced by the upward motion in the north polar region is almost balanced with the longwave (terrestrial infrared) heating anomaly (Figure 6c). Ozone is not a dominant factor of the longwave radiation in the stratosphere. Hence it is not possible that the ozone decrease anomaly in the north polar region contributes significantly to the polar temperature anomaly in Figure 4d. Similarly, a heating anomaly around 30°N, 20–50 hPa in Figure 4d is in a balance with dynamical heating due to downward motion anomaly and the longwave cooling, with a small contribution from the ozone heating anomaly. These results indicate that the contribution of ozone variability to the extratropical signal is small.
4.3. An Analysis Based on the 10 hPa QBO
Figure 2 indicates that the zonal wind data around 10 and 50 hPa are nearly anticorrelated, thus the mechanism suggested in section 3 should also be valid in a composite based on the zonal wind direction around 10 hPa. We examined the anomalies in the composite based on 10 hPa. The results show that these anomalies are nearly the same in magnitude with opposite sign to those calculated based on the 50 hPa QBO, although the statistical significance of zonal wind in the high latitude is lower than that of Figure 2 with 90% in the JRA-25. Thus, our conclusions are also valid in the analysis that is based on the 10 hPa QBO.
4.4. The Effect of the Critical Line at 50 hPa on the Polar Vortex
 In this paper, we suggest a mechanism through which the QBO at 10 hPa could influence the polar vortex, which has a consistency with the circulation anomaly below this altitude through the downward control [Haynes et al., 1991] and with the warm anomaly around 30°N, 20–50 hPa. At 50 hPa, however, the same mechanism does not work as an explanation, because the signs are opposite; that is, the warm anomaly cannot be induced by the E-P flux divergence anomaly at 30°N, 50 hPa. We did not find any clear mechanisms by our CCM experiments and analyses how the anomalies due to the shift of the critical line at 50 hPa have an influence on the polar vortex.
Naoe and Shibata  showed the poleward anomaly in E-P flux at 40°N and 50 hPa in the W–E composite from the MRI CCM simulation. Our CCM simulation also indicates a similar anomaly. However, this is not necessarily the evidence for explaining that the QBO effect at 40°N and 50 hPa comes from the other altitudes (e.g., from 10 hPa), because, for example, it might be possible to explain the poleward anomaly in the easterly phase by enhanced wave amplitude around the poleward side of the critical line at the same pressure level. The cause of the poleward anomaly at 50 hPa and the relationship with the polar vortex are still unclear.
 It is also noteworthy that theoretically a diagnostic study in this work cannot establish which altitudes are causal for the vortex response; it just shows a consistency among several physical processes. Hence we conclude that anomalies induced by QBO at 10 hPa have a consistency with the interaction with the polar vortex while the exact mechanism of 50 hPa QBO influence is unclear but the possibility is not ruled out.
5. Concluding Remarks
 In this study, we examined the effect of zonal wind around the equatorial stratosphere on the extratropical Northern Hemisphere stratospheric winter using the JRA-25 data and the outputs from the CCSR/NIES CCM experiments. When the zonal wind around the equatorial 50 hPa is in the westerly phase of the QBO, the zonal wind around 10 hPa is easterly, which shifts the critical line into the Northern Hemisphere. Then, planetary waves cannot propagate to the equator and instead create a convergence anomaly of the E-P flux around 30°N, while a divergence anomaly around 50 hPa, 30°N results from further equatorward penetration of the planetary wave in the westerly wind region around this altitude. The convergence anomaly around 10 hPa is consistent with the warm anomaly over the midlatitude stratosphere and the westerly wind anomaly around the 60°N. Furthermore, the westerly anomaly could maintain the anomaly through the wave mean flow interaction at the midlatitudes and high latitudes.
 It is noteworthy that the diagnostic study in this work cannot establish which altitudes are causal for the vortex response. Hence the above results indicate that anomalies induced by QBO at 10 hPa has a consistency with the interaction with the polar vortex while the mechanism of 50 hPa QBO influence is unclear but the possibility is not ruled out.
 The QBO may have an effect on the troposphere. Naito and Hirota  showed that the variations of tropospheric temperature and zonal wind in the Northern Hemisphere midlatitudes and high latitudes were related to the QBO phase variation. More numerical experiments using CCM that generates QBO internally may be useful to clarify the stratosphere-troposphere coupling and the climate effect associated with the QBO.
 The authors thank anonymous reviewers for their useful comments that improved the manuscript. They also thank R. J. Wilson in Geophysical Fluid Dynamics Laboratory for helpful discussion and correction of English in the manuscript. The Grid Analysis and Display System (GrADS) was used to draw Figures 1–4 and 6. The CCM calculations were performed using a supercomputer system (NEC SX-8R/128M16) at the Center for Global Environmental Research (CGER), National Institute for Environmental Studies (NIES). This work was supported by the Global Environmental Research Fund (GERF) of the Ministry of the Environment (MOE) of Japan (A-071 and A-0903) and by grants-in-aid for scientific research from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan (17340140 and 19340138) and partly supported by a grant-in-aid for the Japan Society for the Promotion of Science (JSPS) Fellows and JSPS Asian CORE Program. The reanalysis data set used for this study was provided from the cooperative research project of the JRA-25 long-term reanalysis by the Japan Meteorological Agency (JMA) and Central Research Institute of Electric Power Industry (CRIEPI).