Precipitating solar protons contribute to ozone depletion in the atmosphere; α particles and electrons also precipitate during solar energetic particle (SEP) events. If the SEP is accompanied by a shock, then magnetospheric particles can also be injected into the atmosphere as the shock hits the magnetosphere. Both particle species in both particle populations show distinct energy spectra (and thus penetration depth in the atmosphere) and precipitate in different regions: the SEP inside the polar cap and the magnetospheric particles inside the auroral oval. In this paper, we reevaluate the 3-D spatial and temporal precipitation patterns of these particle populations for the October–November 2003 event and compare the results to conventional approaches using only protons in evaluating SEP consequences. The main results are as follows: (1) The 3-D model AIMOS gives a very differentiated picture of the global ionization maps; (2) if only protons are considered, the differences between the 3-D model and the conventional approach of homogeneous precipitation inside the polar cap are small in NOx production and ozone depletion in the mesosphere and stratosphere; and (3) the consideration of electrons in addition to protons leads to significant increases in atmospheric ionization in the mesosphere, less so in the stratosphere. This is reflected in changes in the chemical composition as shown here for ozone depletion and an increase of NOx.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Large solar energetic particle events cause ozone destruction [Crutzen et al., 1975; Heath et al., 1977; Jackman et al., 2000; Randall et al., 2005]. Conventional modeling of the ozone loss considers solar protons only; for instance the electrons are assumed to contribute less than 10% to the total ionization [Jackman and McPeters, 1985]. An analysis of the total particle inventory in the October–November 2003 period suggests that depending on the parent flare the relative contributions to ion pair production of protons and electrons can be quite variable [Wissing and Kallenrode, 2009]. In the study of solar events, often only solar protons are considered, neglecting the contribution of magnetospheric particles. This approach has been justified in some way by the reasonable agreement between model results and measurements, as, e.g., shown for the July 2000 solar event by Jackman et al. , or for the October–November 2003 solar event by Jackman et al. [2005a, 2008] and Rohen et al. . However, there is mounting evidence for a significant influence of magnetospheric electrons on the composition of the middle atmosphere, both from observations [e.g., Thorne, 1977; Randall et al., 2005, 2007; Sinnhuber et al., 2006; Seppälä et al., 2007] and from model studies [e.g., Callis et al., 1996, 2001; Codrescu et al., 1997; Rozanov et al., 2005]. As a substantial amount of magnetospheric particles can be injected into the atmosphere if a shock or coronal mass ejection arrives at Earth, both solar and magnetospheric particles might contribute to the observed changes during solar particle events. Both particle populations show distinct spatial and temporal precipitation patterns: solar energetic particles precipitate along open field lines into the polar cap while magnetospheric particles are injected into the polar oval. With increasing geomagnetic activity the polar cap expands and particles are injected at lower latitudes [Leske et al., 1997, 2001].
Wissing and Kallenrode  suggested a combined approach in the Atmospheric Ionization Module Osnabrück (AIMOS). AIMOS uses two polar orbiting satellites complemented by a geostationary one to model atmospheric ionization by protons and electrons of solar and magnetospheric origin. This combination allows the 2-D modeling of the horizontal particle precipitation pattern depending on geomagnetic activity. Thus both particle populations and the spatial variation of their precipitation are modeled. As already shown by Wissing and Kallenrode  for the October–November 2003 period, in the solar particle population electron precipitation can exceed 30% at certain heights during certain times of the event. The consideration of atmospheric ionization by the formerly neglected magnetospheric particles will increase total ionization even more. Consequently, also subsequent models for atmospheric chemistry and circulation will yield different results. In this paper we present a case study for the October–November 2003 event to show whether and how these additional particles and ionization show up in the modeling of ozone chemistry.
 The paper is structured as follows. In section 2 we describe the data and models. Section 3 shows the results: relative contributions of different particle populations to ionization, the consequences of the additional ionization in a conventional chemistry model (Bremen three-dimensional chemistry and transport model) extending well into the mesosphere. The results are summarized and implications for atmospheric modeling are discussed in section 4.
2. Data and Models
 The modeling chain consists of two parts: (1) the ionization model AIMOS, which calculates atmospheric ion pair production rates from particle fluxes observed in space and inside the magnetosphere, and (2) chemistry/circulation models which process these ionization rates to calculate electron densities and ozone depletion, which in turn can be compared to observations.
 AIMOS [Wissing and Kallenrode, 2009] is a 3-D numerical model of atmospheric ionization due to precipitating particles with high spatial resolution for an atmosphere extending from ground up to 1.7 × 10−5 Pa, corresponding to an upper boundary between 250 to 600 km. The spatial grid is borrowed from HAMMONIA [Schmidt et al., 2006] with 3.6° × 3.6° in the horizontal and 67 logarithmically equidistant height layers. Ionization rates are calculated from the observed particle spectra by a Monte Carlo approach based on the GEANT-4 simulation package [Agostinelli et al., 2003].
 Particle data to calculate the horizontal precipitation patterns and the low energetic part of the particle spectra for both particle populations are taken from the Polar Orbiting Environmental Satellites (POES) NOAA 15 and NOAA 16; both flying in a Sun-synchronous orbit with a height of 850 km and an inclination of 98°. Equatorial crossing in the southward direction nominally occurs at 0730 UT for NOAA 15 and at 0200 for NOAA 16. Particle measurements are performed with the Space Environment Monitor SEM-2 [Evans and Greer, 2005] which consists of the Total Energy Detector (TED) measuring low energetic particles and the Medium Energy Proton and Electron Detector (MEPED). The combined instruments cover electrons from 150 keV to 2.5 MeV and protons from 150 keV to 6.9 MeV. Protons with higher energies (4–500 MeV) are taken from the Energetic Particle Sensor (EPS) [National Aeronautics and Space Administration, 1996] on GOES 10 or 11. For the higher electron energies, AIMOS extends the spectrum to 5 MeV because SOHO/COSTEP [Klassen et al., 2005] and SAMPEX [Mewaldt et al., 2005] observations suggest that the higher electron energies track the lower ones quite well.
 AIMOS considers ion pair production rates up to heights of some hundred kilometers; since this study focuses on ozone depletion due to precipitating particles, AIMOS is combined with an atmospheric model with a strong focus on chemistry, the Bremen three-dimensional chemistry and transport model.
 The Bremen three-dimensional chemistry and transport model is a combination of the Bremen transport model developed by B.-M. Sinnhuber [B.-M. Sinnhuber et al., 2003] with the chemistry code of the Bremen two-dimensional model of the stratosphere and mesosphere [M. Sinnhuber et al., 2003; Winkler et al., 2008]. The Bremen three-dimensional chemistry and transport model is driven by analyzed wind fields and temperatures from the European Centre for Medium-Range Weather Forecasts (ECMWF). It runs on 28 isentropic surfaces from 330 to 3402 Kelvin (about 10 to 65 km) with a horizontal resolution of 3.75° × 2.5°, and a vertical resolution of about 1 km in the lower stratosphere, increasing to about 4 km at 60 km altitude. The vertical motion perpendicular to the isentropes is described by diabatic heating and cooling. Diabatic heating and cooling rates are calculated using the MIDRAD radiation scheme [Shine, 1987]. Advection is calculated by using the second-order moments scheme of Prather . The neutral model chemistry includes about 180 gas phase, photochemical, and heterogeneous reactions and 57 tracers and uses the recent set of recommendations for kinetic and photochemical data of the Jet Propulsion Laboratory [Sander et al., 2006]. NOx and HOx production due to atmospheric ionization are parameterized in such a way that 1.25 NOx (55% NO, 45% N) [Porter et al., 1976] and up to 2 HOx constituents depending on altitude and ionization rate [Solomon et al., 1981] are produced per ion pair, as described, e.g., by Jackman et al. [2005a] and Rohen et al. .
 Some consequences of the 3-D model already have been discussed by Wissing and Kallenrode , namely, (1) during quiet times, the major contribution to ionospheric ionization is from electrons in both the polar cap (solar electrons) as well as in the auroral oval (magnetospheric electrons) with the ionization in the auroral oval exceeding that in the polar cap; (2) during solar particle events the dominant effect in the polar cap in the stratosphere and mesosphere is from solar protons although solar electrons can contribute up to 30% to the ionization; (3) during strong shocks following a solar particle event, in the auroral oval magnetospheric electrons and protons lead to ionization rates of up to some ten % of the ones of solar particles; and (4) independent of particle source and precipitation site, in general ionization by electrons is more important in the thermosphere.
Figure 2 (top) shows the spatial distribution of the total electron production (TEP) rate, that is, the vertically integrated ion pair production rate, for electrons (Figure 2, left) and protons (Figure 2, right) during 23 October (doy 296). As the Kp index is approximately 1.4, which is very low, particle flux is at background level typical of very quiet times. At (almost) any given point the contribution of electrons to TEP exceeds that of protons with maximum TEP rates around the geomagnetic poles. The sharp separation between a polar cap and an auroral oval in individual energy channels as described by Wissing et al.  is not visible. TEP is calculated from the entire energy spectrum of each species, and the size and location of polar cap and auroral oval depend on particle energy and species. Thus TEP tends to smear out these features. Nonetheless, the maximum TEP gives some indication of the location of the auroral oval. At slightly higher Kp the sharp separation of oval and cap will be observable.
 Please note that the white color within the TEP ionization graphs represents areas where the particle precipitation (within the examined energy range) is assumed to be negligible at any time. Therefore these areas have been excluded in the model. Precipitation in the South Atlantic Anomaly is (at the moment) not covered by the model either. The ionization within the colored area is not calculated continuously but in seven zones for each hemisphere and each local time sector. These zones have been arranged by similar precipitation properties. Hence terrace structure and sudden edges result from binning [Wissing and Kallenrode, 2009].
 In Figure 2 (bottom) the same data are shown for 24 October (doy 297). As shown in Figure 1, a shock hits the Earth increasing the TEP rates. Proton induced ionization rises by a factor of ten.
 The electron domination of TEP in Figure 2 should be interpreted with care: as discussed by Wissing and Kallenrode , at all times ionization by precipitating electrons is dominant in the thermosphere while this is not necessarily the case in the stratosphere and mesosphere. However, as indicated in their Figure 8, during quiet times ionization by electrons is dominant also in the mesosphere. A conclusion for the stratosphere cannot be drawn from the model because of the limited energy range of particle observations: electrons with the highest observed energies precipitate only as far as 50 km, thus the model does not give ionization rates for precipitating electrons with higher energies for which no observations are available.
 During time period “B” (doy 312 to doy 315) in Figure 3 fluxes of solar energetic particles are low but geomagnetic activity increases during day 313 and stays at a rather high level. Figure 3 shows the TEP maps for electrons (Figure 3, left) and protons (Figure 3, right). During all four days, the auroral oval is clearly visible in both particle species and precipitating electrons dominate TEP rates. With increasing geomagnetic activity (from the first to the second to the third pair of panels), TEP rates increase and the auroral oval expands. This pattern reflects the typical features expected for the precipitation of magnetospheric particles.
 Let us now turn to the two particle events marked by the bar “A” (doy 300 to doy 304) in Figure 1. Figure 4 shows daily averaged TEP rates for electrons (Figure 4, left) and protons (Figure 4, right) for this time period. TEP rates on doy 300 are dominated by the decay phase of a solar particle event (flare on doy 299) and contributions from a shock originating in an even earlier event. Figure 5 (top) displays an ion pair production of >101 s−1 cm−3 peaking at 70 km and indicating precipitation of high energetic protons due to the former event. However, electron ionization dominates upper altitudes; therefore TEP rates are dominated by electrons as well and TEP maximum indicates the energy-averaged location of the auroral oval.
 Doy 301 (Figure 4, second pair of panels) is dominated by a large solar energetic particle event originating in a X17 flare starting at 1215. During this time period, dominant TEP contribution is from protons in the polar cap. The contribution of precipitating electrons to TEP shows the same spatial pattern as on the previous day, indicating a strong magnetospheric contribution.
 The shock accompanying the flare of doy 301 arrives in the morning of day 302 (Figure 4, third pair of panels, and Figure 5, middle). The contribution of precipitating electrons to the TEP map shows (1) a general increase in ion pair production rates, (2) an expansion of the auroral oval toward the equator (expansion is most obvious looking at northern Russia or Canada) and (3) strong impact on low altitudes. These effects are expected during strong geomagnetic storms. The contribution of precipitating protons to TEP shows a slightly different pattern: compared to the previous day, the ion pair production rate is increased due to the higher particle fluxes and the maximum also expands equatorward as indicated by the 1014 contour line. However, as ionization in polar cap and aurora cannot be distinguished on doy 301 to doy 303 the strong solar particle ionization covers the auroral expansion. While at doy 304 geomagnetic disturbance still is at a high level (Kp = 6.3 compared to 3.8 at doy 301, 7.0 at doy 302 and 7.3 at doy 303) a decreasing solar proton flux exposes the expansion.
 The strong particle precipitation at doy 302, however, occurs in the polar cap while the auroral oval does not silhouette against the cap. The explanation for the apparently contradictory behavior is obvious from Figure 1: later on day 302 a new flare leads to a fresh increase in solar energetic particles which then precipitate deep into an already expanded polar cap (see also Figure 5). The maximum ionization moves to lower altitudes and is mostly generated by protons. The entirely different behavior of electrons and protons nicely demonstrates the complexity in the relative contributions of both particle species to TEP.
 On doy 303 (Figure 4, fourth pair of panels) the situation is much simpler: the fluxes of solar energetic particles decrease but the shock accompanying the flare from the previous day arrives. Consequently, the spatial pattern in both particle species shows a pronounced auroral oval, indicating strong precipitation of magnetospheric particles as expected during a geomagnetic disturbance. The polar cap and auroral oval still are expanded equatorward. On the following day, the pattern continues, also particle fluxes and thus TEP rates are lower.
 At the end of the October event the lack of high energetic particles causes an uplift of the zone of maximum ionization. Figure 5 (bottom) displays our results for doy 313 showing a maximum ionization at 120 km for protons.
4. Atmospheric Consequences
Section 3 demonstrated that AIMOS gives a differentiated picture of atmospheric ionization due to precipitating particles. In this section we will demonstrate that this also has consequences for the results of atmospheric modeling. We have incorporated ionization rates produced with AIMOS into the Bremen 3d CTM in three different scenarios: scenario A, the conventional scenario using only averaged proton rates; scenario B, a scenario using only protons but with the 3-D distribution of ionization rates provided by AIMOS; and scenario C, a scenario considering the full AIMOS solution, that is, electrons and protons of both solar and magnetospheric origin. Additionally, a “base” model scenario was carried out for the same time range without atmospheric ionization impacts. This is used as a reference for the particle impacts of model scenarios A, B, and C. In model scenario A, atmospheric ionization is allowed only into regions of geomagnetic latitudes poleward of 60°; in scenarios B and C, the spatial distribution of ionization is determined by the AIMOS model results, i.e., based on measured particle fluxes. Deviations from the “base” model scenario will be given as follows: NOx: scenario - base; ozone: 100* (scenario-base)/base.
 Atmospheric ionization leads primarily to the formation of NOx and HOx. Both NOx and HOx destroy ozone in catalytic cycles, albeit in different altitudes: NOx dominates ozone loss in the middle stratosphere, HOx in the upper stratosphere and mesosphere [Lary, 1997]. Thus, immediate ozone loss in the mesosphere during a particle precipitation event will be dominated by HOx, which is short-lived and relaxes to background values immediately with ionization rates. NOx, however, can be quite long-lived especially during high-latitude winter, when it can be transported down into the middle stratosphere; thus, long-term impacts of the particle events on stratospheric ozone are due mainly to the production of NOx [e.g., Jackman et al., 2000, 2005b].
Figure 6 shows the ozone depletion following particle precipitation for the time interval from Figure 2 (doy 296 and 297, correlating to 23 and 24 October, respectively) for the three scenarios described above. As expected due to lack of high energetic protons (see Figure 1 in this paper or Figure 8 of Wissing and Kallenrode ), ozone depletion due to atmospheric ionization is negligible during this time interval for scenarios considering solar protons only (left for averaged precipitation, middle for precipitation as calculated with AIMOS). If electrons and magnetospheric particles are considered (Figure 6, right), variations in ozone are observed on both days: Small but significant ozone changes of 2–5% are observed compared to the “base” model run even during geomagnetic quiet times (doy 296) at high northern latitudes. On day 297, which shows increased geomagnetic activity, ozone depletion exceeds 15% in high latitudes. These results suggest that, given the AIMOS ionization rates, even during geomagnetic quiet times electron precipitation has an impact on the ozone budget of the polar mesosphere. This impact is probably restricted to high-latitude winter where ozone recovery is slow. However, these results suggest that at least during polar night, models predicting middle atmosphere ozone need to take electron precipitation into account even during geomagnetically quiet times. This might also explain the large year-to-year variation of middle stratospheric ozone observed during high-latitude polar winter [Sinnhuber et al., 2006].
 As ozone loss in this altitude is mainly due to HOx, whose photochemical lifetime is in the order of minutes in the lower mesosphere and stratosphere, the spatial distribution is mainly due to the precipitation patterns. This is certainly true in the Southern Hemisphere, where recovery of ozone is very fast. At very high northern latitudes it is already dark enough that ozone recovery, which depends on photolysis of O2, does not take place, and here the spatial distribution might also be affected by transport of ozone-poor air in regions without solar illumination.
Figure 7 shows the longitudinally averaged relative ozone depletion as a function of altitude for the same time period and scenarios as in Figure 6. No significant ozone depletion is observed for scenarios A and B (protons only) at any altitude. In scenario C, ozone losses are restricted to altitudes above 45 km and latitudes poleward of 45°N and 30°S, where they exceed 15% on doy 297. That no significant ozone losses are observed below 45 km in scenario C is not surprising because the data set underlying AIMOS only considers electrons up to energies corresponding to a stopping height of about 50 km.
Figure 8 shows the same three model scenarios, A, B, and C, at an altitude of 56 km for the time period also used in Figure 3 (doy 312–315, corresponding to 8 to 11 November) which was geomagnetically active with increased ionization along the auroral oval. Some residual changes in O3 can be observed in all three scenarios during this time series, especially at northern high latitudes. At this altitude, these are probably due to an incomplete recovery of ozone after the large solar events on 29–30 October and 3–4 November during polar night rather than to a continuing catalytic ozone loss. The overall ozone depletion is largest in scenario C. However, this is probably also a remnant of the solar events and not due to the additional precipitating magnetospheric particles during days 312–315, because their particle spectrum is rather steep and consequently most of the TEP is in the thermosphere rather than in the mesosphere.
Figure 9 shows the same three model scenarios for the particle events and shocks discussed in Figure 4 (doy 300–304, corresponding to 27–31 October). The amount of ozone loss in high latitudes is quite similar in all three scenarios. Model runs A and B also have a very similar spatial pattern, with only small deviations at the edge of the southern polar cap. This shows that the assumption of a homogeneous polar cap with edges defined by 60° geomagnetic latitude is reasonable if only solar particles are considered. In scenario C, ozone losses during the event are quite comparable to scenarios A and B at high latitudes. The spatial distribution of scenario C is different from scenarios A and B insofar as small but significant ozone loss values extend far into mid latitudes compared to scenarios A and B. This is most pronounced on 29 and 30 October, at the east coast of North America, and extending from south Australia to the tip of South America in several regions.
 Let us now turn to longer-term consequences of particle precipitation. Outside of polar night, the lifetime of NOx in the upper stratosphere and lower mesosphere is in the order of days to weeks, the lifetime of ozone in the order of minutes (see Figure 10); within polar night, the lifetime of ozone increases to several years. As NOx does not contribute to catalytic ozone loss above 45 km, ozone is not a good indicator of medium-term atmospheric ionization effects outside polar night. Therefore, in the following we will discuss the medium-term effects of particle precipitation by considering NOx outside of polar night, and medium to long-term effects by considering NOx and ozone within polar night. We will limit ourselves to scenarios B and C, that is, the correct spatial pattern for protons only (B) and for protons and electrons (C).
Figure 11 shows a global picture of NOx volume mixing ratios at 56 km for selected time intervals during the event relative to the “base” scenario. The consideration of all particles (Figure 11, right) leads to much higher NOx values at high latitudes. In addition, the latitudinal extension of the region affected by particle precipitation, which has already been discussed for ozone, is occasionally observed in the NOx production. This can be seen quite clearly, for example, on 29 October, where enhanced NOx values reach the tip of South America in scenario C, but not in scenario B, or on 4 November, where NOx enhancements reach down into the gulf of Persia in scenario C, but not in scenario B.
 However, the spatial distribution of NOx enhancements into lower latitudes is different to that of ozone loss shown in Figure 9. Ozone depletion in this altitude range is driven by HOx, which is very short-lived. Regarding latitudes southward of 80°N, ozone recovers quickly. Therefore the distribution of ozone depletion directly reflects areas of particle precipitation. However, NOx is considerably longer lived than ozone everywhere but in polar night (see Figure 10), and the spatial distribution of NOx enhancements also reflects horizontal transport of enhanced NOx values into lower latitudes.
 In Figure 12, an example of the evolution of NOx and ozone over a longer time series (from 20 October to 31 December) is shown exemplary for the Arctic station of Ny Ålesund (78°55′N, 11°57′E), relative to a model run without atmospheric ionization. A location in high northern latitudes was chosen because there both NOx and ozone are comparatively long-lived, the lifetime of ozone increasing from several hours in early November (see Figure 10) to several years during polar night, and downward transport of NOx as well as ozone-poor air into the midstratosphere is possible during polar winter. As already seen in Figures 6 and 7, some ozone loss and NOx production occurs already before the solar events on 29 October in model scenario C due to enhanced geomagnetic activity from 24 October. However, ozone loss and NOx production during the solar event are actually quite similar in model scenarios B and C. After the event, the lower edge of the significantly affected area descends down from around 40 km altitude to around 28 km; this is observed in both model scenarios, and both for NOx enhancements and ozone depletion. Both NOx enhancements and ozone depletion are larger in model scenario C. Ozone loss in model scenario C exceeds this in model scenario B at the end of December by 5–10% at 30 km altitude, and by more than 10% between 30 and 40 km altitude. However, it is not clear from comparing model scenarios B and C whether this is due to additional ionization at the edge of the polar cap region during the solar events, or due to the additional impact of magnetospheric particle precipitation during the geomagnetically disturbed times before and after the solar event.
 In this paper we present different model runs for the October–November 2003 event using the Atmospheric Ionization Module Osnabrück in combination with the Bremen three-dimensional chemistry and transport model. First, we examine the relative contributions of electrons and protons to the total electron production (TEP) rates. The results confirm that AIMOS produces the typical features expected from solar and magnetospheric particles and partly already discussed by Wissing and Kallenrode . Second, we use the Bremen three-dimensional chemistry and transport model to evaluate whether the use of AIMOS ionization rates influences modeling of NOx production and ozone depletion compared to the simpler assumption of solar energetic protons only. The main results are as follows:
 1. If only proton precipitation is considered, total ozone depletions are almost identical for the assumption of homogeneous precipitation inside the polar cap and the spatial precipitation pattern determined with AIMOS. A minor difference between the spatial pattern of proton induced Ozone depletion is visible in the Southern Hemisphere on 29 October, the main phase of the event. This might reflect the large aberration of the geomagnetic south pole (79.5°S 108.4°E as given by IGRF-10 model for 2000, and commonly used as central point for the polar cap ionization) compared with the magnetic south pole (64.7°S 138.3°E as measured by Barton  for the same year). AIMOS uses satellite measurements only to describe the spatial pattern. Therefore estimations on the central point are not necessary.
 2. Our model suggests that total ionization column by electrons almost always exceeds the one of the protons. Exceptions are the polar cap regions within the maximum of the October event. As most of the electron impact concentrates on altitudes above 80 km an increased altitude range may be beneficial for GCM models.
 3. During solar particle events, the direct impact of the electrons is small and due to model constraints restricted to altitudes above 50 km. However, considering the electrons as well extends the area affected by particle precipitation somewhat to lower latitudes. This results in a somewhat fuzzy edge of the polar cap, as, e.g., observed by SCIAMACHY during the October–November 2003 SPE [Rohen et al., 2005].
 4. Model results suggest that at least at high latitudes during polar winter, mesospheric ozone can be affected significantly by geomagnetic electrons. A small impact of electron precipitation is predicted using the AIMOS ionization rates even during geomagnetic quiet times, increasing to quite significant values of ozone loss during geomagnetically disturbed times even in the absence of solar protons. This might go some way toward explaining the observed variability of ozone in the upper stratosphere and mesosphere at high latitudes during polar winter [Sinnhuber et al., 2006; M. Palm et al., Stratospheric and mesospheric O3 above Spitsbergen modeled by a 3-D chemical transport model and measured by ground-based millimeter wave radiometry, submitted to Journal of Geophysical Research, 2009].
 5. It has been shown that considering additional magnetospheric electrons also has a significant (10–15%) impact on stratospheric ozone during polar winter. However, it is not clear whether this is due to additional magnetospheric electron precipitation into the edge of the polar cap during the solar events, or due to magnetospheric electron precipitation during days of enhanced geomagnetic activity before and after the solar events.
 6. Our calculations imply that electron ionization strongly affects NOx concentration and should not be neglected (see Figure 11). The computations include electrons up to energies of 2.5 MeV (extrapolated to 5 MeV), which directly produce NOx above about 45 km. Lower altitudes, which are not directly affected by electron precipitation, also show increased NOx concentrations (see Figure 12).
 These results suggest that electron precipitation should be considered in modeling ozone in the stratosphere and mesosphere, not only during large solar events and geomagnetically highly disturbed times, but also during geomagnetically quiet times, at least during polar winter.
 This work was supported by the Deutsche Forschungsgemeinschaft DFG under contracts DFG-Ka1297/7-1 and DFG-Ka1297/8-1.