Carrington event solar protons

Wednesday, September 17, 2008
1
The atmospheric impact of the Carrington event solar protons 1
Craig J. Rodger 2
Department of Physics, University of Otago, Dunedin, New Zealand 3
Pekka T. Verronen 4
Finnish Meteorological Institute, Helsinki, Finland 5
Mark A. Clilverd and Annika Seppälä 6
Physical Sciences Division, British Antarctic Survey (NERC), Cambridge, United Kingdom 7
Esa Turunen 8
Sodankylä Geophysical Observatory, University of Oulu, Sodankylä, Finland 9
Abstract. The Carrington event of August/September 1859 was the most significant solar 10
proton event (SPE) of the last 450 years, about four times larger than the solar proton fluence of 11
the largest event from the "spacecraft era" (August 1972). Recently, much attention has focused 12
upon increasing our understanding of the Carrington event, in order to better quantify the 13
impact of extreme space weather events. In this study the Sodankylä Ion and Neutral Chemistry 14
(SIC) model is used to estimate the impact of the Carrington event to the neutral atmosphere 15
and the ionosphere, and the disruption to HF communication. We adopt a reported intensity- 16
time profile for the solar proton flux, and examine the relative atmospheric response to 17
different SPE-energy spectra, and in particular, the comparatively soft energy spectrum of the 18
August 1972 or March 1991 SPE which are believed to provide the best representation of the 19
Carrington event. Our calculations indicate that large changes in electron density and 20
atmospheric constituents occur during the period of SPE-forcing, depending upon the nature of 21
the spectrum and also on the hemisphere considered. However, the most important SPE-driven 22
Wednesday, September 17, 2008
2
atmospheric response is an unusually strong and long-lived Ox decrease in the upper 23
stratosphere (Ox levels drop by ~40%) primarily caused by the very large fluxes of >30 MeV 24
protons. This depletion is an indication of the extreme changes possible for the largest SPE. We 25
find that there are comparatively small long-term differences in the atmospheric and 26
ionospheric response between the 3 suggested SPE spectra. 27
28
1. Introduction 29
The Carrington event of August/September 1859 was the most significant solar proton event 30
(SPE) of the last 450 years, identified through impulsive nitrate events in polar ice [McCracken 31
et al., 2001]. The >30 MeV solar proton fluence determined from the ice cores indicate it was 32
twice as large as the next largest event (1895), and roughly four times larger than the solar 33
proton fluence of the largest event from the "spacecraft era" which occurred in August 1972. 34
The Carrington SPE was associated with the 1–2 September 1859 magnetic storm, the most 35
intense in recorded history [Tsurutani et al., 2003]. The space weather events of 36
August/September 1859 are now particularly famous due to Carrington's visual observation of 37
a white-light solar flare for the first time [Carrington, 1960]. The associated magnetic 38
disturbances produced widespread auroral displays and disruption to telegraph transmissions 39
which attracted much public attention and were widely reported in the newspapers and 40
scientific articles [see the review Boteler, 2006]. 41
Recently, much attention has focused upon increasing our understanding of the Carrington 42
event, in order to better quantify what extreme space weather events could do to our current 43
technological society. For example, estimates suggest a potential economic loss of billion due to lost revenue (~US$44 billion) and the cost of replacement of GEO satellites 45
(~US$24 billion) caused by a "once a century" single storm similar to the Carrington event 46
Wednesday, September 17, 2008
3
[Odenwald et al., 2006]. These authors estimate that 80 satellites in low-, medium, and 47
geostationary- Earth orbits might be disabled as a consequence of a superstorm event with 48
additional disruptions caused by the failure of many of the satellite navigation systems (e.g., 49
GPS). Ionising radiation doses from the SPE have been estimated to be as high as 54 krad (Si) 50
[Townsend et al., 2003], levels which are not only highly life-threatening for crews of manned 51
missions, but present a significant hazard to onboard electronics. 52
Solar proton events produce large ionization changes in the polar ionosphere which can drive 53
significant changes in atmospheric chemistry and communications disruption. Over the years 54
several studies of Solar Proton Event effects on the atmosphere have been published. The 55
earlier work of Crutzen and Solomon [1980], McPeters et al. [1981], and Solomon et al. [1983] 56
has been followed by several studies, notably the work of Jackman and coauthors [Jackman 57
and McPeters, 1985; Jackman and Meade, 1988; Jackman et al., 1990, 1993, 1995, 2000]. 58
SPEs result in enhancements of odd nitrogen (NOx) and odd hydrogen (HOx) in the upper 59
stratosphere and mesosphere [Crutzen et al.,1975; Solomon et al., 1981; Jackman et al., 1990, 60
2000]. NOx and HOx play a key role in the ozone balance of the middle atmosphere because 61
they destroy odd oxygen through catalytic reactions [e.g., Brasseur and Solomon, 1986, pp. 62
291-299]. Ionization changes produced by a 20 MeV proton will tend to peak at ~60 km 63
altitude [Turunen et al., Fig 3, 2008]. Ionization increases occurring at similar altitudes, caused 64
by solar proton events are known to lead to significant local perturbations in ozone levels 65
[Verronen et al., 2005], with polar ozone levels decreasing by >50% for large SPE. However, 66
the effect on annually averaged global total ozone is considered to be relatively small, of the 67
order of few tenths of a percent at the maximum [Jackman et al., 1996]. Changes in NOx and 68
O3 consistent with solar proton-driven modifications have been observed [Jackman et al. ,2001; 69
Seppälä et al., 2004; Verronen et al., 2005]. It is well-known that particle precipitation at high 70
latitudes produce additional ionization leading to increased HF absorption at high-latitudes 71
Wednesday, September 17, 2008
4
[MacNamara, 1985], in extreme cases producing a complete blackout of HF communications 72
in the polar regions. 73
In order to consider the impact of the Carrington event solar protons upon the Earth's 74
atmosphere, information on the fluence and energy spectrum of the SPE is required. An 75
estimate of the odd nitrogen increases and ozone decreases due to the Carrington SPE has been 76
undertaken [Thomas et al., 2007], using a Greenland ice core derived >30 MeV fluence of 77
2.7×1010 cm-2 and a spectrum taken from the very energetic and spectrally hard 19 October 78
1989 SPE. The total ionization was distributed over the 2 day duration uniformly (i.e. as a step 79
function), leading to a localized maximum column ozone depletion which was ~3.5 times 80
greater than that of the 1989 event. As noted in this study, the use of the 19 October 1989 81
spectrum to represent the 1859 Carrington SPE was a "best guess" approach, given the total 82
lack of direct proton spectral measurements in that era. 83
Limits on the Carrington event spectrum have been provided by measurements of the 84
cosmogenic isotope 10Be, also found in polar ice cores. Analysis of the 10Be concentrations 85
suggest that the spectral hardness of the Carrington event was significantly softer than those of 86
September-October 1989 SPE [Beer et al., 1990]; any increase in 10Be associated with the 87
Carrington event was found to be less than the 9% standard deviation of the annual data. It has 88
been suggested that the Carrington event may have had an energy spectrum very similar to 89
those measured for the August 1972 or March 1991 SPE [Smart et al., 2006]. The later study 90
also constructed an intensity-time profile of the solar particle flux, by assuming that the 91
Carrington solar event is part of the class of interplanetary shock-dominated events where the 92
maximum particle flux is observed as the shock passes the Earth. 93
In this study we make use of the new findings as to the nature of the Carrington event, 94
adopting both the Smart et al. [2006] intensity-time profile for the >30 MeV solar proton flux, 95
and the softer energy spectrum required to reproduce the cosmogenic isotope concentrations. 96
Wednesday, September 17, 2008
5
We make use of the Sodankylä Ion and Neutral Chemistry (SIC) model to estimate the impact 97
of the Carrington event to the neutral atmosphere and the ionosphere, and the disruption to HF 98
communication. We go on to consider the potential "worst case" significance of a Carrington- 99
level "superstorm". 100
2. Spectrum and intensity-time profile of the Carrington SPE 101
As noted above, cosmogenic isotope concentrations have indicated that the Carrington SPE 102
was significantly softer than those of September-October 1989 SPE. Smart et al. [2006] 103
concluded that either of the comparatively soft SPE energy spectra from August 1972 or March 104
1991 could be representative of that for the Carrington event, given the cosmogenic isotope 105
observations. In contrast, some previous studies into the potential doses to humans and 106
electronics, and the atmospheric impact, have used hard SPE energy spectra, particularly 29 107
September 1989 and 19 October 1989 [see the discussion in Townsend et al., 2006]. In this 108
study we model the SPE spectra using a Weibull distribution, which has been shown to provide 109
an accurate representation of the measured proton spectra for these events [Xapsos et al., Table 110
1, 2000]. The Weibull distribution fit for differential SPE fluxes are described through the 111
expression 112
( ) α α α kE E k
dE
d − Φ =
Φ − exp 1
0 (1) 113
where E is the energy in MeV and Φ
0 , k, and α are the Weibull fitting parameters. k and α are 114
taken from Xapsos et al. [2000], while Φ
0 is scaled to reproduce the >30 MeV proton fluence 115
for the Carrington event (1.9×1010 cm-2 [McCracken et al., 2001]). These values are given in 116
Table 1. The SPE fluxes are expressed with units of protons cm-2s-1sr-1MeV-1. Figure 1 shows a 117
comparison between the differential fluences which have been used to describe the Carrington 118
event, based on four different SPE. The differential fluences shown in Figure 1 have been 119
Wednesday, September 17, 2008
6
normalised to the Carrington-level >30 MeV fluence of 1.9×1010 protons cm-2. As noted by 120
Townsend et al. [2006], both of the SPEs which occurred in 1989 were spectrally hard, and 121
rather similar to one-another. In contrast, the SPEs which occurred in August 1972 and March 122
1991 were much softer, with the August 1972 SPE having an unusually soft spectrum. These 123
two soft-spectra SPEs provide two possibilities through which we can estimate the impact of 124
the Carrington event upon the neutral atmosphere, following the approach of Smart et al. 125
[2006] to treat these as "indicative spectra". The harder spectra from September-October 1989 126
provide an approach by which comparisons can be made with earlier studies, and also an 127
estimate of the possible "extreme" worst-case for a Carrington-level SPE with a hard spectra. 128
As the September and October 1989 SPE have very similar spectra, we arbitrarily select 129
October 1989. Note, however, that the differences between the normalised energy spectra for 130
the two hard SPEs and that of March 1991 are much smaller across the energy range 2- 131
40 MeV. Protons in this energy range deposit most of their energy in the altitude range ~50- 132
85 km [Turunen et al., Fig. 3, 2008], where SPE-induced changes to the neutral atmosphere are 133
largest [e. g., Verronen et al., 2005]. As such, it is instructive to contrast the atmospheric 134
impact of the differing spectra, as this may be less significant than Figure 1 suggests. 135
The upper panel of Figure 2 shows the intensity-time profile for >30 MeV solar proton fluxes 136
during the Carrington event [after Fig. 12, Smart et al., 2006]. This profile is combined with the 137
SPE energy spectra of Figure 1 to produce three different time-varying differential proton 138
fluxes for proton energies of 1-2000 MeV across the time period of the Carrington event, as 139
shown in the 3 lower panels of Figure 2. In those panels the time-varying differential proton 140
flux is shown with units of log10[protons cm-2s-1sr-1MeV-1]. The second and third panels of 141
Figure 2 represent differing possible differential fluxes for the Carrington SPE, while the lower 142
panel represents the "worst case" of a Carrington-level SPE with a hard spectrum. For time 143
periods outside the Smart et al. intensity-time profile, the proton fluxes are set to zero. A 144
Wednesday, September 17, 2008
7
widely accepted SPE definition requires the >10 MeV proton flux to be >10 cm-2s-1sr-1MeV-1. 145
The beginning and end of the time profile shown in Figure 2 are below this level, and thus we 146
have confidence that the fluxes shown in Figure 2 describe the entire SPE event. 147
3. Sodankylä Ion Chemistry Model 148
Using the Sodankylä Ion and Neutral Chemistry (SIC) model we consider the atmospheric 149
consequences of the Carrington SPE using the time-varying proton fluxes from Figure 2. SPE 150
produce ionization increases in the polar mesosphere and upper stratosphere, which in turn alters 151
atmospheric chemistry through changes in HOx and NOx. The SIC model is a 1-D chemical 152
model designed for ionospheric D-region studies, solving the concentrations of 65 ions at 153
altitudes across 20–150 km, of which 36 are positive and 29 negative, as well as 15 minor neutral 154
species. Our study made use of SIC version 6.9.0. The model has recently been discussed by 155
Verronen et al. [2005], building on original work by Turunen et al. [1996] with neutral species 156
modifications described by Verronen et al. [2002]. A detailed overview of the model was given 157
in Verronen et al. [2005], but we summarize the key characteristics of the model here to provide 158
background for this study. 159
In the SIC model several hundred reactions are implemented, plus additional external forcing 160
due to solar radiation (1–422.5 nm), electron and proton precipitation, and galactic cosmic 161
radiation. Solar flux is calculated with the SOLAR2000 model (version 2.27, now the Solar 162
Irradiance Platform, SIP) [Tobiska et al., 2000]. The scattered component of solar Lyman-α flux 163
is included using the empirical approximation given by Thomas and Bowman [1986]. The SIC 164
code includes vertical transport [Chabrillat et al., 2002] which takes into account molecular 165
[Banks and Kockarts, 1973] and eddy diffusion with a fixed eddy diffusion coefficient profile 166
which has a maximum of 1.2×106 cm2s-1 at 110 km. The background neutral atmosphere is 167
calculated using the MSISE-90 model [Hedin, 1991] and tables given by Shimazaki [1984]. The 168
Wednesday, September 17, 2008
8
SIC-models does not calculate temperature variations, leaving these fixed by MSIS. As such no 169
transport driven by adiabatic heating or cooling are included in the SIC results. Such changes in 170
vertical transport have been calculated for large SPE events [Jackman et al., 2007], but were 171
insignificant below 60 km altitudes. In the SIC model transport and chemistry are advanced in 172
intervals of 5 minutes. Within each 5 minute interval exponentially increasing time steps are 173
used because of the wide range of chemical time constants of the modeled species. 174
3.1 Control Run 175
In order to interpret the SPE-driven changes, a SIC modeling run has also been undertaken 176
without any SPE-forcing (i.e., zero proton fluxes), termed the "control" run. The SIC model is 177
run for the northern hemisphere location (70ºN, 0ºE) and southern hemisphere location (70ºS, 178
0ºE) starting on 27 August 1859 and continuing for 24 days. These locations were selected as the 179
geomagnetic cutoff energy is small for sufficiently high magnetic latitudes, such that the proton 180
flux spectra is essentially unaffected by the geomagnetic field, particularly for the mesospheric 181
altitudes of interest. Modeling of the 1850 geomagnetic field suggests there is little change in 182
cutoff rigidities for these locations relative to the modern field [Shea and Smart, Fig. 4, 2006], 183
and hence modern rigidity calculations can be applied [e.g., Rodger et al., 2006a]. In addition, 184
for these locations UT=LT, making interpretation easier. Finally, the northern location is the 185
same as has been used in some previous SIC-modeling studies into SPE-effects [e.g., Verronen 186
et al., 2005; Clilverd et al, 2006; Seppälä et al., 2006], allowing direct comparisons. 187
We assume active solar cycle phase for the SOLAR2000 output (F10.7 = 158.2×10-22 Wm-2Hz- 188
1, F10.7A = 167.7 ×10-22 Wm-2Hz-1), and drive the MSIS model with Ap = 138 based on the 189
mean storm value determined for the Carrington period [Nevanlinna, 2006]. Note that the 190
geomagnetic amplitude index C9 from the St. Petersburg observatory (Russia) reached peak 191
values of 8 and 9 for the times of the two peaks in the SPE fluxes [Nevanlinna, Fig 3., 2006], 192
which will also ensure that the cutoff rigidities are very low for our modeling locations. We 193
Wednesday, September 17, 2008
9
therefore assume that the geomagnetic cutoff energy is zero throughout the calculation for both 194
locations. 195
The results of this no-forcing control SIC-run, shown in Figure 3, represent the calculation of 196
"normal" conditions, and hence allow an indication of the significance of the SPE-driven 197
changes. The top panel of Figure 3 shows the normal diurnal variation in electron number 198
density, the second panel shows HOx number density (H + OH + HO2), the third panel NOx 199
number density (N + NO + NO2), and the lower panel shows Ox (O + O3). We use HOx, NOx and 200
Ox rather than NO and O3 as there are substantial diurnal variations in both the latter populations, 201
which would lead to distracting features in the relative change plots presented below. In all cases 202
these panels have units of log10[cm-3]. 203
The diurnal variation in the constituents is most clearly seen in the electron density and HOx 204
panels of Figure 3, but is much weaker for the NOx and Ox panels. This is because the chemical 205
lifetimes for the NOx and Ox species are relatively long, while the rapid changes taking place 206
during the diurnal cycle occur inside the family of species (i.e. NO2 and NO). There is a gradual 207
change present in the southern hemisphere HOx and NOx panels, due to increasing levels of 208
sunlight caused by the seasonal lengthening of the periods with daytime conditions as seen in the 209
electron density panels. 210
211
3.2 Proton Forcing 212
Ionization rates are calculated for the three possible representations of the Carrington SPE as 213
described in Verronen et al. [2005], and are shown in Figure 4 with units of log10[cm-3s-1]. As 214
the March 1991 spectrum leads to more high energy protons (>300 MeV) present in the SPE 215
than the August 1972 spectrum, the ionization rates are more significant at lower altitudes 216
(~20 km). However, the Weibull-fitted SPE spectrum for March 1991 also leads to additional 217
lower energy protons (<20 MeV) relative to August 1972, causing the ionization rates to be 218
larger than the August 1972 case for altitudes >~65 km. The hard spectra case (October 1989) 219
Wednesday, September 17, 2008
10
has significant ionization rates present for the lowest altitudes considered in our modeling (at 220
~20 km), where the rates are several orders of magnitude higher than for either of the other 221
spectra. In all cases the first ionization pulse is similar to the peak ionization rates seen for 222
other large SPE events [e.g., Verronen et al., Fig. 1, 2005, Seppälä et al., Fig. 1, 2006], while 223
the second pulse on 2 September 1859 is about one order of magnitude larger. 224
4. Modeling results 225
The ionization rates shown in Figure 4 are used to drive the SIC model. Hence we examine 226
the altitude and time variation in the electron number density and the neutral atmospheric 227
species (e.g., NOx (N + NO + NO2), HOx (H + OH + HO2), and Ox (O + O3)), during the 228
Carrington SPEs. The atmospheric changes modeled in our study mostly occur in the 229
mesosphere and upper stratosphere, as determined by energy spectra of the precipitating 230
protons. In the mesosphere changes in O3 (or Ox) are primarily caused by increases in HOx, 231
although NOx does play some role near 50 km and is important in Ox chemistry in the upper 232
stratosphere. Ionization-produced NOx and HOx leads to the Ox changes as shown in the 233
following figures. 234
4.1 Electron density variation 235
Figure 5 presents the Carrington SPE electron number density changes caused by the three 236
selected SPE-spectra, relative to the control runs shown in Figure 3, shown for the northern 237
hemisphere (left) and southern hemisphere (right) cases. The figure shows the ratio to represent 238
the very large changes which occur. The electron density plots show very large increases 239
during the period of direct SPE-forcing (i.e., 27 August-8 September 1859), with the two pulse 240
time-structure of the SPE clearly seen in the electron density relative changes. The peak 241
electron density changes in the 2nd pulse are roughly one order of magnitude larger than for the 242
1st pulse, as expected from the differences in the SPE forcing. The largest enhancements occur 243
Wednesday, September 17, 2008
11
in the southern hemisphere, which is dominated by night-time ionospheric conditions. The 244
three SPE spectra lead to rather similar electron density increases, of about 103-104 times in the 245
40-100 km altitude range. The primary difference between the 3 spectra considered is between 246
the August 1972 and October 1989 cases below 40 km altitude, to the relative hardness of the 247
spectra. Once the SPE forcing ends, the electron density rapidly returns to normal levels at 248
most altitudes. The exception to this is a long-lived factor of 3-9 increase which occurs at 249
~80 km altitude, caused by the Lyman-α ionization of increased NOx present after the SPE. 250
Note that this electron density enhancement feature does not completely disappear, and is still 251
present 10 days after the end of the SPE forcing. However, we are unable to determine the true 252
recovery time of this enhancement, as it lasts beyond the timescale over which our model runs 253
can be considered realistic. The length of the SIC-run needs to be limited due to the increasing 254
significance of horizontal transport, in addition to vertical transport from adiabatic heating, 255
neither of which are included in the 1D SIC model. In reality, the relatively small electron 256
density enhancement feature is likely to dissipate more rapidly than shown here, due to 257
transport mixing out the long-lived NOx increase. 258
4.2 HOx variation 259
Figure 6 presents the variation of HOx, in the same format as Figure 5. Increases in HOx are 260
only significant around the times of the SPE-forcing, as the lifetime of HOx is very short, and 261
the majority of the SPE-produced HOx rapidly decreases once forcing ends. As in the electron 262
case, the increases appear more significant in the southern hemisphere as the background level 263
of HOx is larger in the daytime than the night (Figure 3). All three spectra lead to peak HOx 264
increases of about one order of magnitude from ~20-85 km, with the two "hard" SPE spectra 265
causing significant HOx increases to the lowest altitudes considered. Order-of-magnitude HOx 266
increases have been seen also during smaller, more recent SPE, one example being the January 267
2005 event [Verronen et al., 2006; Seppälä et al., 2006]. However these high enhancements 268
Wednesday, September 17, 2008
12
have typically been restricted to mesospheric altitudes between 50-80 km. After the forcing, all 269
the SPE spectra produce a long-lived HOx increase of ~40% located around 60-80 km altitudes, 270
present during the nighttime periods. This persistent HOx increase is related to the increase of 271
NO. Larger amounts of NO lead to increased ionization in the D-region, even at night-time 272
because of Lyman-α radiation scattered from the geocorona being an important source of ions. 273
Increased ionization leads to more HOx production through ion chemistry during both day and 274
night. However, this occurs on a relatively low level so that this change is only seen at night 275
when the background production of HOx, which is dependent upon solar radiation, is low. The 276
persistent HOx increase will influence Ox destruction at these altitudes, leading to a long-lived 277
loss. 278
4.3 NOx variation 279
The SPE-produced NOx increases are shown in Figure 7, again in the same format as Figure 5. 280
In this case, the NOx increases are more significant in the northern hemisphere as the 281
background levels of NOx are significantly lower in the sunlight hemisphere (12 times less NOx 282
in the northern hemisphere), and hence the relative NOx increase in the northern hemisphere is 283
~5 times larger than the southern hemisphere. The production rate is the same in both 284
hemispheres, and while the loss rate is larger in the more sunlit northern hemisphere, the 285
relative peak increases are also larger. The faster decay rates in the northern hemisphere due to 286
additional levels of sunlight can be seen in this figure, such that the NOx increases at the end of 287
the SIC modeling period are larger in the southern hemisphere, even though the peak change is 288
larger in the northern. The most significant NOx increases occur from 40-85 km altitude, where 289
the background NOx levels are very low (Figure 3, panel 3). The SPE-produced increases are 290
roughly 100-1000 times the background levels, and thus are much like creating lower 291
thermospheric NOx concentrations at mesospheric altitudes where NOx concentrations are 292
normally very low. Again, there is very little difference between the NOx changes produced by 293
Wednesday, September 17, 2008
13
the March 1991 and October 1989 SPE spectra, while the softer August 1972 spectra leads to 294
11-14 times less NOx around 80 km altitude than for the harder spectra (depending on the 295
hemisphere considered). The NOx produced at altitudes below ~60 km altitudes is very long 296
lived in all conditions as NOx is normally destroyed by solar radiation, which has been largely 297
absorbed at higher altitudes causing very little photodissociation of low-altitude enhancements. 298
Once again, this is an area in which transport needs to be considered, and would be a 299
reasonable topic for a future study. 300
4.4 Ox variation 301
The effect of the HOx and NOx increases on Ox is shown in Figure 8. In this case the relative 302
changes (the ratio between the SPE run and the control run) are shown on a linear scale. During 303
the peak SPE-forcing periods, Ox concentrations drop by 80-90% across an altitude range of 304
50-80 km, with minimum values of 11-13% of the ambient Ox. The two hard spectra 305
representations produce unusually broad Ox decreases stretching over a wider altitude range 306
than seen for most large SPE. For example, the large SPEs which occurred in January 2005 307
produced Ox decreases of about 80% over 70-80 km altitude [Fig2.,Seppälä et al., 2004], while 308
the Carrington SPE is likely to have led to a ~90% decrease over the wider altitude range of 309
~60-80 km. These very large decreases in Ox which occur during the SPE-forcing are caused by 310
HOx. However, this quickly returns to near normal levels at most altitudes (Figure 6) once the 311
proton forcing has finished, after which the mesospheric Ox largely recovers. A long-lived but 312
relatively small nighttime Ox decrease of 5%, at 60-80 km altitude remains after the forcing, 313
caused by the previously identified HOx feature in Figure 6. However, after the SPE forcing 314
there is a significant long lived Ox decrease (Ox levels drop of ~40% when compared with 315
normal levels) at ~45 km altitude, i.e. in the upper stratosphere. This is an unusually large 316
decrease for a direct SPE effect upon these relatively low altitudes and is produced by the long- 317
lived NOx with an order of magnitude increase occurring at ~45 km altitude (Figure 7). The 318
Wednesday, September 17, 2008
14
first pulse in the SPE leads directly to a ~10-20% Ox decrease at these altitudes, which is more 319
typical for very large SPE [e.g. Seppälä et al., 2004]. All the three possible Carrington spectra 320
create this large low-altitude Ox decrease. Since protons of energy 30 MeV penetrate to ~50 km 321
and the three proton flux spectra are normalized to the same value for proton energies 322
>30 MeV, this was not unexpected. This is an indication of the extreme changes possible for 323
the largest SPE, and is a feature not seen in "normal" large SPE, even those with unusually hard 324
spectra [Seppälä et al., 2004, 2008], where the relative SPE-produced NOx increase are 325
insignificant due to the very large background NOx concentrations at these altitudes (Figure 3). 326
4.5 Synthesis 327
The three possible proton spectra representing the Carrington event lead to rather different 328
atmospheric ionization rates (Figure 4), which in turn produce somewhat different responses 329
for the electron number density profiles and neutral atmospheric constituents. There is also 330
some difference in the relative response between the northern and southern hemisphere due to 331
the relative levels of sunlight, either by pre-conditioning the background conditions (e.g., 332
electrons, HOx, NOx), or by driving the direct loss rates (e.g., NOx). In all cases the largest 333
changes occur during the two-pulse solar proton event itself, across the ~12 days in which there 334
is direct SPE forcing. During the period of direct SPE forcing the nature of the change depends 335
somewhat upon the SPE-spectra and hemisphere considered. The first pulse of the SPE leads to 336
changes which are similar to those produced by the largest SPE previously considered, while 337
the second pulse generally drives considerably larger changes. After the period of SPE-forcing 338
has finished there is much less difference in the calculated chemical effects between the 339
different SPE-spectra and hemisphere calculations than the period during the SPE forcing. This 340
is particularly the case for odd oxygen (Ox), likely the most important long-lived change driven 341
by large SPE. In all cases considered here the Carrington SPE produces a significant and 342
unusually strong long-lived Ox decrease (Ox levels drop by ~40%) at ~45 km altitude, i.e. in the 343
Wednesday, September 17, 2008
15
upper stratosphere due to NOx increases. As the nature of this NOx increase does not vary 344
significantly by SPE-spectra or hemisphere, there is relatively little variation in the long lived 345
Ox decrease. 346
Thomas et al. [2007] used ionization rates scaled from the October 1989 SPE to describe the 347
impact of the Carrington SPE upon long-lived ozone levels using a two dimensional 348
atmospheric model. Figure 9 shows the variation in the O3 column above 30 km altitude 349
determined from our calculations, for the three possible Carrington spectra and considering 350
both hemispheres. The maximum decrease in the >30 km O3 column is ~9% for both 351
hemispheres, with the somewhat softer spectra (March 1991) leading to maximum decrease of 352
~7%. As there is no immediate impact of the Carrington SPE upon O3 below 30 km altitude, 353
this 7-9% variation represents all the change in total column O3 which would be produced by 354
the SPE on short-time scales, before transport processes become significant. Transport 355
processes are likely to cause a decent in altitude of the SPE-produced NOx inside the winter 356
pole, leading to larger decreases in Ox and to more significant decreases in column O3. As the 357
one dimensional SIC model does not extend to low altitudes, and does include some significant 358
sources of vertical and horizontal transport, it is not possible to make a direct comparison with 359
the results of Thomas et al. [2007] from their two dimensional model. However, the small 360
differences between the calculated >30 km O3 column indicates the atmospheric response of 361
the Carrington SPE is not strongly dependent upon the spectra, suggesting that the calculations 362
of Thomas et al. [2007] are likely to be reasonable, at least within the overall uncertainties 363
associated with modeling this event. 364
The primary differences between the southern and northern hemispheric runs are due to the 365
very differing levels of sunlight. In order to further test this, and to consider the possible 366
"extreme" effect of the Carrington SPE, we have also undertaken a SIC-modeling run at the 367
southern hemisphere location (70ºS, 0ºE) starting from 15 July, and thus considering a period in 368
Wednesday, September 17, 2008
16
which there is almost no direct solar illumination. While it is complex to directly compare the 369
results of this run with the earlier calculations described above due to the very different 370
background conditions, there are no dramatic differences between the southern hemisphere 371
polar winter run starting 15 July and the run for the actual Carrington event times. The minor 372
differences (not shown) are dependent upon the levels of sunlight, as expected. For example, 373
the NOx loss at ~80 km is very low, such that almost all of the NOx produced by the SPE 374
remains to the end of the modeling run. In addition, the effect of the SPE during the forcing 375
period is more significant, with deeper longer-lived Ox losses from 60-80 km altitude that 376
recover more slowly (lasting 3-4 days more). However, the long-lived Ox decrease at 40-50 km 377
altitude is slightly less significant in the polar night run (decreases of ~40% rather than ~60%). 378
This is because the catalytic cycles of NOx require atomic oxygen to be present, which is only 379
released by photodissociation. While there are numerous minor differences between the polar 380
night runs and the existing southern hemisphere runs, they are small enough that we can 381
conclude the existing northern and southern hemisphere calculations provide a sufficiently 382
accurate indication of the significance of the Carrington-event of 1859. Additional 383
interhemispheric differences will also arise due to differences between the circulation patterns 384
in both hemispheres and their seasonal variability, which is not captured in the existing 1D 385
model used in our study. While this is likely to be significant when considering the long-time 386
scale response, it is unlikely to produce large changes in the calculations as presented. Over 387
time the relative importance of horizontal and vertical transport will increase, processes which 388
are not fully included in the one dimensional SIC model. However, the SIC modeling does 389
allow comparison between the immediate effects of the three different Carrington spectra in 390
either hemisphere. As such, these calculations should also provide a reasonable estimate of the 391
immediate impact of a future Carrington-like event striking the Earth's atmosphere. 392
Wednesday, September 17, 2008
17
5. Effect on HF communication 393
The additional ionospheric ionization caused by solar proton events lead to "polar blackouts", 394
also known as "polar cap absorption (PCA)" events, which are disruptions to HF/VHF 395
communications in high-latitude regions caused by attenuation in the ionospheric D-region 396
[Davies, 1990]. The additional attenuation can make HF communications impossible throughout 397
the polar regions, areas where HF communication is particularly important for the international 398
aviation industry; on some occasions major airlines have cancelled trans-polar flights due to such 399
space weather events [Jones et al., 2005], while the practice of changing the flightpaths to avoid 400
the poles leads to increased fuel consumption. 401
In order to characterize the ionospheric significance of the Carrington-SPE on HF attenuation 402
levels, we consider the variation with time of the Highest Affected Frequency (HAF) during the 403
SPE. The HAF is defined as the frequency which suffers a loss of 1 dB during vertical 404
propagation from the ground, through the ionosphere, and back to ground. Radio frequencies 405
lower than the HAF suffer an even greater loss. Here we determine the HAF by contrasting the 406
SPE-produced electron density changes with those expected from Solar Flares, following the 407
approach outlined in Rodger et al. [2006b]. As an example, an X20 flare, which has peak 0.1- 408
0.8 nm X-ray fluxes of 2.0 mW m-2, produces a HAF of 38 MHz. Flares of this magnitude lead 409
to "extreme" Radio Blackouts, with essentially no HF radio contact with mariners or en-route 410
aviators. NOAA has defined a Space Weather Scale for Radio Blackouts [Poppe, 2000], ranging 411
from R1 describing a minor disruption due to an M1 flare (10 µW m-2 peak 0.1-0.8 nm X-ray 412
flux) to R5 for the extreme blackout case described above. We will employ this scale to provide 413
an indication of the severity of the SPE-induced polar blackouts. 414
Figure 10 shows the HF blackout estimates for the Carrington SPE. The left panels shows the 415
blackouts estimated for the August 1972 spectra, while the right panels are for the March 1991 416
spectra, where the northern hemisphere results are in black and the southern hemisphere results 417
Wednesday, September 17, 2008
18
in red. The HF blackouts estimated for October 1989 are very similar to the March 1991 case, 418
due to the very similar electron density profiles (Figure 5), and hence are not shown in Figure 10. 419
The upper panels of Figure 10 show the equivalent peak X-ray flux in the 0.1-0.8 nm 420
wavelength range which would cause the same ionosphere electron density change during a solar 421
flare, represented using the H' parameter of Wait and Spies [1964]. For both the northern 422
hemisphere and southern hemisphere cases, and both spectra, the SPE-produced disruptions are 423
equivalent to very large equivalent peak X-ray fluxes. For context, there are about 175 solar 424
flares with peak X-ray fluxes of X1 or above per solar cycle (~16 per year), and 8 flares >X10 425
per cycle. The upper panel of Figure 10 also indicates the X45 threshold, representing the largest 426
solar flare peak X-ray flux measured since about 1976 [Thomson et al., 2004; 2005]. In both 427
cases the SPE-produced ionospheric disturbance peaks around the X45 threshold, and is larger 428
than the X10 threshold for ~15.5 hours. In contrast, the X45 flare of 4 November 2003 led to X- 429
ray fluxes which were >X10 for ~20 min [Thomson et al., Fig. 2, 2004]. 430
The lowest panels of Figure 10 presents the Highest Affected Frequency calculated from the 431
equivalent peak X-ray powers following the empirically derived relationship between HAF and 432
solar 0.1-0.8 nm X-ray flux provided by the Space Environmental Forecaster Operations 433
Manual (1997). The NOAA Radio Blackout Scale has been added for comparison. While the 434
peak HAF is smaller for the softer August 1972 spectra, in both cases and hemispheres the 435
Carrington SPE would have produced polar blackouts equivalent to the "Extreme" threshold of 436
the NOAA Radio Blackout Scale for a time period of ~15 hours, and led to some disturbances in 437
polar HF communications for ~11-12 days. While this is long in comparison to most SPE 438
disruptions, it is not wildly larger than the few days of disruption caused by most large SPE- 439
events. As such, the polar communications disruptions associated with the Carrington SPE would 440
not be severe, despite the exceptional nature of the SPE itself. 441
Wednesday, September 17, 2008
19
6. Discussion and Summary 442
The Carrington event of August/September 1859 was the most significant solar proton event 443
(SPE) of the last 450 years, with about four times larger solar proton fluence than the largest 444
event from the "spacecraft era" (August 1972). The space weather event which occurred at this 445
time produced the most intense geomagnetic storm in recorded history and the first visual 446
observation of a white-light solar flare. Recently, much attention has focused upon increasing 447
our understanding of the Carrington event, in order to better quantify the impact of extreme 448
space weather events. In this study we have used new findings as to the nature of the 449
Carrington event, adopting an intensity-time profile for the >30 MeV solar proton flux, and 450
examining the relative atmospheric response to three different SPE-energy spectra which have 451
previously been used to represent the Carrington SPE. Cosmogenic isotope concentrations have 452
indicated that the Carrington SPE likely had a comparatively soft energy spectrum, for example 453
similar to those measured for the August 1972 or March 1991 SPE, rather than the October 454
1989 SPE spectra sometimes used to model the impact of the Carrington SPE. The Sodankylä 455
Ion and Neutral Chemistry (SIC) model has been used to estimate the impact of the Carrington 456
event to the neutral atmosphere and the ionosphere, and the disruption to HF communication. 457
As seen in the SIC-output plots described in this paper, large changes in electron density and 458
atmospheric constituents occur during the period of SPE-forcing, which depend upon the nature 459
of the spectrum and also on the hemisphere considered. This is particularly significant for the 460
electron density increases. However, the most important SPE-driven atmospheric response is 461
the long-lived Ox decreases in the upper stratosphere (Ox levels drop by ~40% of the normal 462
level). This change does not significantly vary between the 3 spectra, or between the 2 463
hemispheres, due to the very small differences in the long-lived low-altitude NOx increases 464
which produce the Ox decreases. All the three possible Carrington spectra create this large low- 465
Wednesday, September 17, 2008
20
altitude Ox decrease due to the very large >30 MeV proton fluxes present in the normalized 466
spectra. This is an indication of the extreme changes possible for the very largest SPE, and is 467
not a feature seen previously for "normal" large SPE, even those with unusually hard spectra. 468
We have also characterized the ionospheric significance of the Carrington-SPE on HF 469
attenuation levels. Should a Carrington-level event occur in the current era, it would cause 470
disruptions to HF/VHF communications in high-latitude regions, making HF communications 471
impossible throughout the polar regions for some time. This could have a significant impact on 472
the routing of trans-polar aeroplane travel. The Carrington SPE would have produced polar 473
blackout equivalent to the "Extreme" threshold of the NOAA Radio Blackout Scale for a time 474
period of ~15 hours, and led to some disturbances in polar HF communications over a time 475
period of ~11-12 days, (i.e., during the period of direct SPE forcing). While this is long in 476
comparison to most SPE disruptions, it is not wildly larger than the few days of disruption 477
caused by most large SPE-events. As such, the polar communications disruptions associated 478
with the Carrington SPE would not be particularly severe, despite the exceptional nature of the 479
SPE itself. 480
In general, the atmospheric and ionospheric response is somewhat different between the 481
August 1972 or March 1991 SPE spectra, while the calculations using the March 1991 and 482
October 1989 SPE spectra are very similar to one another. Thus while cosmogenic isotope 483
concentrations from ice-cores indicate that the Carrington-SPE was comparatively soft, the 484
conclusions of previous modeling studies into the atmospheric response of the Carrington SPE 485
which have used rather hard spectra, and specifically the October 1989 SPE spectra, are likely 486
to be reasonable, at least within the overall uncertainties associated with modeling this event. 487
This is particularly important to the conclusions of Thomas et al. [2007], who used ionization 488
rates scaled from the October 1989 SPE to describe the impact of the Carrington SPE upon 489
long-lived ozone levels. These authors concluded that the globally averaged column-ozone 490
Wednesday, September 17, 2008
21
would decrease by as much as 4%, recovering slowly over several years. While this ozone 491
depletion is small in a global sense, it is accompanied by much larger decreases in the poles, 492
with long-lived polar ozone losses which are similar to those calculated in our study. As noted 493
by Thomas et al., even small increases in UVB can be harmful to many life forms. In addition, 494
changes in the chemical balance of the upper and middle stratosphere may be associated with 495
changes in polar winds and temperatures, and may even lead to few degree variations in sea- 496
level temperatures [Rozanov et al., 2005]. 497
498
Acknowledgments. CJR would like to thank Gillian Rodger of Manchester for her support. 499
PTV and AS are partly funded by the Academy of Finland through project 123275, 500
Thermosphere and Mesosphere affecting the Stratosphere. AS would also like to thank the 501
Academy of Finland for their support of her work through the EPPIC project. 502
References 503
Banks, P. M., and G. Kockarts (1973), Aeronomy, vol. B, chap. 15, Academic Press. 504
Beer, J., A. Blinov, G. Bonani, R. C. Finkel, H. J. Hofmann, B. Lehmann, H. Oeschger, A. Sigg, 505
J. Schwander, T. Staffelbach, B. Stauffer, M. Suter, and W. Wötfli (1990), Use of 10Be in 506
polar ice to trace the 11-year cycle of solar activity, Nature, 347, 164–166, 507
doi:10.1038/347164a0. 508
Boteler, D. H. (2006), The super storms of August/September 1859 and their effects on the 509
telegraph system, Advances Space Res., 38(2), 159-172. 510
Brasseur, G., and S. Solomon (1986), Aeronomy of the Middle Atmosphere, second ed., D. 511
Reidel Publishing Company, Dordrecht. 512
Carrington, R.C. (1860), Description of a singular appearance seen in the Sun on September 1, 513
1859. Monthly Not. R. Astron. Soc. 20, 13–15. 514
Wednesday, September 17, 2008
22
Clilverd, M. A., A. Seppälä, C. J. Rodger, N. R. Thomson, P. T. Verronen, E. Turunen, Th. 515
Ulich, J. Lichtenberger, and P. Steinbach, Modeling polar ionospheric effects during the 516
October-November 2003 solar proton events, Radio Sci., 41(2), doi: 10.1029/2005RS003290, 517
2006. 518
Chabrillat, S., G. Kockarts, D. Fonteyn, and G. Brasseur (2002), Impact of molecular diffusion 519
on the CO2 distribution and the temperature in the mesosphere, Geophys. Res. Lett., 29, 1-4. 520
Crutzen, P. J., I. S. A. Isaksen, and G. C. Reid (1975), Solar proton events: Stratospheric sources 521
of nitric oxide, Science, 189, 457–458. 522
Crutzen, P., and S. Solomon (1980), Response of mesospheric ozone to particle precipitation, 523
Planetary and Space Science, 28, 1147–1153. 524
Davies, K. (1990), Ionospheric radio, Peter Peregrinus, London. 525
Hedin, A. E. (1991), Extension of the MSIS Thermospheric model into the middle and lower 526
Atmosphere, J. Geophys. Res., 96, 1159-1172. 527
Jackman, C. H., and R. D. McPeters (1985), The responce of ozone to solar proton events during 528
solar cycle 21: A theoretical interpretation, J. Geophys. Res., 90, 7955–7966. 529
Jackman, C. H., and P. E. Meade (1988), Effect of solar proton events in 1978 and 1979 on the 530
odd nitrogen abundance in the middle atmosphere, J. Geophys. Res., 93, 7084–7090. 531
Jackman, C. H., A. R. Douglass, R. B. Rood, R. D. McPeters, and P. E. Meade (1990), Effect of 532
solar proton events on the middle atmosphere during the past two solar cycles as computed 533
using a two-dimensional model, J. Geophys. Res., 95, 7417–7428. 534
Jackman, C. H., J. E. Nielsen, D. J. Allen, M. C. Cerniglia, R. D. McPeters, A. R. Douglass, and 535
R. B. Rood (1993), The effects of the October 1989 solar proton events on the stratosphere as 536
computed using a three-dimensional model, Geophys. Res. Lett., 20, 459–462. 537
Jackman, C. H., M. C. Cerniglia, J. E. Nielsen, D. J. Allen, J. M. Zawodny, R. D. McPeters, A. 538
R. Douglass, J. E. Rosenfield, and R. B. Hood (1995), Two-dimensional and three- 539
Wednesday, September 17, 2008
23
dimensional model simulations, measurements, and interpretation of the October 1989 solar 540
proton events on the middle atmosphere, J. Geophys. Res., 100, 11,641–11,660. 541
Jackman, C. H., Fleming, E. L., Chandra, S., Considine, D. B., and Rosenfield, J. E., Past, 542
present, and future modeled ozone trends with comparisons to observed trends (1996), J. 543
Geophys. Res., 101 , 28 753–28 768. 544
Jackman, C. H., E. L. Fleming, and F. M. Vitt (2000), Influence of extremely large solar proton 545
events in a changing stratosphere, J. Geophys. Res., 105, 11,659–11,670. 546
Jackman, C. H., R. D. McPeters, G. J. Labow, E. L. Fleming, C. J. Praderas, and J. M. Russel 547
(2001), Northern hemisphere atmospheric effects due to the July 2000 solar proton events, 548
Geophys. Res. Lett., 28, 2883–2886. 549
Jackman, C. H., R. G. Roble, and E. L. Fleming (2007), Mesospheric dynamical changes induced 550
by the solar proton events in October–November 2003, Geophys. Res. Lett., 34, L04812, 551
doi:10.1029/2006GL028328. 552
Jones, J. B. L., R.D. Bentley, R. Hunter, R.H.A. Iles, G.C. Taylor and D.J. Thomas (2006), 553
Space weather and commercial airlines, Advances Space Res., 36(12), 2258-2267. 554
MacNamara, L. F. (1985), High frequency radio propagation, in Handbook of Geophysics and 555
the Space Environment, edited by A. S. Jursa, Chapter 10, pp. 45-65, Air Force Geophys. 556
Lab., U. S. Air Force. 557
McCracken, K. G., G. A. M. Dreschhoff, E. J. Zeller, D. F. Smart, and M. A. Shea (2001a), Solar 558
cosmic ray events for the period 1561–1994. 1. Identification in polar ice, 1561–1950. J. 559
Geophys. Res. 106 (A10), 21585–21598. 560
McPeters, R. D., and C. H. Jackman (1985), The response of ozone to solar proton events during 561
solar cycle 21: the observations, J. Geophys. Res., 90, 7945–7954. 562
Nevanlinna, H. (2006), A study on the great geomagnetic storm of 1859: Comparisons with other 563
storms in the 19th century, Adv. in Space Res., 38 (2), 180-187. 564
Wednesday, September 17, 2008
24
Odenwald, S., J. Green, and W. Taylor (2006), Forecasting the impact of an 1859-calibre 565
superstorm on satellite resources, Advances Space Res., 38(2), 280-297. 566
Poppe, B. (2000), New scales help public, technicians understand space weather, Eos Trans., 567
Vol. 81 (29), 322, July 18. 568
Rodger, C. J., M. A. Clilverd, P. T. Verronen, Th. Ulich, M. J. Jarvis, and E. Turunen (2006a), 569
Dynamic geomagnetic rigidity cutoff variations during a solar proton event, J. Geophys. Res., 570
111, A04222, doi:10.1029/2005JA011395. 571
Rodger, C. J., M. A. Clilverd, Th. Ulich, P. T. Verronen, E. Turunen, and N. R. Thomson 572
(2006b), The atmospheric implications of Radiation Belt Remediation, Annales Geophys., 573
24(7), 2025 - 2041, SRef-ID: 1432-0576/ag/2006-24-2025. 574
Seppälä, A., P. T. Verronen, E. Kyrölä, S. Hassinen, L. Backman, A. Hauchecorne, J. L. Bertaux, 575
and D. Fussen (2004), Solar proton events of October-November 2003: Ozone depletion in 576
the Northern hemisphere polar winter as seen by GOMOS/Envisat, Geophys. Res. Lett., 577
31(19), L19,107, doi:10.1029/2004GL021042. 578
Seppälä, A., P. T. Verronen, V. F. Sofieva, J. Tamminen, E. Kyrölä, C. J. Rodger, and M. A. 579
Clilverd (2006), Destruction of the tertiary ozone maximum during a solar proton event, 580
Geophys. Res. Lett., 33, L07804, doi:10.1029/2005GL025571. 581
Seppälä, A., M. A. Clilverd, C. J. Rodger, P. T. Verronen, and E. Turunen (2008), The effects of 582
hard spectra solar proton events on the middle atmosphere, J. Geophys. Res., in press. 583
Shea, M. A., D. F. Smart, K. G. McCracken, G. A. M. Dreschoff, and H. E. Spence (2006), Solar 584
proton events for 450 years: The Carrington event in perspective, Adv. Space Res., 38 (2), 585
232-238. 586
Shea, M. A., and D. F. Smart (2006), Geomagnetic cutoff rigidities and geomagnetic coordinates 587
appropriate for the Carrington flare epoch, Adv. Space Res. 38 (2006), pp. 209–214. 588
Wednesday, September 17, 2008
25
Shimazaki, T. (1984), Minor Constituents in the Middle Atmosphere (Developments in Earth 589
and Planetary Physics, No 6), D. Reidel Publishing Co., Dordrecht, Netherlands. 590
Smart, D. F., M. A. Shea, and K. G. McCracken (2006), The Carrington event: Possible solar 591
proton intensity–time profile, Adv. Space Res., 38 (2), 215-225. 592
Solomon, S., D. W. Rusch, J.-C. Gérard, G. C. Reid, and P. J. Crutzen (1981), The effect of 593
particle precipitation events on the neutral and ion chemistry of the middle atmosphere: II. 594
Odd hydrogen, Planet. Space Sci., 8, 885–893. 595
Solomon, S., G. C. Reid, D. W. Rusch, and R. J. Thomas (1983), Mesospheric ozone depletion 596
during the solar proton event of July 13, 1982 Part II. Comparisons between theory and 597
measurements, Geophys. Res. Lett., 10, 257–260. 598
Space Environmental Forecaster Operations Manual, page 4.3.1, 55th Space Weather Squadron, 599
Falcon AFB, USAF, 21 October 1997. As referenced by the NOAA Space Environment 600
Center at http://www.swpc.noaa.gov/dregion/dregionDoc.html 601
Rozanov, E., L. Callis, M. Schlesinger, F. Yang, N. Andronova, and V. Zubov (2005), 602
Atmospheric response to NOy source due to energetic electron precipitation, Geophys. Res. 603
Lett., 32, L14811, doi:10.1029/2005GL023041. 604
Thomas, B. C., C. H. Jackman, and A. L. Melott (2007), Modeling atmospheric effects of the 605
September 1859 solar flare, Geophys. Res. Lett., 34, L06810, doi:10.1029/ 2006GL029174. 606
Thomas, L., and M. R. Bowman (1986), A study of pre-sunrise changes in negative ions and 607
electrons in the D-region, Ann. Geophys., 4, 219-228. 608
Thomson, N. R., C. J. Rodger, and R. L. Dowden (2004), Ionosphere gives size of greatest solar 609
flare, Geophys. Res. Lett., 31(6), L06803, 10.1029/2003GL019345. 610
Thomson, N. R., C. J. Rodger, and M. A. Clilverd (2005), Large solar flares and their 611
ionospheric D-region enhancements, J. Geophys. Res., 110, A06306, 612
doi:10.1029/2005JA011008. 613
Wednesday, September 17, 2008
26
Tobiska, W. K., T. Woods, F. Eparvier, R. Viereck, L. D. B. Floyd, G. Rottman, and O. R. White 614
(2000), The SOLAR2000 empirical solar irradiance model and forecast tool, J. Atmos. Terr. 615
Phys., 62, 1233-1250. 616
Townsend, L. W. (2003), Carrington flare of 1859 as a prototypical worst-case solar energetic 617
particle event, IEEE Trans. Nucl. Sci., 50(6), 2307-2309. 618
Tsurutani, B. T., W. D. Gonzalez, G. S. Lakhina, and S. Alex (2003), The extreme magnetic 619
storm of 1–2 September 1859, J. Geophys. Res., 108(A7), 1268, doi:10.1029/2002JA009504. 620
Turunen, E., H. Matveinen, J. Tolvanen, and H. Ranta (1996), D-region ion chemistry model, in 621
STEP Handbook of Ionospheric Models, edited by R. W. Schunk, pp. 1-25, SCOSTEP 622
Secretariat, Boulder, Colorado, USA. 623
Turunen, E., P. T. Verronen, A. Seppälä, C. J. Rodger, M. A. Clilverd, J. Tamminen, C. F. Enell 624
and Th. Ulich (2008), Impact of different precipitation energies on NOx generation during 625
geomagnetic storms, J. Atmos Sol.-Terr. Phys., in press. 626
Verronen, P. T., E. Turunen, Th. Ulich, and E. Kyrölä (2002), Modelling the effects of the 627
October 1989 solar proton event on mesospheric odd nitrogen using a detailed ion and neutral 628
chemistry model, Ann. Geophys., 20, 1967-1976. 629
Verronen, P. T., A. Seppälä, M. A. Clilverd, C. J. Rodger, E. Kyrölä, C-F. Enell, Th. Ulich, E. 630
and Turunen (2005), Diurnal variation of ozone depletion during the October-November 2003 631
solar proton event, J. Geophys. Res., 110(A9), doi:10.1029/2004JA010932. 632
Verronen, P. T., A. Seppälä, E. Kyrölä, J. Tamminen, H. M. Pickett, E. and Turunen, Production 633
of odd hydrogen in the mesosphere during the january 2005 solar proton event (2006), 634
Geophys. Res. Lett., 33, L24811, doi:10.1029/2006GL028115. 635
Wait, J. R., and K. P. Spies (1964), Characteristics of the Earth-ionosphere waveguide for VLF 636
radio waves, NBS Tech. Note 300, Nat. Inst. of Stand. and Technol., Gaithersburg, Md.. 637
Wednesday, September 17, 2008
27
Xapsos, M. A., J. L. Barth, E. G. Stassinopoulos, S. R. Messenger, R. J. Walters, G. P. Summers, 638
and E. A. Burke (2000), Characterizing solar proton energy spectra for radiation effects 639
applications, IEEE Trans. Nuclear Sci., 47(6). 640
641
M. A. Clilverd and A. Seppälä, Physical Sciences Division, British Antarctic Survey 642
(NERC), High Cross, Madingley Road, Cambridge CB3 0ET, England, U.K. (e-mail: 643
macl@bas.ac.uk, annika.seppala@fmi.fi). 644
C. J. Rodger, Department of Physics, University of Otago, P.O. Box 56, Dunedin, New 645
Zealand. (email: crodger@physics.otago.ac.nz). 646
E. Turunen, Sodankylä Geophysical Observatory, University of Oulu, Sodankylä, Finland. 647
(email: esa@sgo.fi). 648
P. T. Verronen, Earth Observation, Finnish Meteorological Institute, P.O. Box 503 (Erik 649
Palménin aukio 1), FI-00101 Helsinki, Finland. (email: pekka.verronen@fmi.fi). 650
651
RODGER ET AL.: ATMOSPHERIC IMPACT OF CARRINGTON SPE 652
Wednesday, September 17, 2008
28
653
Table 1. Weibull fitting parameters used in equation 1 to represent the differential SPE fluxes 654
of the Carrington event with the 4 different energy spectra as shown in Figure 1. 655
656
Date Φ
0 k α
4 August 1972 5.0033 × 1010 0.0236 1.108
29 September 1989 4.5751 × 1011 0.877 0.3841
19 October 1989 4.4280 × 1012 2.115 0.2815
23 March 1991 1.4039 × 1012 0.972 0.441
657
658
Figures 659
660
661
Figure 1. Comparison between the normalised differential fluences which have been used to 662
describe the Carrington event, based on four previous SPE. The values shown have been 663
normalised to the Carrington-level >30 MeV fluence of 1.9×1010 protons cm-2. [See the online 664
version for the color version of this figure]. 665
Wednesday, September 17, 2008
29
666
Wednesday, September 17, 2008
30
667
Wednesday, September 17, 2008
31
Figure 2. Proton fluxes used to describe the Carrington event SPE. The upper panel shows the 668
>30 MeV intensity-time proton profile [after Smart et al., Fig. 12, 2006]. The lower 3 panels 669
present the different time-varying differential proton fluxes used in this study to describe 670
Carrington-level SPE. [See the online version for the color version of this figure]. 671
672
673
Wednesday, September 17, 2008
32
Figure 3. The results of a SIC modeling run without any SPE-forcing (i.e., zero precipitating 674
proton fluxes), showing the calculated "normal" conditions for the northern (left) and southern 675
(right) hemispheres. Units are shown in log10[cm-3]. [See the online version for the color 676
version of this figure]. 677
678
679
Wednesday, September 17, 2008
33
680
Wednesday, September 17, 2008
34
Figure 4. Atmospheric ionization rates calculated from the SPE fluxes shown in the lower 3 681
panels of Figure 2, given in units of log10[cm-3s-1]. [See the online version for the color version 682
of this figure]. 683
684
685
Figure 5. SPE-driven changes in electron number density determined from the SIC model for 686
the varying SPE spectra, and show as the ratio to the control run (Figure 3). The left panels are 687
for the northern hemisphere, while the right are the southern hemisphere. [See the online 688
version for the color version of this figure]. 689
690
691
Wednesday, September 17, 2008
35
692
Figure 6. SPE-driven changes in odd hydrogen (HOx) determined from the SIC model for the 693
varying SPE spectra, and show as the ratio to the control run (Figure 3). The left panels are for 694
the northern hemisphere, while the right are the southern hemisphere. [See the online version 695
for the color version of this figure]. 696
697
Wednesday, September 17, 2008
36
698
Figure 7. SPE-driven changes in odd nitrogen (NOx) determined from the SIC model for the 699
varying SPE spectra, and show as the ratio to the control run (Figure 3). The left panels are for 700
the northern hemisphere, while the right are the southern hemisphere. [See the online version 701
for the color version of this figure]. 702
703
Wednesday, September 17, 2008
37
704
Figure 8. SPE-driven changes in odd oxygen (Ox) determined from the SIC model for the 705
varying SPE spectra, and show as the linear ratio to the control run (Figure 3). The left panels 706
are for the northern hemisphere, while the right are the southern hemisphere. [See the online 707
version for the color version of this figure]. 708
709
Wednesday, September 17, 2008
38
710
Figure 9. SPE-driven percentage changes in the >30 km altitude O3 total column for varying 711
Carrington SPE spectra. The upper panel shows the changes for the northern hemisphere (NH), 712
while the lower is for the southern hemisphere (SH). [See the online version for the color 713
version of this figure]. 714
715
Wednesday, September 17, 2008
39
716
717
Figure 10. Estimate of the severity of the Carrington-SPE forced HF polar blackout for the 718
August 1972 (left) and March 1991 (right) SPE-spectra. The upper panel shows the equivalent 719
peak X-ray fluxes in the 0.1-0.8 nm range which would cause the same ionospheric change 720
during a solar flare. The lower panel is the Highest Affected Frequency calculated from the 721
equivalent peak X-ray flux. The NOAA Radio Blackout Scale has been added for comparison. 722
[See the online version for the color version of this figure]. 723