This study investigates the interannual variability of zonal mean temperature in the wintertime Northern Hemisphere stratosphere and mesosphere measured by Aura-MLS (Microwave Limb Sounder). Results show that the wintertime Arctic temperature is modulated by the phase of the equatorial quasi-biennial oscillation (QBO) wind. On the whole, the 40-hPa QBO easterly phase corresponds to a warmer (colder) northern polar stratosphere (mesosphere) and vice versa. Accordingly, composite differences show that the planetary waves in the winter Arctic stratosphere and lower mesosphere are stronger when the equatorial 40-hPa QBO in its easterly phase. The presented findings are consistent with the established relationship between the QBO phase and the northern winter polar vortex.
 The state of stratospheric polar vortex dominates the dynamical and chemical features in the wintertime Northern Hemisphere (NH) polar middle atmosphere. Both observational and modeling studies have established that the NH winter polar vortex is modulated by the phase of the equatorial quasi-biennial oscillation (QBO). When the equatorial winds near 40 or 50-hPa are easterly, the wintertime stratosphere often experiences a weaker polar night jet and warmer polar temperature than when the QBO is in its westerly phase [e.g., Holton and Tan, 1980; Dunkerton and Baldwin, 1991; Gray et al., 2001].
 In the wintertime Arctic, a cool polar mesosphere is often associated with a warm polar stratosphere and vice versa [e.g., Xu et al., 2009]. Considering the QBO effect on the winter NH stratosphere, a QBO-like modulation of temperature is to be expected in the winter NH polar mesosphere. Until very recently there was little observational evidence for the QBO modulation of zonal mean temperature in the polar mesosphere. With the launch of upper atmosphere research satellite missions that have slow or no (sun-synchronous) precession in local time, remote sensing instruments onboard these satellites have opened up new possibilities for the observations of the global temperature in the stratosphere and mesosphere. Huang et al.  addressed the QBO amplitude of the stratospheric and mesospheric zonal mean temperature measured by the SABER on TIMED (2002–2004) and by the MLS on UARS (1992–1994) using a least squares fitting procedure. Pancheva et al.  applied a similar approach to the SABER/TIMED temperature measurements between 2002–2007 to obtain the QBO variability of the zonal mean temperature and migrating temperature semidiurnal amplitude in the mesosphere and stratosphere. However, these studies are limited to low to middle latitudes since TIMED and UARS cannot provide the continuous coverage for polar latitudes due to 180° yaw maneuvers.
Karlsson et al.  found a correlation between the winter stratospheric temperatures and summer mesospheric temperatures based on the interannual variability of the monthly means from the extended Canadian Middle Atmosphere Model (CMAM). Xu et al.  demonstrated that such an interhemispheric link could be reproduced using the temperatures measured by the Microwave Limb Sounder (MLS) instrument on the Aura satellite. The global temperature-correlation figures [Xu et al., 2009] showed variations from winter to winter, which were considered to be associated with the interannual variability of the winter polar vortex. As a complementary work to Xu et al. , this paper will examine the relationship between the phase of equatorial wind and interannual variability of the polar temperature measured by Aura-MLS.
2. Data Description
 The Aura satellite is in a sun-synchronous orbit at ∼705 km, sampling 82°S to 82°N with an ascending node at 13:45 local solar time (LST) and a descending node at 01:45 LST. The Aura-MLS instrument observes thermal emission at millimeter and sub-millimeter wavelengths. For the Aura-MLS version 2.2 temperature data used here, the useful pressure range is 316-0.001 hPa (∼8–97 km) with precision ranging from 0.6 to 2.5 K and vertical resolution ranging from 3.5 to 15 km [Schwartz et al., 2008].
 The Aura-MLS measurements became operational starting in August 2004, so this work will discuss the interannual variability only for the past 6 winters (2004/2005 to 2009/2010). Figure 1 provides the time-sequences of the Aura-MLS observed temperature averaged over the entire polar cap north of 70°N at 10-hPa and 0.01-hPa for the 6 winters. A substantial interannual variability of the polar temperature in the wintertime NH stratosphere and mesosphere is illustrated in Figure 1. This is not surprising since there is evident interannual variation in the stratospheric polar vortex. The 2004/2005 winter stratosphere was cold with a strong polar vortex in December and January. In contrast, the winters 2005/2006, 2008/2009 and 2009/2010 had the major sudden stratospheric warmings (SSW) in mid- or late-January characterized by displacement or splitting of the vortex [Manney et al., 2009]; whereas the 2006/2007 and 2007/2008 winters were typified by FUB (Free University of Berlin) statistics as having a major SSW in February.
 The interannual variability of vortex structure and dynamics is expected to be correlated with the equatorial QBO [Holton and Tan, 1980; Dunkerton and Baldwin, 1991]. However, as noted by Dunkerton and Baldwin  the SSW effect on interannual variation depends on the timing and duration of the warming, so we need to choose an appropriate time interval when correlating the temperature anomaly with the equatorial QBO data. In the past 6 winters, the major SSWs occurred in either January or February. In order to maximize the contribution of SSWs to the temperature anomaly this study focuses on the monthly mean temperature of January, which will be compared with the monthly mean QBO wind data. For other individual months (December or February) and a 3-month (December, January and February, DJF) or 2-month (DJ or JF) average, the SSW effect on the temperature anomaly would be mitigated since the stratospheric temperatures were often extremely low before or after the duration of a major SSW (e.g., the winters 2005/2006, 2008/2009, and 2009/2010, Figure 1). Figure 2a shows the standard deviations of the January mean values for Aura-MLS zonal mean temperature, which are for the interannual variations over 6 years (2005 to 2010). As expected, the standard deviations standing for interannual variability of the January temperature are large (≥3 K) in the NH polar region.
Figure 2b provides the correlation between the January values for 40-hPa QBO wind and Aura-MLS zonal mean temperature at each pressure/latitude grid for period 2005–2010. The monthly mean QBO wind data observed at Singapore (1°N, 104°E) are obtained from the Institute of Meteorology, Free University of Berlin (http://www.geo.fu-berlin.de/en/met/ag/strat/produkte/qbo/index.html). The 40-hPa QBO winds are selected since they can reflect the optimum northern hemisphere QBO effects [Dunkerton and Baldwin, 1991; Baldwin and Dunkerton, 1998]. The 40-hPa QBO winds are negatively (positively) correlated with the zonal mean temperatures in the Arctic stratosphere (mesosphere) (Figure 2b). A Monte-Carlo shuffling method is applied to estimate the significance of correlation. The correlations are significant at the 90% level in the Arctic mesosphere, while the correlations in the Arctic stratosphere are less significant. The latter is consistent with the imperfect relationship between the QBO wind and the polar stratospheric temperature (shown later). The correlations in the NH polar region in Figure 2b are stronger than those (not shown) using other individual months (D or F) or a 2–3 month average (DJF, DJ, or JF).
Figure 3 shows the time series for the interannual variations in the January values of Singapore 40-hPa zonal wind (red), Aura-MLS average temperature poleward of 70°N at 0.01-hPa (blue) and 10-hPa (green), and UKMO (United Kingdom Meteorological Office, also known as MetO) zonal mean zonal wind at 10 hPa, 61°N (black).The time sequences in Figure 3 are consistent with the correlation results in Figure 2. Overall, the easterly (westerly) equatorial QBO wind (Figure 3, red) is associated with a weaker (stronger) zonal wind (Figure 3, black) and higher (lower) temperature in high-latitude stratosphere (Figure 3, green). Such a QBO effect agrees with the established knowledge [e.g., Holton and Tan, 1980; Dunkerton and Baldwin, 1991]. Due to the stratosphere-mesosphere connection, correspondingly, the colder (warmer) polar mesosphere usually occurred in the easterly (westerly) QBO phase (Figure 3, blue). However, note that the relationship between the QBO wind and the polar stratospheric temperature is not perfect (Figure 3, red and green): the QBO effect was of opposite sign (i.e., the easterly QBO wind is associated with colder stratosphere) for January 2008 and January 2010. Associated with the lower temperatures, stronger zonal winds occurred in high-latitude stratosphere for January 2008 and January 2010 (Figure 3, black) when the equatorial winds were easterly. Similarly, the association between the polar mesospheric temperature and the QBO wind is not perfect for January 2008 and January 2010 (Figure 3, blue and red), although the tendency of the temperature anomaly was correct.
 In Figure 4, we provide the composite difference (easterly minus westerly phase) of the January values for Aura-MLS zonal mean temperature and temperature planetary wave number 1 (PW1) amplitude when these data are partitioned according to the phase of the equatorial QBO wind at 40-hPa. As expected, the Arctic temperatures are significantly lower (higher) in the mesosphere (stratosphere) during the easterly QBO phase than during the westerly QBO phase. The PW activity in high-latitude stratosphere and low mesosphere is substantially enhanced when the equatorial QBO in its easterly phase. This is consistent with the accepted explanation for the NH QBO effect: the easterly equatorial winds can effectively block the equatorward propagation of the winter hemisphere's PW activity; the latter tend to propagate higher and more poleward during periods of the easterly equatorial winds than during those with the westerly equatorial winds. Consequently, the polar vortex is more likely to be disturbed by PWs when the 40-hPa QBO winds are easterly, increasing the likelihood of SSW events and a warmer (colder) polar stratosphere (mesosphere). It should be noted that consistency between composite differences of zonal mean circulation anomalies and wave amplitudes is expected because these quantities are not independent [Dunkerton and Baldwin, 1991].
 This study shows a link between the phase of equatorial QBO wind and interannual variability of the wintertime NH polar temperature measured by Aura-MLS, with easterly QBO wind at 40-hPa correlated with a colder (warmer) polar mesosphere (stratosphere) and vice versa. The observation of such a QBO effect is consistent with the previous studies for the relationship between the equatorial QBO phase and the NH polar vortex. The mechanism behind this link involves the modulation of the vertical and meridional propagation of PW activities originating in the troposphere by the phase of equatorial zonal wind [Holton and Tan, 1980]. Note that the sample size is small in this case and the issue is likely to be better resolved with a longer observation data record.
 Consistent with the imperfect association between the extratropical and tropical QBOs for the longer record [Dunkerton and Baldwin, 1991; Lu et al., 2008], we also notice that the correlation between the QBO phase and the Aura-MLS polar temperature anomaly is not perfect. Considering the timing effect of stratospheric warming, e.g., the duration of the stratospheric disturbance is too short during January to dominate the monthly mean temperature anomaly for 2008 and 2010 (Figure 1), we could see an increased correlation if some averaging periods finer than monthly scale are selected. However, note that the imperfect correlation between the QBO phase and the polar temperature is not surprising because the equatorial QBO phase is not the only factor leading to interannual variability of the polar vortex.
 Funding for this work was from CANDAC-PEARL. Thanks are given to the Aura team for their MLS dataset as well as to the Institute of Meteorology, FUB for providing access to the QBO wind data. We are also grateful to the UKMO and the BADC for providing access to the stratospheric assimilated data. Sincere thanks to the two anonymous reviewers for their helpful comments on the original manuscript.
 The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.