It is well known that the interannual variability of the stratospheric winters over the Arctic is very large. On the basis of data for more than 60 winters, this variability has been studied with the aim of understanding and possibly forecasting the type of the coming winter, in the stratosphere and also in the troposphere. Today, there is general agreement that the variability of the stratospheric circulation during the Arctic winters is influenced by different forcing mechanisms: by the tropospheric planetary waves which penetrate into the stratosphere, by the Quasi-Biennial Oscillation (QBO) and the Southern Oscillation (SO) in the tropics which influence the stratospheric polar vortex, and by the 11-year sunspot cycle which interacts with the QBO and probably also with the SO. For the winter 2008/2009, all of the known signals pointed to a stable, cold stratospheric polar vortex throughout the winter, but in the real atmosphere a major midwinter warming developed in January and February with record-breaking temperatures. The synoptics of this winter will be discussed in the context of all of the above-mentioned forcing mechanisms.
 One of the key questions of Climate and Weather of the Sun-Earth System (CAWSES) is to understand the variability of the Arctic winters and to try to predict the influence of the stratospheric processes on the troposphere. Meteorological data (temperatures and winds) from the Arctic are now available for more than 60 years. They are mainly derived from the reanalyses made by NCEP/NCAR [Kalnay et al., 1996]. Figure 1a shows the monthly mean 30-hPa temperatures over the North Pole in January (since 1948), and Figure 1b gives the monthly mean 30-hPa heights in February, since 1942 [Labitzke et al., 2006]. The large variability between winters is obvious: very warm winters, like, for example, January 1970, 1985 and 2006 alternate with very cold winters, like, for example, 1976 and 2000. This leads to large standard deviations (sigma = 8 K over the North Pole in January, or about 500 gpm in February). In February 2009 the 30-hPa heights over the North Pole reached a new (positive) record (Figure 1b), although the forecasts were for a record minimum. The variability of the Arctic winters is influenced by different factors, for example, (1) the planetary waves which penetrate from the troposphere into the stratosphere [Geller and Alpert, 1980]; (2) the Quasi-Biennial Oscillation (QBO) [Holton and Tan, 1980]; (3) the Southern Oscillation (SO) in the tropics and its respective Warm Extremes (WE) or Cold Extremes (CE) which are connected to a warmer or colder polar stratosphere [van Loon and Labitzke, 1987]; and (4) the 11-year sunspot cycle (SSC) which changes the sign of the influence of the QBO during solar maxima [Labitzke, 1987]; [Labitzke and van Loon, 1988]. But other factors such as internal variance also play an important role.
2. Evolution of the Winter 2008/2009
 Time series of daily data of selected parameters are given in Figures 2 and 3 in order to describe best the large variability during this winter. The first and second time series in Figure 2 show the daily temperatures over the North Pole for the 10- and 30-hPa level, respectively. The dashed lines are the long-term mean (30 years) of the daily data. The very strong and fast warming over the polar region is the main event of the winter: at the 10-hPa level the data show an increase of the temperatures of about 70 K within about a week, and the temperatures stay above the long-term mean for 1 month. The next three plots give the amplitudes of the planetary waves 1 and 2 at 60°N, for the 10- and 30-hPa temperatures and for the 30-hPa heights. Until the beginning of January the circulation is determined by the planetary wave 1, but this wave was not particularly large as compared with other winters. The temperature wave 2 was weak, and the height wave 2 similar to wave 1. From about 7 January the circulation is dominated by wave 2, in the height and temperature fields, and this continues for about 1 month (see Figures 3 and 4).
Figure 3 shows derived quantities (daily values, 1 November 2008 to 31 March 2009) of different dynamical parameters which are generally used to interpret the dynamics of the stratospheric circulation. Figure 3a shows the daily data of the zonal mean zonal winds at 60°N at the 1-hPa level (at a height of about 42 km). Here, the wind reaches a maximum of 90 m/s from the west on 7 January. This wind decreased rapidly after 16 January, changed within a few days to east (40 m/s) and increased afterward slowly during February.
Figure 3b shows the zonal mean zonal winds at the 10- and 30-hPa levels, in 60°N, at a height of about 28 and 22 km, respectively. The development of the winds at these levels is clearly very similar to the winds at the 1-hPa level, and the change to easterlies is only few days later. After the breakdown of the polar vortex (Figures 4 and 5) the winds remain weak through the rest of this winter.
 The heat and geopotential fluxes (Figure 3c), as well as the Eliassen-Palm vector (60°N/10 hPa; Figure 3d) are useful derivatives which give some indication of the coupling between the troposphere and the stratosphere [Eliassen and Palm, 1961]. During periods of strong coupling, i.e., when the planetary waves are strong, the fluxes are strong as well. If a breakdown of the polar vortex is completed, the fluxes decrease to very small values (only weak winds).
2.1. Description of the Synoptic Development of the Major Midwinter Warming
Figure 4 shows synoptic 10-hPa maps (geopotential heights to the left and temperatures to the right) for three selected days which highlight the most important phases of this winter, in the middle stratosphere. Until 8 January 2009 (Figure 4, top) the circulation in the stratosphere is undisturbed with a strong cyclonic vortex and very weak planetary wave activity (Figures 2 and 3). The zonal mean zonal winds at 60°N have reached the maximum of this winter (Figure 3b), and the vortex is very cold with a minimum of −93°C at 10 hPa over Iceland (Figure 4, right).
 Within the next 2 weeks the major midwinter warming (MMW) developed and on 24 January (Figure 4, middle) the criteria for a MMW are fulfilled: A stratospheric warming is called “major” if the gradients of the temperature and of the mean zonal winds are reversed at the 10-hPa level, between 60°N and the North Pole. The temperature over Iceland rose from −93°C on 8 January to −20°C on 24 January, while the main warming is situated over Greenland, with plus 8°C.
 The increase in wave activity during the development of the MMW is documented best with the EPV (Figure 3d, bottom), as well as the decrease during February and March which hints at a complete decoupling of the stratosphere from the troposphere. On 1 February 2009 (Figure 4, bottom), a strong anticyclone has developed over the Arctic, with temperatures close to summer conditions. This complete reversal from the winter to a summer circulation lasted well into April and the vortex did not recover to full strength, as seen in Figure 3b and resulted in 30-hPa heights almost 3 standard deviations above those known for the last 67 winters (Figure 1b).
2.2. Anomalies From December Through March
 When looking for an explanation or early indication of such a strong event with such a long time scale, it is imperative to look at the data available before the event. In Figure 5 we compare, for instance, anomalies of the 30-hPa temperatures from December through March of four winters, selected by the criteria of QBO-west and solar minimum, i.e., similar to the winter 2008/2009, which have been shown by, for example, Labitzke , Labitzke and Naujokat  and van Loon and Labitzke  to be prone to a cold, undisturbed winter. Obviously, the anomalies are similar in all four winters, all resulting in a cold and undisturbed late winter.
 For comparison, we show in Figure 5 (bottom) the anomalies of the winter under investigation, and the difference between the four cold winters and the winter 2008/2009 is very clear. With the current understanding, based on 67 winters, it was not possible to imagine the developments which took place.
3. Global Dimension of the MMW
 Our description of the MMW was so far limited to the Northern Hemisphere and mainly to the middle stratosphere. But in order to comprehend the global scale of these events it is mandatory to use the NCEP/NCAR reanalyses which are available since 1948 from the ground up to 10 hPa [Kalnay et al., 1996]. Figure 6 shows, for February 2009, vertical meridional sections of the monthly mean anomalies: Figure 6 (top) for the temperatures (K), Figure 6 (middle) for the geopotential heights (gpdam), and Figure 6 (bottom) for the zonal mean zonal winds (m/s). The MMW is, of course, the dominating feature, with positive temperature anomalies larger than 3 sigma in the Arctic tropopause region (about 300 hPa). The whole middle and lower stratosphere shows positive anomalies larger than 2 or 1 sigma over the Arctic, and at the same time a large region with negative anomalies above 2 sigma in the tropics and subtropics, far into the Southern Hemisphere [Kodera and Kuroda, 2002].
 The height anomalies are consistent with the temperature anomalies and lead to the wind anomalies which reflect the weakening (and reversal; see Figure 3) of the stratospheric polar night jet stream. The large stratospheric warmings over the Arctic indicate strong downwelling, which is compensated by concurrent upwelling (cooling) outside the Arctic, far into the Southern Hemisphere [Salby and Callaghan, 2004]. Through these interactions the state of the circulation during the northern winters determines the conditions outside the Arctic [Labitzke and Kunze, 2009].
 Over the equator the QBO is well pronounced with positive anomalies, indicating west winds, in the upper troposphere and lower stratosphere (up to 20 hPa), and negative anomalies, indicating east winds, above.
4. Where Do We Stand?
 As discussed at the beginning, the Arctic winters are influenced by different factors. Up to now, we have (at best) data for 68 winters. According to the discussion above, the winter studied here should have been cold and undisturbed, because the QBO is in the west phase (at about 45 hPa, in January and February [Holton and Tan, 1980]). On the other hand, we are in the minimum of the 11-year solar sunspot cycle (SSC) and according to statistical investigations we could expect a cold undisturbed winter [e.g., Labitzke and van Loon, 1988; van Loon and Labitzke, 1994; Labitzke, 2006]. The SO was in the neutral state during the beginning of the winter and should not have an influence. Figure 7 shows two scatter diagrams with the monthly mean 30-hPa heights in February (Figure 1b) plotted against the SSC (given here is the 10.7 cm solar flux, a radio wave which is highly and positively correlated with the UV part of the solar spectrum [Hood, 2003]). The data are grouped into winters in the east phase of the QBO (Figure 7, left) and winters in the west phase of the QBO (Figure 7, right).
 It is obvious that the correlations are opposite in the two different phases of the QBO: while the winters in the east phase of the QBO are negatively correlated, the winters in the west phase are positively correlated. But as far as February 2009 is concerned, the value (and the whole winter) is an outlier: it should be with the coldest winters (Figure 7 (right), left and low down) according to Holton and Tan  and Labitzke and van Loon . February 1997 is another outlier with the opposite facts. During the winter 1996/1997 the SSC was in a minimum and the QBO was in its east phase. Thus, the statistical analyses of Figure 7 indicate a larger probability for a disturbed stratospheric vortex (i.e., larger values of the geopotential height at the North Pole) in February 1997.
 These outliers show the importance of the internal variability which so far is not completely understood.
 (1) The QBO and SSC explain a significant portion of the variance of the winter monthly values of Arctic temperatures, winds, and geopotential heights. (2) The behavior of the Arctic stratosphere is strongly related to the QBO phase and the SSC, yet other factors such as internal variance also play an important role. (3) The Arctic 2008/09 winter is an excellent example of how these other factors, such as internal variability, can sometimes overwhelm the strong statistical relationships involving the QBO and the SSC. (4) This emphasizes the probabilistic nature of forecasting Arctic stratosphere behavior.
 We acknowledge the suggestions made by reviewer 1 (Marvin Geller) which we incorporated in our text. We also thank the members of the Stratospheric Research Group, FUB, for professional support. The 10.7-cm solar flux data are from the World Data Center A, Boulder, Colorado. NCEP reanalysis data are provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, from their Web site at http://www.cdc.noaa.gov/.