GREAT FEP EVENTS AND SPACE WEATHER
1. RADIATION HAZARD LEVELS, ALERT FOR EVENT START,
PROBABILITIES OF FALSE AND MISSED ALERTS
L.I. Dorman 1,2, L.A. Pustil’nik 1, A. Sternlieb 1, and I.G. Zukerman 1
1
Israel Cosmic Ray/Space Weather Center and Emilio Segre’ Observatory, affiliated to Tel Aviv University,
Technion and Israel Space Agency, P.O.Box 2217, Qazrin 12900, ISRAEL
2
Cosmic Ray Department of IZMIRAN, Russian Academy of Science, Troitsk 142092, Moscow Region, RUSSIA
ABSTRACT
According to NOAA Space Weather Scales (SWS) are dangerous Solar Radiation Storms S5-extreme (flux level of
particles with energy > 10 MeV more than 105 cm-2sec-1), S4-severe (flux more than 104 cm-2sec-1), and S3-strong
(flux more than 103 cm-2sec-1). We assume to modernize this NOAA SWS by using fluence instead of flux and by
adding two much more rarely but more dangerous Solar Radiation Storms: S6-very extreme and S7-ultra extreme.
Solar Radiation Storms S3 – S7 are dangerous for communication and operation systems, for computer memories,
for astronauts in space station, for passengers and crew in commercial jets, and even for technology and people on
the ground (especially during S7 – S5 storms). The problem is how to forecast exactly these dangerous
phenomena. We show that exact forecast can be made by using high-energy particles (few GeV/nucleon and
higher) which transportation from the Sun is characterized by much bigger diffusion coefficient than for small and
middle energy particles. Therefore high energy particles came from the Sun much more early (8-20 minutes after
acceleration and escaping into solar wind) than main part of smaller energy particles caused dangerous situation
for electronics (more than 30-60 minutes later). We describe here principles and experience of automatically
working of program "FEP-Search" what determined on the basis of one-minute data the start of great FEP event
automatically by simultaneously increasing on 2.5 St. Dev. in two or three sections of neutron supermonitor. The
next 1-min data the program "FEP-Search" uses for checking that the observed cosmic ray intensity increase
reflects the beginning of real great FEP or not. If yes, automatically starts to work on line the programs "FEPResearch". We determine also the probabilities of false and missed alerts, and show that they are negligible.
GREAT FEP EVENTS IN THE PAST AND MODERNIZATION OF NOAA SPACE WEATHER SCALE
NOAA Space Weather Scale establishes 5 gradations of FEP events, what are called Solar Radiation Storms:
from S5 (the highest level of radiation, corresponded to the flux of solar protons with energy >10 MeV about
10 5 protons. cm 2 . sec1 ) up to S1 (the lowest level, the flux about 10 protons. cm 2 . sec1 for protons with energy
>10 MeV). From our opinion, by ground level CR neutron monitors and muon telescopes it is possible monitoring
and forecast (by using much higher energy particles than smaller energy particles caused the main radiation
hazard) FEP events of levels S5, S4 and S3. With increasing of FEP event level of radiation will increase the
accuracy of forecasting. According to Dorman (2002a,b), in the first, for satellite damage and influence on people
health and technology, on communications by HF radio-waves more important is the total fluency of FEP during
the event than the protons flux what is used now in NOAA Space Weather Scale; in the second, the level S5
(corresponds to the flux 105 protons.cm 2 . sec 1 , or fluency F  109 protons.cm2 for protons with
Ek  10 MeV ) is not maximal, but can be much higher and with much smaller probability than S5 (Dorman and
Venkatesan, 1993; Dorman and Pustil’nik,
1995, 1999). As it was shown recently by
McCracken et al. (2001), the dependence of
event probability from fluency can be
prolonged
at
least
up
to
F  2 1010 protons.cm2 for protons with
Ek  30 MeV , what was observed in FEP of
September 1869 according to data of nitrate
contents in polar ice (see Fig. 1).
Fig. 1. The dependence of FEP events
probability (number of events per year) from
the value of fluence according to direct
satellite and NM data, nitrate in polar ice data
and cosmogenic nuclide data on the moon
(McCracken et al., 2001).
This type of great dangerous events is
very rarely (about one in few hundred years).
According to Fig 1 it is not excluded that in principle can occurred very great FEP events with fluency in 10 and
even in 100 times bigger (correspondingly one in few thousand and one in several ten thousand years). So, we
suppose to correct the very important classification, developed by NOAA, in two directions: to use fluency F of
FEP during all event (in units protons..cm 2 ) instead of flux I, and to extend levels of radiation hazard. As result,
the modernized classification of FEP events is shown in Table 1.
Table 1. Extended NOAA Space Weather Scale for Solar Radiation Storms
FEP events radiation hazard
S7
S6
S5
Ultra
extreme
Biological: Lethal doze for astronauts, for passengers and crew on
commercial jets; great influence on people health and gene
mutations on the ground
Satellite operations: very big damages of satellites electronics and
computers memory, damage to solar panels, loosing of many
satellites
Other systems: complete blackout of HF (high frequency)
communications through polar and middle-latitude regions, big
position errors make navigation operations extremely difficult.
Very
extreme
Biological: About lethal doze for astronauts, serious influence on
passengers and crew health on commercial jets; possible influence
on people health and genes mutations on the ground
Satellite operations: a big damages of satellites electronics and
computers memory, damage to solar panels, loosing of several
satellites
Other systems: complete blackout of HF communications through
polar regions, some position errors make navigation operations
very difficult.
Extreme
Biological: unavoidable high radiation hazard to astronauts on
EVA (extra-vehicular activity); high radiation exposure to
passengers and crew in commercial jets at high latitudes
(approximately 100 chest X-rays) is possible.
Fluence
Frequ30MeV
ency
protons
1011
One in
few
thousan
d years
1010
One in
few
hundred
years
10
9
One in
20-50
years
Satellite operations: satellites may be rendered useless, memory
impacts can cause loss of control, may cause serious noise in
image data, star-trackers may be unable to locate sources;
permanent damage to solar panels possible.
Other systems: complete blackout of HF (high frequency)
communications possible through the polar regions, and position
errors make navigation operations extremely difficult.
S4
S3
Severe
Biological: unavoidable radiation hazard to astronauts on EVA;
elevated radiation exposure to passengers and crew in commercial
jets at high latitudes (approximately 10 chest X-rays) is possible.
Satellite operations: may experience memory device problems
and noise on imaging systems; star-tracker problems may cause
orientation problems, and solar panel efficiency can be degraded.
Other systems: blackout of HF radio communications through the
polar regions and increased navigation errors over several days are
likely.
108
One in 34 years
Strong
Biological: radiation hazard avoidance recommended for
astronauts on EVA; passengers and crew in commercial jets at high
latitudes may receive low-level radiation exposure (approximately
1 chest X-ray).
Satellite operations: single-event upsets, noise in imaging
systems, and slight reduction of efficiency in solar panel are likely.
Other systems: degraded HF radio propagation through the polar
regions and navigation position errors likely.
107
One per
year
Let us note, that the expected frequency of FEP events in the last column of Table 1 is averaged over solar cycle.
Really, this frequency is much higher in periods of high solar activity than in periods of low solar activity (Dorman
et al., 1993; Dorman and Pustil’nik, 1995, 1999).
SHORT DESCRIPTION OF THE METHOD OF AUTOMATICALLY SEARCH OF THE START OF
GROUND FEP EVENTS
Let us consider the problem of automatically searching for the start of ground FEP events. Of course, the
patrol of the Sun and forecast of great solar flares are very important, but not enough: only very small part of great
solar flares produce dangerous FEP events. In principal this exact forecast can be made by using high-energy
particles (few GeV/nucleon and higher) whose transportation from the Sun is characterized by much bigger
diffusion coefficient than for small and middle energy particles. Therefore high-energy particles arrive from the
Sun much earlier (8-20 minutes after acceleration and escaping into solar wind) than the lower energy particles
that cause a main dangerous situation in space and in atmosphere (see Table 1). The flux of high-energy particles
is very small and cannot be dangerous for people and electronics. The problem is that this very small flux is very
difficult to measure with enough accuracy on satellites to use for forecast (it needs very large effective surfaces of
detectors and very large weight). From other side, high-energy particles of galactic or solar origin are measured
continuously by ground-based neutron monitors, ionization chambers and muon telescopes with very large
effective surface areas (many square meters) that provide very small statistical errors. It was shown on the basis of
data in periods of great historical FEP events (as the greatest of February 23, 1956 and many tens of others), that
one-minute on-line data of high energy particles could be used for forecasting of incoming dangerous fluxes of
particles with much smaller energy. The method of coupling (response) functions (Dorman, 1957; Dorman et al.,
2000; Clem and Dorman, 2000) allows us to calculate the expected flux above the atmosphere, and out of the
Earth's magnetosphere from ground based data. Let us describe the principles and on-line operation of programs
"FEP-Search-1 min", "FEP-Search-2 min", and so on, developed and checked in the Emilio Segre' Observatory of
ICRC. The determination of increasing flux is made by comparison with intensity averaged from 120 to 61 minutes
before the present Z-th one-minute data. For each Z minute data, starts the program "FEP-Search-1 min". The
program for each Z-th minute determines the values
k  Z  60


D A1Z  ln I AZ  
 ln I Ak  60  1 ,
k  Z 120


(1)
k  Z  60


DB1Z  ln I BZ  
 ln I Bk  60  1 ,
k  Z 120


(2)
where I Ak and I Bk are one-minute total intensities in the sections of neutron super- monitor A and B.
If simultaneously
D A1Z  2.5, D B1Z  2.5,
(3)
the program "FEP-Search-1 min" repeat the calculation for the next Z+1-th minute and if Eq. 3 is satisfied again,
the onset of great FEP is determined and programs "FEP-Collect" and "FEP-Research" described in the next
section, are started.
If Eq. 3 is not satisfied, the program "FEP-Search-2 min" searches for the start of an increase by using twomin data characterized by  2   1 2 . In this case, the program "FEP-Search-2 min" will calculate values



2
k  Z  60

 ln I Ak  60
k  Z 120

2 ,
(4)



2
k  Z  60

 ln I Bk  60
k  Z 120

2 ,
(5)

D A2 Z   ln I AZ   ln I A, Z 1


DB 2 Z   ln I BZ   ln I B, Z 1

If the result is negative (no simultaneous increase in both channels of total intensity ≥ 2.5  2 , i.e. the
condition D
≥ 2.5, D
≥ 2.5 fails), then "FEP-Search-3 min" uses the average of three minutes Z-2, Z-1 and
A2Z
Z with  3  1
B2Z
3 . If this program also gives a negative result, then the program "FEP-Search-5 min" uses the
average of five minutes Z-4, Z-3, Z-2, Z-1 and Z with  5  1 5 . If this program also gives negative result, i.e.
all programs "FEP-Search-K min"(where K=1,2,3,5) give negative result for the Z-th minute, it means that in the
next 30-60 minutes there will be no radiation hazard (this information can be also very useful). After obtaining this
negative result, the procedure repeats for the next, Z+1-th minute, and so on. If any positive result is obtained for
some Z-minute, the "FEP-Search" programs checked the next Z+1- minute data. If the obtained result is again
positive, then the program "FEP-Spectrum" determined the FEP spectrum (see Paper 2, Dorman et al., 2003a).
Let us note that the similar program will be work in the case of three sections A, B and C: instead of Eq. 3 the
necessary condition for determining of great FEP starting will be
D A1Z  2.5 , D B1Z  2.5 , DC1Z  2.5
(6)
THE PROBABILITY OF FALSE ALARMS
The Case of Two Independent Channels
Because the probability function 2.5  0.9876 , that the probability P1 of an accidental increase with
amplitude more than 2.5 in one channel will be
P1 = 1  2.5 2  0.0062 min 1 ,
(7)
that means one in 161.3 minutes (in one day we expect 8.93 accidental increases in one channel). The probability
P2 of accidental increases simultaneously in both channels will be
P2 = 1  2.5 22  3.845  10 5 min 1
(8)
that means one in 26007 minutes  18 days. The probability P22 that the increases of 2.5 will be accidental in
both channels in two successive minutes is equal to
P22 = 1  2.5 24  1.478 10 9 min 1
(9)
that means one in 6.76  108 minutes 1286 years. If this false alarm (one in about 1300 years) is sent, it is not
dangerous, because the first alarm is preliminary and can be cancelled if in the third successive minute is no
increase in both channels bigger than 2.5 (it is not excluded that in the third minute there will be also an
accidental increase, but the probability of this false alarm is negligible :
1  2.5 26  5.685 1014 min 1
(10)
that means one in 3.34  10 7 years. Let us note that the false alarm can be sent also in the case of solar neutron
event (which really is not dangerous for electronics in spacecrafts or for astronauts health), but this event usually is
very short (only few minutes) and this alarm will be automatically canceled in the successive minute after the end
of solar neutron event.
The Case of Three Independent Channels
In this case the probability P3 of accidental increases simultaneously in three channels will be
P3 = 1  2.5 23  2.384  10  7 min 1
(11)
that means one in about 8 years. The probability P33 that the increases of 2.5 will be accidental in three channels
in two successive minutes is equal to
P33 = 1  2.5 26  5.685 10 14 min 1
(12)
7
that means one in 3.34  10 years. Therefore in the case of three independent channels we don’t need to wait of
the result fore the third minute. For the not dangerous very short (only few minutes) solar neutron event, the alarm
will be automatically canceled in the successive minute after the end of solar neutron event (the same situation as
in the case of two independent channels).
THE PROBABILITY OF MISSED TRIGGERS
The Case of Two Independent Channels
The probability of missed triggers depends very strong from the amplitude of increase. Let us suppose for
example that we have a real increase with amplitude A = k. The trigger will be missed if in any of both channels
and in any of both successive minutes the increase of intensity will be less than 2.5 (as a result of negative
statistical fluctuations). For realization of this case, the statistical fluctuation in one of channels must be negative
with amplitude more than (k-2.5). The probability P1  k  2.5  of this negative fluctuation in one channel in
one minute is equal
P1  k  2.5  = 1  k  2.5 2 ,
(13)
and the probability Pmt 2  A  k  of missed trigger for two successive minutes of observation simultaneously in
two channels is 4 times larger:
Pmt 2  A  k  = 2  1  k  2.5 .
(14)
For example, if k = 7 (that for ESO corresponds to an increase of about 9.8 %), the probability Pmt 2  A  7  of
missed trigger will be
Pmt 2  A  7  = 1.36 10 5 .
(15)
It means that missed trigger is expected only one per about 70000 events. In Table 2 are listed probabilities Pmt 2
of missed triggers as a function of the amplitude of increase A (in ).
Table 2. Probabilities Pmt 2 and Pmt 3 of missed triggers as a function of amplitude of increase A (in )
Channels
Two
A, 
Pmt 2
4.5
5.0
5.5
6.0
6.5
7.0
7.5
9.110 2
2.5 102
5.4 103
9.3 104
1.3 10 4
1.4 10 5
1.110 6
Three
Pmt 3
0.136
3.8 102
8.110 3
1.4 10 3
2.0 104
2.1105
1.6 10 6
The Case of Three Independent Channels
In this case instead of Eq. 14 will be (see the last row in the Table 2):
Pmt 3  A  k  = 3  1  k  2.5 .
(16)
DISCUSSION AND CONCLUSION
1. We assumed extended NOAA Space Weather Scale for Solar Radiation Storms (see Table 1).
2. Considered method of automatically searching for the onset of great, dangerous FEP on the basis of one-minute
NM data practically does not give false alarms (even for two channels the false preliminary alarm is expected
only one in about 1300 years, and for false final alarm one in 3.34 107 years).
3. None dangerous solar neutron events also can be separated automatically.
4. We estimated the probability of missed triggers; it was shown that for events with amplitude of increase A  7
the probability of a missed trigger for two and three independent channels is smaller than 1.4 10 5 and
2.1105 , correspondingly (this probability decreases sufficiently with increasing of amplitude A, see Table
2).
5. Historical ground FEP events show very fast increase of amplitude at the start of event (for example, in great
FEP event of February 23, 1956 amplitudes of increase in the Chicago NM were at 3.51 UT – 1%, at 3.52 UT
– 35%, at 3.53 UT – 180%; in this case, the missed trigger can be only for the first minute at 3.51 UT).
6. The described method can be used in many Cosmic Ray Observatories where one-minute data are detected.
ACKNOWLEDGEMENTS
Our great thanks to Prof. Yuval Ne’eman for constant interest and support of the work of Israel Cosmic Ray
Center and Emilio Segre’ Observatory, and J. Allen, A.V. Belov, E.A. Eroshenko, N. Iucci, K.G. McCracken, M.
Murat, M. Parisi, M.A. Shea, D.F. Smart, M. Storini, G. Villoresi, and V.G. Yanke for useful discussions.
The work of NM on Mt. Hermon is supported by Tel Aviv University, UNIRoma-Tre, and IFSI-CNR
Collaboration. This research is partly supported by EU INTAS Grant 00810.
REFERENCES
Carmichael, H., Space Sci. Rev., 1, 28, 1962.
Clem J.M. and L.I. Dorman “Neutron monitor response functions”, Space Science Rev., 93, 335-360, 2000.
Dorman L.I., Cosmic Ray Variations. Gostekhteorizdat, Moscow, 1957.
Dorman I.V., L.I. Dorman, and D. Venkatesan, Proc. 23-th ICRC, Calgary, 4, 79-82, 1993.
Dorman L.I. and L.I. Miroshnichenko, Solar Cosmic Rays. FIZMATGIZ, Moscow, 1968.
Dorman L.I. and L.A. Pustil'nik, Proc. 24-th ICRC, Rome, 4, 86-89, 1995.
Dorman L.I. and L.A. Pustil’nik, Proc. 26-th ICRC, Salt Lake City, 6, 407-410, 1999.
Dorman L.I. and D. Venkatesan “Solar cosmic rays”, Space Sci. Rev., 64, 183-362, 1993.
Dorman, L.I., G. Villoresi, N. Iucci, M. Parisi, and N.G. Ptitsyna, J. Geophys. Res., 105, 21047, 2000.
Duggal, S.P., Rev. Geophys. Space Phys., 17, 1021, 1979.
McCracken K.G., D.F. Smart, M.A. Shea, and G.A.M. Dreschhoff, Proc. 27-th ICRC, Hamburg, 8, 3209, 2001.
Stoker P.H. “Relativistic solar cosmic rays”, Space Sci. Rev., 73, 327, 1995.
E-mail address of Lev I. Dorman: lid@physics.technion.ac.il, lid1@ccsg.tau.ac.il
Manuscript received
; revised
; accepted