We investigate the influence of major sudden stratospheric warming (SSW) and elevated stratopause (ES) events in the Northern Hemisphere winter on the transport of NOx produced by energetic particle precipitation (EPP) from the mesosphere–lower thermosphere to the stratosphere using the Whole Atmosphere Community Climate Model (WACCM). Increases in NOx following a major SSW and/or ES event are in excess of 100% compared to winters when no major SSW or ES event occurred. The increase in NOx is attributed to an increase in the descending branch of the residual circulation () following the event. The timing of the event strongly affects the amount of NOx that descends to the stratosphere: the earlier the event occurs, the more NOx descends to the stratosphere. We also quantify the amount of NOx produced by EPP descending to the stratosphere in each winter and find that the largest increases in NOx are in years that have a major SSW followed by an ES event early in the season (December or early January). The strength of following an event shows a very strong seasonal dependence and explains why the timing of the event affects the transport of NOx.
 Energetic particle precipitation regularly produces nitric oxide (NO) in the mesosphere and lower thermosphere [e.g., Crutzen, 1979; Thorne, 1980]. The photochemical lifetime of NO is only a few days in the sunlit mesosphere; however, in the polar night, NO can persist for months and be transported to the stratosphere without being photochemically destroyed. As NO descends in the polar region, a portion is converted to nitrogen dioxide (NO2) through reaction with ozone (O3). When NOx (NO + NO2) created by energetic particle precipitation (EPP-NOx) reaches the stratosphere, it participates in catalytic destruction of O3, which, since O3 is a radiatively active gas, can affect the thermal structure of the middle atmosphere. The process whereby EPP influences the stratosphere indirectly is called the EPP indirect effect (EPP IE) [Randall et al., 2006]. This is in contrast to EPP that directly affects the stratosphere when higher energy electrons and protons deposit their energy in situ.
 The amount of EPP-NOx that descends to the stratosphere, i.e., the strength of the EPP IE, depends on both the level of EPP and atmospheric transport. This dependency results in pronounced hemispheric differences. For example, the interannual variability in dynamics in the Southern Hemisphere (SH) is small, and the SH winter stratosphere is characterized by a strong and steady polar vortex. Randall et al.  showed that the amount of EPP-NOx descending to the SH winter stratosphere is highly correlated with several measures of EPP activity, such as the Ap index and medium energy electron hemispheric power. In the Northern Hemisphere (NH), where higher planetary wave activity means that minor stratospheric warmings are common and a major sudden stratospheric warming (SSW) event occurs roughly every other year, the amount of EPP-NOx descending to the stratosphere is strongly influenced by dynamics. For example, Sinnhuber et al.  found a positive correlation between the AE index and mesospheric NOx in the NH between December and March, but the correlation abruptly stops in March and does not extend into the stratosphere. They attribute the breakdown of the correlation to the high dynamical variability in the NH.
 The stratopause lowers in altitude during a major SSW event; following this, the stratosphere becomes isothermal for a few days before the stratopause reforms. After some major SSW events, the stratopause reforms as high as ∼80 km, an altitude normally associated with the mesosphere. This is referred to as an elevated stratopause (ES) event and has received considerable attention in the most recent decade [e.g., Manney et al., 2008, 2009a, 2009b, 2010]. ES events occurred in the NH winters of 2003–2004, 2005–2006, and 2008–2009 and had a pronounced impact on the descent of polar NOx [e.g., Hauchecorne et al., 2007; López-Puertas et al., 2006; Randall et al., 2005, 2006, 2009; Siskind et al., 2007]. It is particularly interesting that the level of EPP was relatively low in 2005–2006 and 2008–2009, yet the amount of EPP-NOx reaching the upper stratosphere was just as high or higher than that in years with a higher level of EPP [e.g., Holt et al., 2012; Randall et al., 2006, 2009; Seppälä et al., 2007], highlighting the importance of dynamics in the NH.
 The goal of the present study is to further our understanding of the mechanisms controlling the transport of EPP-NOx in the polar winter using the Whole Atmosphere Community Climate Model (WACCM). Specifically, we investigate the effects of major SSW and ES events on the EPP IE. Section 2 briefly describes the model and simulations used here. Section 3 describes the methods used to identify major SSW and ES events in WACCM. In section 4, we discuss differences in polar NOx evolution with respect to month of major event occurrence. We also quantify the EPP IE and compare years with no major SSW or ES event, years with a major SSW or ES event, and years with a major SSW event followed by an ES event. Section 5 includes the summary and conclusions.
2 Model: Whole Atmosphere Community Climate Model
 WACCM is a global, 3-D climate model developed by the National Center for Atmospheric Research (NCAR) that extends from the Earth's surface to the thermosphere [Marsh et al., 2013]. WACCM is an optional superset of the Community Atmosphere Model [Neale et al., 2010], which is the atmospheric component of the Community Earth System Model. WACCM has 66 pressure levels ranging from the surface to 5.1×10−6 hPa. The vertical resolution varies and is 3.5 km above ∼65 km, 1.75 km below ∼50 km, 1.1–1.4 km below ∼30 km, 1.1 km in the troposphere, and much higher in the planetary boundary layer. The model resolution of the simulation used in this study is 1.9°×2.5° (latitude × longitude).
 WACCM includes fully interactive chemistry, radiation, and dynamics. The chemistry module is based on the Model for OZone And Related chemical Tracers and is described in detail by Kinnison et al. . Processes of the mesosphere/lower thermosphere (MLT) are based on the NCAR thermosphere-ionosphere-mesosphere electrodynamics general circulation model (TIME-GCM), which includes a parameterization for auroral EPP [Roble and Ridley, 1987]. The parameterization inputs hemispheric power, which itself is parameterized by the Kp index, and outputs ion-pair production rates. The reader is referred to Marsh et al.[2007, 2013] for an in-depth description of auroral and solar forcing in WACCM. For detailed descriptions of WACCM gravity wave drag and vertical diffusion parameterizations, see Garcia et al.  and Richter et al. .
 Major SSW and ES events commonly occur during Arctic winters in WACCM simulations with a frequency that agrees well with observations. For example, de la Torre et al.  found that the frequency of major SSW events in WACCM is very similar to that found in reanalysis data, although the major SSW events are generally prolonged in WACCM and occur disproportionately often in December. Chandran et al.  found that the frequency of ES events in WACCM agrees well with the frequency of ES events in the Modern Era Retrospective Analysis for Research and Applications (MERRA) data set, although they noted that in WACCM, ES events are more likely to occur with a vortex splitting event while in MERRA, they are more likely to occur with a vortex displacement. The EPP IE is also a self-generated feature of free-running WACCM; however, WACCM underestimates the EPP IE compared to observations [Smith et al., 2011]. Since WACCM has reasonable major SSW and ES events and a self-generated EPP IE, WACCM is a useful tool to study the effects of major SSW and ES events on the EPP IE. Furthermore, with WACCM, it is possible to keep the level of EPP constant in order to isolate the effects of dynamics on the EPP IE.
 In this study, we have used two WACCM version 4 simulations [Marsh et al., 2013]. The runs include forcing for auroral electrons but not solar protons or higher energy electrons. The runs are referred to throughout the rest of the paper as R1 and R2. Each simulation is a 42-year, perpetual year 2000 A.D. (annually repeating) simulation. The runs are identical except for initial conditions. We report the results separately to give an indication of variability. Seasonally varying solar spectral irradiance was specified, based on the model of Lean et al.  as described in Marsh et al. . Sea surface temperatures (SSTs) were also specified, based on monthly mean SSTs constructed from the monthly mean Hadley Centre sea ice and SST data set versions 1 and 2 of the National Oceanic and Atmospheric Administration weekly optimum interpolation SST analysis [Hurrell et al., 2008]. In order to study the effects of major SSW and ES events on the EPP IE, the level of auroral EPP, which is parameterized by the geomagnetic index Kp, was held constant. For the simulations used in this study, the Kp index was set to 4 (Ap=27). For reference, the average of the daily Ap index (obtained from the National Geophysical Data Center) for 2002–2012 is ∼9.7, where the 90th percentile is 21. The first 2 years of each simulation were discarded to remove any influence of model spin-up, so that each run contains 39 NH winters (November through March).
3.1 Major Sudden Stratospheric Warming Detection
 We use the algorithm of Charlton and Polvani  to detect major SSW events in WACCM. They follow the first criterion of the World Meteorological Organization (WMO): the zonal mean zonal winds at 60°N and 10 hPa become easterly during the winter (November–March). They ignore the second WMO criterion, which the temperature gradient is positive between 60° and 90°N, because it does not significantly change the number of major SSW events identified. The first day that this definition is met is defined as the central warming date. Their algorithm also imposes two additional criteria: (1) the zonal mean zonal winds must remain westerly for 20 consecutive days after a central warming date before another can be identified, and (2) the zonal mean zonal wind must return to westerly for at least 10 days before 30 April. If the second condition is not met, the warming is considered to be a final warming, which initiates the transition from winter westerlies to summer easterlies.
3.2 Elevated Stratopause Detection
 We define an ES event using the method described in de la Torre et al.  and Chandran et al. . The method can be summarized as follows. An ES event is identified as an increase of 15 km or more in the stratopause height from one day to the next between November and March. The stratopause is defined as the maximum temperature between 20 and 100 km. The temperature is averaged poleward of 75°N and smoothed with a 9-day running mean to eliminate short-term excursions in stratopause height that result from transient wave forcing. The central date of the ES event is defined as the date of the displacement in the stratopause.
4 Results and Discussion
4.1 Major SSW and ES Events
 Table 1 shows the major SSW and ES events found in each run using the above definitions of major SSW and ES event. The 18 major SSW and 17 ES events occurred in R1, and 11 major SSW and 7 ES events occurred in R2. The frequency of occurrence for major SSW events is 0.46 per winter for R1 and 0.28 per winter for R2, and the frequency of occurrence for ES events is 0.44 per winter for R1 and 0.18 per winter for R2. The frequency of occurrence for both simulations together is 0.37 major SSW events per winter and 0.31 ES events per winter. Most (∼ 71%) ES events occur after a major SSW event. Table 1 also shows the number of times that a major SSW and ES event occur together (SSW-ES). ES events that follow a major SSW event occur 8 days, on average, after the major SSW event with a range of 2 to 15 days. There are 12 SSW-ES events (0.31 per winter) in R1 and 5 SSW-ES events (0.13 per winter) in R2. Although we do not explicitly calculate the effects of minor SSW events on EPP-NOx transport in this study, we note that there were ∼2.0 minor SSW events per winter in R1 and ∼1.8 minor SSW events per winter in R2. We define a minor warming as an increase of at least 25 K within a period of 7 days below 10 hPa. There were only 2 winters in the 78 model years (∼3%) without any minor or major SSW or ES event (i.e., dynamically calm winters).
The frequency of occurrence in number per winter is given in parentheses.
Number of major SSW events
Number of ES events
Number of SSW-ES events
 There are fewer major SSW events per winter in R1 and R2 than Charlton and Polvani  found using reanalysis data from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis and the European Center for Medium-Range Weather Forecasting 40-year reanalysis (ERA-40) for the period of time that the data sets overlap (45 NH winter seasons from November 1957 through March 2002). Using the algorithm described in section 3.1, they found 0.60 major SSW events per winter in the NCEP/NCAR reanalysis and 0.64 major SSW events per winter in the ERA-40 reanalysis. However, de la Torre et al.  found that a subset of NCEP/NCAR from 1960 to 1989 had only 0.47 major SSW events per winter.
Chandran et al.  also applied their ES algorithm to 32 NH winters from 1979 to 2011 from the MERRA reanalysis data set and found an occurrence frequency of 0.31 ES events per winter. The average number of ES events per winter for R1 and R2 combined (0.31 per winter) agrees with Chandran et al. , although either R1 or R2 alone is outside of the range they found.
 Figure 1 shows the monthly distribution of major SSW and ES events for the two simulations. The majority of major SSW events in R1 occurs in December, whereas the majority of major SSW events in R2 occurs in February and March. Charlton and Polvani  found an occurrence frequency of ∼0.1 December, ∼0.2 January, ∼0.17 February, and ∼0.11 March major SSW events per winter. The largest discrepancy in the simulations here compared to Charlton and Polvani  is the frequency of January major SSW events.
 The discrepancy in both the number of major SSW events and the monthly distribution compared to reanalysis might be explained by the specified SSTs used in the simulations here. For example, Richter et al.  showed that a 29-year WACCM simulation with fixed SSTs had 0.5 major SSW events per winter, whereas a 29-year simulation for the same time period with an active ocean had 0.7 major SSW events per winter. This suggests that the way SSTs are handled might have an effect on major SSW event frequency. Richter et al.  also found earlier major SSW event onset in the simulation with specified SSTs compared to the simulation with an active ocean. And, in fact, WACCM simulations with an active ocean described by Marsh et al.  had a monthly distribution in better agreement with observations than the simulations here and the WACCM simulations in de la Torre et al. . For the purpose of this study, the exact frequency is not critical as long as the run is long enough to get enough major SSW events in each month.
 In R1, ES events occur more frequently in December (∼0.18 per winter), followed by January (∼0.15 per winter) and February (∼0.1). The opposite trend is observed in R2: ∼0.03 December, ∼0.05 January, and ∼0.1 February ES events per winter. In the combined frequency of occurrence, ES events occur with equal frequency in each month. Again, the exact frequency of ES events is not critical to this study.
4.2 Major SSW and ES Event Timing and Time Evolution of Polar NOx
 Figure 2 shows composite NOx mixing ratios averaged poleward of 70°N (area-weighted) as a function of pressure and time for years with (Figure 2a) no event, (Figure 2b) a December event, (Figure 2c) a January event, and (Figure 2d) a February event. Winters from the two simulations were placed into one of the four categories. Only winters with a single event are considered here, where an event is either (1) a major SSW event without an ES event, (2) an ES event without a major SSW event, or (3) a SSW-ES event. Winters that have more than one event (e.g., a major SSW event in December and an ES event in February) are not considered. We have also discarded winters with a minor warming within 60 days of the event. There are 42 winters in Figure 2a, 8 winters in Figure 2b, 6 winters in Figure 2c, and 8 winters in Figure 2d. The day of the event is marked with a red line, where the day of the event is defined as follows. For a major SSW event occurring alone or an SSW-ES event, the day of the event is the central warming date of the major SSW event. For an ES event occurring without a major SSW event, the day of the event is the central date of the ES event minus 8 days to account for the lag of the ES event with respect to the reversal of the zonal mean zonal wind. This has been done to make the events comparable. We chose 8 days because this is the average time between the central date of a major SSW event and the central date of an ES event for an SSW-ES event.
 The tongue of descending NOx takes on a different appearance depending on the month of occurrence of the event. A December event has the largest influence on the magnitude and extent in altitude of the NOx tongue. The 12 ppbv contour line (shown in black) almost reaches 1.0 hPa, whereas in the other composites it remains up at ∼0.1 hPa. During the event, the black contour line is pushed up in altitude compared to that in Figure 2a, indicating upwelling. Following the event, the black contour line extends further down in altitude compared to Figure 2a, indicating enhanced downwelling.
 Figure 3 shows the percent difference between years with (Figure 3a) December, (Figure 3b) January, and (Figure 3c) February events and years with no event. The black contours show where the vertical component of the residual, or transformed Eulerian mean (TEM), circulation () is more than ±50% (+= solid lines; −= dotted lines) different than it is in the case without an event. Here is positive for upwelling and negative for downwelling. The increases in NOx usually correspond to the increases in descent, and the decreases in NOx usually correspond to decreases in descent. In all cases, around the time of the event, there is a decrease in NOx above ∼0.8 hPa (upper stratosphere and mesosphere) and an increase in NOx below ∼0.8 hPa (middle stratosphere). This is a result of the changes in the TEM circulation during the event. Smith et al.  show that the TEM in WACCM during a SSW-ES event has a strong poleward and downward component at ∼1 hPa and an equatorward and upward component at ∼0.1 hPa. This reflects the large EP flux convergence near the level of the maximum zonal mean zonal wind deceleration during a major SSW event, which leads to cooling and induced upwelling above and warming and induced downwelling below that level [Matsuno, 1971; Holton, 1983]. Therefore, the increase in NOx below ∼0.8 hPa during the event is largely caused by the influx of lower latitude air with higher NOx mixing ratios, and the decrease in NOx above ∼0.8 hPa during the event is caused by the upwelling and divergence in the mesosphere that reverses the normal winter downwelling. This is supported by the evolution of NOx in the latitude-pressure plane during an event (not shown). The anomalies propagate downward over time following the event.
 After the event, there is an increase in NOx as the vortex reforms (not shown) and strengthens. The largest NOx increase is in the winters with a December event, the second largest is in the winters with a January event, and the smallest increase is in the winters with a February event. The NOx increases following the later events are confined to higher altitudes; namely, the increase in NOx following the event reaches well below 1.0 hPa in Figure 3a, just above 1.0 hPa in Figure 3b, and just below 0.1 hPa in Figure 3c.
 The spread of the timing of the events, even when grouped by month of occurrence, is sufficient to wash out the descending tongue of NOx in the monthly composites. Figure 4 shows NOx mixing ratios averaged poleward of 70°N (area-weighted) as a function of pressure and time for individual model years with different event timings. It shows a year with (Figure 4a) no major SSW or ES event, (Figure 4b) a December major SSW event, (Figure 4c) a January ES event, and (Figure 4d) a February major SSW-ES event. We note that in Figure 4c, conditions for a major SSW event were almost met, viz., the winds almost became easterly (they reach a minimum of ∼1.7 m/s) at 60°N and 10 hPa. The descent of NOx is interrupted during the major event and resumes, often at a faster rate than before the event as indicated by the steeper slope of the contour lines, after the warming ends. Figure 4 confirms what Figures 2 and 3 show: that the timing of the major event impacts the descent of NOx in the manner described above.
4.3 Quantification of Major SSW and ES Event Effects on the EPP IE
 To quantify the impact of major SSW and ES events on the descent of EPP-NOx in WACCM, we first quantify the amount of EPP-NOx in gigamoles (Gmol) descending to the stratosphere each NH winter season using the method described in Holt et al. . We briefly outline the method here. First, in order to separate the portion of NOx that is created by EPP, we use the relationship between NOx and CH4. In the polar winter upper stratosphere, NOx and CH4 are positively correlated since they are both decreasing with increasing altitude. This relationship remains true unless there is a source of NOx, and the only source of NOx in the polar winter upper stratosphere is EPP. Therefore, EPP-NOx can be identified as an anticorrelation between NOx and CH4. This method of identifying stratospheric EPP-NOx transported from the upper atmosphere was first described by Siskind and Russell III . Once we identify the EPP-NOx density at a particular level (in Gmol per m3) and grid box, we determine the daily flux of EPP-NOx (in Gmol per day) through that level by multiplying the excess density in each grid box by the area of the grid box and in that grid box and then summing over all the grid boxes. We then sum the daily flux over 1 November to 31 March to get the total number of Gmol per NH winter across the chosen level.
 Figure 5 shows the flux of EPP-NOx in Gmol per day across three levels in WACCM for years with (Figure 5a) no event, (Figure 5b) a December event, (Figure 5c) a January event, and (Figure 5d) a February event. The three levels shown here are 0.41 hPa (∼ 52 km, black line), 0.73 hPa (∼ 47 km, dark grey line), and 1.24 hPa (∼ 43 km, light grey line). The largest flux of EPP-NOx into the stratosphere at all levels by far happens in winters with a December event. Winters with a January event also have a larger EPP IE than winters with no event. The EPP-NOx flux for winters with a February event is similar in shape but larger in magnitude compared to the winters with no event. The flux in February is interrupted by the event. The winters with a December or January event are also the only winters for which any significant amount of EPP-NOx is transported to the lower levels (light grey). The EPP IE at 1.24 hPa is nonzero for a small number of dynamically calm winters (not shown) but the magnitude is much smaller (a factor of ∼ 58).
 Table 2 compares the EPP IE at 0.41 hPa, 0.73 hPa, and 1.24 hPa, as defined by the average amount of EPP-NOx descending across each level, for winters with no major SSW or ES event, winters with an event, and winters with a SSW-ES event only. The breakdown month of event and SSW-ES event is also shown. The largest EPP IE at 0.41 hPa for a winter without an event is 0.073 Gmol compared to 0.16 Gmol for a winter with a SSW-ES event. On average, the largest EPP IE occurs in winters with a SSW-ES event. The average EPP IE at 0.41 hPa for winters with a SSW-ES event is ∼3 times larger than the EPP IE for winters with no event. At the lower level (1.24 hPa), the average EPP IE for winters with a SSW-ES event is ∼26 times larger than the average EPP IE for winters with no event. In winters with an event, the largest EPP IE at all levels occurs when the event happens in December, followed by January, February, and March; i.e., in general, the earlier the event, the larger the EPP IE. In fact, the maximum EPP IE for winters with no event is larger than the maximum EPP IE for winters with an event when the event occurs after December, and the average EPP IE for winters with no event is similar to the average EPP IE for February and March events. The maximum EPP IE for winters with no event occurs for a dynamically calm winter with only one minor SSW event at the end of February.
Table 2. Average EPP-NOx (Gmol) Across Three Levelsa
No event 
The number of winters in each category is shown in brackets. Maximum values are shown in parentheses.
All events 
SSW-ES events only 
 The relationship between the timing of the major event and the EPP IE (the total amount of EPP-NOx descending across each level over the entire winter) is shown in Figure 6. Figures 6a–6c show the EPP IE for all events; i.e., the EPP IE from winters that formed the composites in Figures 2b–2d. Figures 6d–6f show the EPP IE for winters with a SSW-ES only. There is a strong anticorrelation between the amount of EPP-NOx and the central warming date, especially for SSW-ES events. At 0.73 hPa and 1.24 hPa, the relationship is no longer linear, so the correlation underestimates the relationship. Similar relationships exist between central warming date and maximum NOx mixing ratios at each level (not shown). The years that have the largest EPP IE are years in which a major SSW event took place in December and, in all but one year, was followed by an ES event. The relationship between the timing of a SSW-ES event and the amount of EPP-NOx crossing each level is slightly stronger than the relationship between the timing of all events and the amount of EPP-NOx crossing each level. SSW-ES events are more likely to occur earlier in the season. This figure also shows that to get significant EPP-NOx across the 1.24 hPa level requires an early (December or early January) SSW-ES event. EPP-NOx descending past this level increases the likelihood that O3 will be destroyed through the NOx catalytic cycle, which becomes important below ∼45 km.
4.4 Major SSW and ES Event Timing and Dynamics
 Figure 7 shows the zonal mean pole to equator temperature difference (ΔT) as a function of pressure and time for years with (Figure 7a) no event, (Figure 7b) a December, (Figure 7c) a January, and (Figure 7d) a February event. Here we define ΔT as the difference between T averaged poleward of 70°N and T averaged between 0° and 30°N: T(φ>70°N)–T(0°N<φ<30°N), where φ is latitude. There is a large decrease in ΔT following the event around 1 hPa, which corresponds to a larger equator-to-pole temperature gradient. The largest response is in years with a December event, followed by years with a January event.
 Figure 8 shows the zonal mean zonal wind (U), the momentum tendency due to gravity waves (dU/dtGW), (i-l), and T as a function of pressure and time for the different cases shown in Figures 2, 3, and 7. The middle atmosphere response to the event depends on the month of occurrence of the event. Following the event, in response to the change in ΔT, the winds between ∼1.0 and 0.1 hPa (upper stratosphere–lower mesosphere) become westerly again and are stronger than before the event. The strongest winds occur in the December composite (b), followed by January and February; i.e., the earlier the event occurs, the more westerly the winds are afterward. The relationship between timing of the event and the strength of the response is mirrored in the other variables: following the major event, dU/dtGW is more negative (easterly), is more negative (descent), and the stratopause is elevated. It is interesting to note that in Figures 7 and 8 the fields are different from the years with no event before the event happens. However, there are not enough years to determine whether the difference is statistically significant.
 In the stratosphere, planetary waves drive the stratosphere away from radiative equilibrium. Following a major SSW event, the stratosphere above the zero-wind line (critical layer) is relaxed to radiative equilibrium in the absence of planetary waves [e.g., Hauchecorne et al., 2007; Liu et al., 2009; Manney et al., 2005; Siskind et al., 2010]. What Figure 8 shows is that the closer to solstice the event happens, the faster the upper stratospheric winds are (largest ΔT from equator to pole) during the recovery. This in turn affects GW filtering and in such a way that the earlier the event, the stronger is. Additionally, since the perturbations due to the event are stronger, the earlier the event occurs, earlier major SSW events are more likely to produce an ES event.
 Figure 9 shows the maximum U at ∼0.6 hPa (Figure 9, left), the maximum westward (most negative) dU/dtGW at ∼0.03 hPa (Figure 9, middle), and the maximum descent (most negative ) at ∼0.08 hPa (Figure 9, right) following the event as a function of the day of the event (as defined in section 4.2). The earlier in the season the event, the larger the maximum westerly U, the larger the westward dU/dtGW and, ultimately, the stronger the descent following the event. Therefore, the EPP IE is larger for earlier central warming dates for two main reasons: (1) the earlier in the winter season, the larger (more negative) is following the event, and (2) the earlier an event occurs in the season, the longer the EPP-NOx descends before the final warming and the less chance that EPP-NOx is photochemically destroyed during transport.
5 Summary and Conclusions
 We have used WACCM to study the effects of major SSW and ES events on the EPP IE in the NH polar winter. The EPP IE refers to NOx that is produced by EPP in the mesosphere and/or thermosphere and is transported to the stratosphere during polar winter. WACCM is a useful tool with which to study the EPP IE because it is possible to keep the level of EPP constant in order to isolate the effects of major SSW and ES events on the EPP IE.
 We quantified the EPP IE as the amount of EPP-NOx descending across three WACCM levels in the upper stratosphere. The results here suggest that the seasonal timing of an event is an extremely important factor controlling the EPP IE: the earlier in the season the event occurs, the more EPP-NOx is brought down and the lower in altitude the EPP IE extends. We found that the winters with the largest EPP IE are the ones with a December major SSW followed by an ES event. Similar relationships were shown between the timing of the event and the dynamical response of the middle atmosphere. We concluded that the EPP IE is larger for earlier central warming dates for two main reasons: (1) the earlier in the winter season, the larger (more negative) is following the major event, and (2) the earlier an event occurs in the season, the longer the EPP-NOx descends before the final warming and the less chance that EPP-NOx is photochemically destroyed during transport.
 On average, the EPP IE in winters with an event is larger than the EPP IE in winters with no event; however, in dynamically calm winters, the EPP IE can exceed the EPP IE in years with a January, February, or March event. This is because the steady descent of EPP-NOx is not interrupted in dynamically calm winters, whereas in winters with a major event later in the season, the normal descent is interrupted and the increase in descent after the event is too small and too late in the season to make up for it. We also found that the EPP IE at the 1.24 hPa level (∼43 km) for winters with a SSW-ES event is ∼26 times higher than it is for winters with no event and ∼58 times higher when the SSW-ES happens in December. This is relevant because the NOx catalytic cycle becomes important below ∼45 km in the presence of sunlight.
 Overall, our results fit well with the picture of the EPP IE that we have from observations. For instance, the NH 2003–2004 EPP IE was by far the largest EPP-NOx enhancement ever observed, and a SSW-ES event took place on 2 January 2004 [Manney et al., 2005]; however, the level of EPP was relatively high before and during the SSW-ES. The Ap index was ∼10.8 for the second half of December 2003 and ∼19.7 for the first half of January 2004. The SSW-ES events in the 2005–2006 and 2008–2009 NH winters had central warming dates on 21 and 24 January, respectively [Manney et al., 2009a]. The EPP IE in 2005–2006 and 2008–2009 was not nearly as large as in 2003–2004, although the comparison is complicated by the fact that the level of EPP was also lower (the monthly average Ap index was ∼8.5 in January 2006 and ∼4.3 in January 2009). Additionally, the descending tongue of EPP-NOx reached lowest in altitude in 2004, followed by 2006 [Randall et al., 2009], which is consistent with what WACCM shows: the earlier the warming, the lower in altitude the EPP-NOx extends.
 A major discrepancy between the WACCM runs used here and observations is that WACCM underestimates the magnitude of the EPP IE compared to observations. Holt et al.  estimated the EPP IE in the 2003–2004 NH winter to be 2.3–2.7 Gmol across 2000 K, which is ∼20 times larger than the largest EPP IE at 0.73 hPa (∼2000 K) in the WACCM runs here. The average Ap index for December 2003 to January 2004 was ∼16.6, so the discrepancy cannot be attributed to the level of auroral EPP. This corroborates the results of Smith et al. , who found that NOx mixing ratios in the winter with the largest EPP IE out of four 53-year free-running WACCM simulations were still lower than observed NOx mixing ratios in the 2008–2009 NH winter. They suggest possible mechanisms for the discrepancy: (1) the poleward branch of the residual circulation in WACCM is too low in altitude, and/or (2) the eddy or molecular diffusion is not strong enough in WACCM. The latter is a topic of current research and has possible implications for the results here. It is possible that a change in diffusion would affect the different cases disproportionately. Additionally, some of the discrepancy between WACCM and observations might be explained by the lack of higher energy particle populations (e.g., solar protons, relativistic electrons, and medium energy electrons) in WACCM.
 This work was supported by the NSF CEDAR program under grant AGS 0940124 and the NASA LWS program under grant NNX10AQ54G. NCAR is sponsored by the National Science Foundation.