solar proton events recorded in the stratosphere during cosmic ray balloon observations in...

Available online at www.sciencedirect.com

www.elsevier.com/locate/asr

Advances in Space Research 45 (2010) 603–613

Solar proton events recorded in the stratosphere during cosmicray balloon observations in 1957–2008

G.A. Bazilevskaya *, V.S. Makhmutov, Y.I. Stozhkov, A.K. Svirzhevskaya, N.S. Svirzhevsky

Lebedev Physical Institute of Russian Academy of Sciences, 53, Leninsky Prospect, 119991 Moscow, Russia

Received 5 December 2008; received in revised form 11 November 2009; accepted 11 November 2009

Abstract

Long-term balloon observations have been performed by the Lebedev Physical Institute since 1957 up to the present time. The obser-vations are taken several times a week at the polar and mid latitudes and allow us to study dynamics of galactic and solar cosmic ray aswell as secondary particle fluxes in the atmosphere and in the near-Earth space. Solar energetic particles (120) – mostly protons – (SEP)events with >100 MeV proton intensity above 1 cm�2 s�1 s�1 were recorded during 1958–2006. Before the advent of the SEP monitoringon spacecraft these results constituted the only homogeneous series of >100 MeV SEP events. The SEP intensities and energy spectrainferred from the Lebedev Physical Institute observations are consistent with the results taken in the adjacent energy intervals by thespacecraft and neutron monitors. Joint consideration of the SEP events series recorded by balloons and by neutron monitors during solarcycles 20–23 makes it possible to restore the probable number of events in solar cycle 19, which was not properly covered by observa-tions. Some correlation was found between duration of SEP event production in a solar cycle and sunspot cycle characteristics.� 2009 COSPAR. Published by Elsevier Ltd. All rights reserved.

Keywords: Solar energetic particle events; Balloon cosmic ray observation; Sunspot solar cycle

1. Introduction

Not much was known about solar energetic particles –mostly protons – (SEP) intrusions into the Earth’s atmo-sphere until the late 1950s. The rare cases of the SEPrecordings by the ground-based installations (Duggal,1979) could not give information about the rate of SEPcoming into the near-Earth space, their morphology andinfluence on the Earth’s magnetosphere and atmosphere.

Every day measurements of charged particle fluxes inthe atmosphere of the northern polar and mid latitudeswere initiated in the USSR during the International Geo-physical Year (July 1957) by Vernov and Charakhchyan(Charakhch’ian and Charakhch’ian, 1959; Charakhch’yan,1964). In 1963 the observations were started in Antarctica.This monitoring although with reduced balloon launchings(several times per week) continues nowadays. The hosting

0273-1177/$36.00 � 2009 COSPAR. Published by Elsevier Ltd. All rights rese

doi:10.1016/j.asr.2009.11.009

* Corresponding author. Tel.: +7 495 4854263; fax: +7 495 4086102.E-mail address: [email protected] (G.A. Bazilevskaya).

institution is the Lebedev Physical Institute of RussianAcademy of Sciences (RAS); the contribution of Arcticand Antarctic Research Institute (Russian Federal Servicefor Hydrometeorology and Environmental Monitoring)and Polar Geophysical Institute (RAS) are acknowledgedamong others.

The results of balloon observations have been used forthe study of galactic cosmic ray modulation and SEP gen-eration and propagation (Stozhkov et al., 2009). The firstSEP event recorded in the stratosphere was that of 8 July1958 (Rymko et al., 1959). Recently, the role of cosmic raysin atmospheric processes has attracted new attention (Sto-zhkov, 2003; Bazilevskaya et al., 2008).

2. Cosmic ray observations in the atmosphere

Observations are performed by the meteorological bal-loons carrying a radiosonde with a sensor of two Geigertubes arranged as a telescope with 2 g cm�2 aluminuminterlayer. The radiosonde transmitter returns to a

rved.

http://dx.doi.org/10.1016/j.asr.2009.11.009

mailto:[email protected]

604 G.A. Bazilevskaya et al. / Advances in Space Research 45 (2010) 603–613

ground-based station records of an upper single counter(threshold energy is Ee = 0.2 MeV for electrons andEp = 5 MeV for protons), a telescope (Ee = 5 MeV,Ep = 30 MeV), and information on the atmospheric depth(residual atmospheric pressure, i.e. the air mass above theradiosonde), which can be converted to altitude using thestandard atmosphere. The radiosonde also records muonswith energy above several MeV; a single counter is sensitiveto c-rays (Ec P 20 keV) but with an efficiency lower than1% (whereas the efficiency of charged particle recording is�100%). Homogeneity of the data is maintained by theuse of standard detectors (kept unchanged during thewhole period of measurement) and careful calibration,using the control group of counters produced in varioustime periods during the long-term observations. Thedetailed description of the radiosonde and data acquisitionis given in Bazilevskaya and Svirzhevskaya (1998).

The permanent ground-based stations of balloonlaunching where SEPs are observed are situated in Mur-mansk region (geographical coordinates 67�570 N, 33�030

E before 2002 and 67�330 N, 33�200 E after 2002, geomag-netic cutoff rigidity Rc = 0.6 GV (Smart and Shea, 2008)),

Murmansk reg., Sept. 2001

0

2000

4000

6000

8000

1 10 100 1000

Atmospheric depth, g/cm2

Co

un

t ra

tes

per

min

0

Mirny, Sept. 2001

0

2000

4000

6000

8000

1 10 100 1000Atmospheric depth, g/cm2

Co

un

t ra

tes

per

min

0

a

c d

Fig. 1. Count rates of single counters (light symbols) and telescopes (dark sybackground count rates: (a) SEP intrusion, Murmansk region, 26.09.2001, 06201034 UT; (c) SEP intrusion and simultaneous electron precipitation, Mirny, 26test, Murmansk region, 05.10.1962, 0711–0924 UT and 1557–1809 UT.

Mirny, Antarctica, (66�340 S, 92�550 E, Rc = 0.03 GV),and, in very rare occasions, Moscow region (55�560 N,37�310 E, Rc = 2.35 GV).

Generally, galactic cosmic rays (GCRs) incident on theEarth’s atmosphere develop the nucleon-electromagneticcascades so that the particle flux dependence on the atmo-spheric depth X (the so-called transition curve) has a Pfot-zer maximum around 20–90 g cm�2 (depending on the sitelatitude and the solar cycle phase). The transition curve isdistorted during some perturbations in the near-Earthspace or due to radioactive pollution of the atmosphere.Intrusion of SEPs into the Earth’s atmosphere changesthe transition curve as it is plotted in Fig. 1a: both a singleGeiger counter and a telescope respond to the SEPs start-ing at the same altitude. The count rate increases withthe altitude of the radiosonde. At the polar latitudes thedistortion of the transition curve similar to that due toSEPs may be caused by precipitation of magnetosphericelectrons into the Earth atmosphere. These electrons areabsorbed at the height of � 50 km but they generate X-rayspenetrating down to � 20 km. The X-rays cause enhance-ment in the count rate of a single counter but not of a tele-

Murmansk reg., Sept. 2001

0

2000

4000

6000

8000

1 10 100 1000


Co

un

t ra

tes

per

min

0

Murmansk reg., Oct. 1962

0

2000

4000

6000

8000

1 10 100 1000Atmospheric depth, g/cm2

Co

un

t ra

tes

per

min

0

b

mbols, multiplied by 5) versus atmospheric depth. Solid curves show the–0820 UT; (b) electron precipitation, Murmansk region, 12.09.2001, 0704–.09.2001, 0710–0820 UT; (d) radioactive pollution due to nuclear weapon

G.A. Bazilevskaya et al. / Advances in Space Research 45 (2010) 603–613 605

scope thereby allowing to distinguish between the SEP andmagnetospheric events (Fig. 1b). It should be mentionedthat SEP and magnetospheric electron precipitation intothe atmosphere might occur simultaneously. As shown inFig. 1c in this case the count rates of both a single counterand a telescope increase toward the atmosphere boundary,however starting at different altitudes (Bazilevskaya et al.,2003). In the early 1960s, the nuclear tests in the atmo-sphere created the radioactive clouds, which occasionallywere recorded by the Lebedev Physical Institute radioson-des. This occurred at various latitudes, and the count rateenhancement was localized in the certain limits of altitudes.In these cases only a single counter responded. The minorgrowth in a telescope count rate was due to the accidentalcoincidences (Fig. 1d).

Subtraction of a background fluxes as measured on theeve of a SEP event gives the fluxes due to the SEP intrusionat different atmospheric depth. Majority of SEP causingeffects in the atmosphere are protons with E = �100–1000 MeV. Protons of lower energy are absorbed in theupper layers of the atmosphere; for instance, protons withE = 75 MeV do not penetrate deeper than 35 km(X > 5.2 g cm�2). The fluxes of SEPs with E > 1000 MeV/nucleon, which can generate the nucleon-electromagneticcascades, are relatively small due to the falling energy spec-trum. The 100–500 MeV protons lose their energy mainlythrough ionization. The proton path in the air r dependson the energy E as

EðrÞ ¼ arb;

where E is in MeV, a = 29.4, b = 0.57 for r 6 95 g cm�2

and a = 20.6, b = 0.65 for r > 95 g cm�2. In the energyinterval of �100–500 MeV, the energy spectrum of solarprotons can always be fitted by the power-law

dJ=dE ¼ AE�c:

Proton intensity in vertical direction (h = 0) at the atmo-spheric depth X is:

JðX ; 0Þ ¼ AZ 1

EðX ÞE�c dE ¼ BX�m;

where B = �Aa�c+1/(�c + 1), m = b(c � 1). Intensity ofprotons coming from the direction h at the depth X is:

JðX ; hÞ ¼ BX

cos h

� ��m

:

Omni-directional flux of protons is:

IðX Þ ¼ 2pBX�m

Z p=2

0

ðcos hÞm sin hdh ¼ 2pBX�m=ðmþ 1Þ:

Thus, an isotropic flux of solar protons incident on theatmosphere with a power-law energy spectrum in the inter-val where the ionization losses dominate absorbs in theatmosphere also according to the power-law. Intensity ofprotons J(X, 0) as determined from observations at thedepth X is equal to the proton intensity with energy

E > aXb at the atmospheric boundary. It can be estimatedfrom the data of a Geiger counter as

Jð� EðX Þ; 0Þ ¼ IðX Þð1þ mÞ=2p

¼ DN cnðX Þð1þ mÞ=ð2pGcÞ;

where Gc is a counter geometrical factor, DNcn is differencebetween its count rates at the depth X during a SEP eventand in the absence of SEPs. The intensity is found from thetelescope data as

Jð� EðX Þ; 0Þ ¼ DNtðX Þ=GtðmÞ;

where Gt(m) is the geometrical factor of a telescope, DNt istelescope count rates due to SEPs.

In the case when SEPs are mostly protons with isotropicangular distribution on the atmospheric boundary theenergy spectra inferred from the data of a single counterand a telescope coincide. The correction of J(X, 0) fornuclear interaction of protons is done by taking intoaccount the input of secondary cascade protons, whichcan be recorded by a radiosonde. Depending on the slopeof the energy spectrum, the correction is between 1.6 and2.1 for the 500 MeV protons.

With the aim to check validity of retrieving of SEPfluxes from the balloon observations the transport of solarprotons through the atmosphere was simulated using thePLANETOCOSMICS code based on GEANT4 (Mak-hmutov et al., 2007). Starting with a given energy spectrumof solar protons on the atmospheric boundary and takinginto account ionization, elastic and inelastic nuclear inter-action, multiple scattering, bremsstrahlung, pair produc-tion, Compton scattering, photoelectron effect, and thedecay of particles the response of a radiosonde to theSEP intrusion at different altitudes was calculated. Theresults of simulation are in excellent agreement with theresults of observations as illustrated in Fig. 2.

Since the energy spectrum of SEPs is rather steep, theeffect in the stratosphere is usually observed only at polarlatitudes – in the Arctic and Antarctic. In very rare occa-sions when the radiosonde appears to be at high altitudeduring the initial phase of a SEP event the transition curveat mid latitudes also changes but still has a maximumbecause the main contribution comes from the severalGeV protons which generate hadron-electromagnetic cas-cades. For such events the solar proton intensity withenergy more than geomagnetic cutoff may be determined.The SEP effect at the mid latitudes does not last longerthan several hours. If, during a SEP event, balloon mea-surements were made at the same mid latitude location asa neutron monitor, the SEP event would be recorded fora longer time by the neutron monitor than by the balloonsensor because the neutron monitor has a larger geometri-cal factor.

Reliable identification of a SEP event by a radiosondecan be done when at the height above �25 km the countrate of a counter exceeds the background by �300–500 min�1, and the count rate of a telescope, by �50–

Murmansk region, 12 October 1981, 1211-1329 UT

0.01

0.1

1

10

100

1 10 100 1000


Par

ticle

flux

, cm

-2s-1

p

p+e+/- +0.01*gamma+mu+/-

observation

Fig. 2. Comparison between the observed particle fluxes and results ofsimulation with GEANT 4 for the SEP event of 12 October 1981.Background due to galactic cosmic rays is removed. Ionization losses andnuclear interactions of solar protons are taken into account in the resultsof observations.


100 min�1. Consideration of SEP events recorded by bal-loons in the stratosphere and matching with the spacecraftresults shows that majority (not all) of such events arethose with intensity J (P 100 MeV) P 1 cm�2 s�1 sr�1 inthe maximum of the intensity–time profile. The accuracyof the intensity determination is about 30%. The accuracyof the spectrum power-law index is �0.5.

It should be stressed that the solar proton intensity asdetermined from the balloon measurements is usually con-sistent with simultaneous spacecraft results in the adjacentlow-energy range, although there is systematic discrepancybetween the data of balloon measurements and the>700 MeV energy channel of the GOES satellites. On theother hand, the balloon data are in reasonable agreementwith the intensity at energy above 500 MeV inferred fromthe neutron monitor measurements (Bazilevskaya and Mak-hmutov, 1983). Until now, the balloon observations are usedfor confirmation of the SEP flux values deduced from theneutron monitor data during GLEs (Vashenyuk et al., 2008).

The instantaneous energy spectrum at the top of theatmosphere may be derived from the data of each flightof a radiosonde during a SEP event. Frequent launchingsof radiosondes during an event enable us to estimate theintensity–time profile of solar protons (Fig. 3, upper panel).However, there are inevitable gaps in the time profiles(solar protons are detected only while a radiosonde is situ-ated at rather high altitudes). In particular, the beginningof an event is usually not observed. On the other hand,the time profiles of ionizing particle fluxes at different levelsof the atmosphere can be derived from balloon observa-

tions as depicted in Fig. 3 (lower panel). It is seen that inthe beginning of event of 20 January 2005 the solar protonspenetrated into the atmosphere over Murmansk as deep as9 km. However, the maximum intensity enhancement was�60% and lasted during less than �2 h. At altitudes above27 km the particle intensity increased by a factor of �60and was still above the GCR background 28 h after begin-ning of SEP intrusion.

3. Lebedev Physical Institute series of SEP events observed

on balloons

As stated above simultaneous records of a single counterand a telescope provide reliable detecting of SEP events bythe Lebedev Physical Institute radiosondes (which are fur-ther called just balloon events). Unfortunately, till the endof 1960s the radiosondes carried single counters and tele-scopes alternately. Therefore some uncertainties were pres-ent in the SEP events identification because of possiblecontribution of electron precipitation. For separation ofSEP and precipitation events we used some peculiaritiesintrinsic in the electron precipitation (as observed in thestratosphere), but not in the SEP events, namely, ratherfast variability with the time scale of several minutes andoverall duration of less than several hours. Revision ofevents for 1958–1965 resulted in excluding some eventsrecorded by a single counter and not found in the next bal-loon flight several hours later. Also, the lists presented in(Shea and Smart, 1990) proved that events of 22.08.1958and 28.09.1961 were missed because the flight ceiling alti-tude of our balloons was low. With advent of the satelliteera it became possible to check the Lebedev Physical Insti-tute list against the spacecraft results. It was found that theevents 28.08.1966, 23.05.1967, 29.09.1968, 25.07.1989,25.06.1992, and 04.11.1997, which might be recorded, wereactually lost by our observations. These events are includedin the ongoing analysis and are given in Table 1. Table 1contains all the SEP events both recorded and missed bythe Lebedev Physical Institute observations. The time andsite of the first observation, the total number and sites ofthe balloon launchings during an event are indicated. Also,a list of doubtful events is given. In total, the SEP balloonseries contains 120 events (1958–2006) including 8 eventsmissed by the Lebedev Physical Institute observations.The balloon event occurrence together with sunspot num-ber Rz is shown in Fig. 4 (Stozhkov et al., 2009).

Thus, the homogeneity of the balloon SEP event dataseries is provided by the careful calibration of detectors,and the reliability of the data is confirmed by consistencywith results in the adjacent energy ranges and agreementwith the simulation results.

It is interesting to consider the rate of SEP events takinga cumulative number of events for each solar cycle. Thecumulative SEP event number, Nc, just summarizes SEPevents within a solar cycle from its beginning versus timeelapsed. Nc for the balloon events and GLEs in the cycles20–23 are plotted in Fig. 5 together with the sunspot

0.1

1

10

100

1000

10000

20 20.5 21 21.5

January 2005

Inte

nsi

ty, c

m-2

s-1sr

-1

bal., E>150 MeV

bal., Е> 300 MeV

G10,E>50 MeV

G10,11, E>100MeV

NM Apatity, arb.units

G10, E>10 MeV

G10, E>30 MeV

1

10

100

1000

10000

20 20.5 21 21.5January 2005

Flu

x, c

m-2

s-1

1

10

100

1000X=17.44 g/cm2 (27.37 km)

X=30 g/cm2 (23.9 km)

X=50 g/cm2 (20.7 km)

X=100 g/cm2 (16.29 km)

X=200 g/cm2 (11.9 km)

X=300 g/cm2 (9.28 km)

X=8.57 g/cm2 (31.9 km)G11

Fig. 3. Upper panel: intensity time profile of solar protons during event of 20 January 2005. The data of GOES 10 and 11 (http://spidr.ngdc.noaa.gov/spidr, balloons and the Apatity neutron monitor (http://pgia.ru/CosmicRay/) are presented. Lower panel: fluxes of charged particles in the atmospheredue to galactic and solar cosmic rays as measured by a single counter of a radiosonde at various levels of the atmosphere X (g/cm2).


number (ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/SUNSPOT_NUMBERS/). Both balloon events and GLEsare presented. Cycle 19 is not presented in Fig. 5 becausethe balloon observations started in July 1957 (the 40thmonth of the cycle) and did not cover the whole solar cycle.Duration of time of SEP event production inside each solarcycle may be called a SEP event production cycle. Table 2summarizes the temporal characteristics of the SEP eventproduction cycles 19–23 and the total number of SEPevents recorded (Nct). The following temporal characteris-tics are presented: the time from the beginning of a solarcycle till the first SEP event recording (Tb); duration ofSEP events occurrence inside a solar cycle (Tin); the timefrom the last SEP event recording and the end of a solarcycle (Ta). The dates and durations of solar cycles are alsogiven. Shea and Smart (2008) compared the total numbersof the ground level enhancements (GLEs) for the cycles 19–23 and did not find much difference between them. Thesame is true for the balloon SEP events. On the other hand,the temporal features of the SEP event production arerather variable, as it is seen from Table 2.

4. SEP event production in the solar cycle 19

The solar cycle 19 was not fully covered by the balloonobservations but some neutron monitors were operatingsince early 1950s (Shea and Smart, 1990). The first GLEwas recorded in the month 23 from the cycle beginning, i.e.similar to the other cycles (see Table 2). However, the GLEswere absent during 34 months from 31 August 1956 till 16July 1959. In total 10 GLEs were found in 1956–1964, soNct for the GLEs in the cycle 19 seems to be rather low.The number of operating neutron monitors was growingrapidly in the beginning of the 19th cycle: 2, 3, 4, and 5 sta-tions worked in 1953, 1954, 1955, and 1956, while 42 stationswere operating in 1957 (ttp://www.env.sci.ibaraki.ac.jp/database/html/WDCCR/data_e.html), among them severalpolar stations (Churchill, College, Resolute Bay, Thule).However, time resolution was not high enough (Shea andSmart, 1990) leading to possible gaps in the GLE list. Ourattempt to restore the actual number of GLEs in the cycle19 bases on (1) cumulative number of SEP events versus timeelapsed from the solar cycle beginning and (2) regression

Table 1List of balloon SEP events in the solar cycles 19–23 (see text). The observations in the Lebedev Physical Institute started in July 1957. “Mu” stands for theMurmansk region, “Mi”, for the Mirny observatory (Antarctica), “Mo” for the Moscow region. “d” means the day of a month.

# Day Month Year Site of the firstobservation

Time of the firstobservation, UT

Whole time ofobservations inthe stratosphere

Number ofballoon launches

Sites ofobservations

1 8 7 1958 Mu 0757–0922 8d 1 Mu2 22 8 1958 Missed3 26 8 1958 Mu 0808–0831 26d 1 Mu4 10 5 1959 Mu 11d0809–0850 11–13d 4 Mu5 9 7 1959 Mu 0853–0901 9–13d 7 Mu6 14 7 1959 Mu 0830–0901 14–16d 4 Mu7 16 7 1959 Mu 17d0732–0758 17–21d 6 Mu8 1 4 1960 Mu 1004–1041 1d 2 Mu9 28 4 1960 Mu 0807–0838 28–29d 4 Mu10 4 5 1960 Mu 1100–1134 4–5d 4 Mu11 13 5 1960 Mu 0814–0858 13d 2 Mu12 3 9 1960 Mu 0814–0859 3–5d 11 Mu13 12 11 1960 Mu 14d0801–0830 14d 4 Mu14 15 11 1960 Mu 0729–0828 15–19d 11 Mu15 20 11 1960 Mu 21d1307–1336 21d 1 Mu16 12 7 1961 Mu 2019–2041 12–13d 2 Mu17 18 7 1961 Mu 19d0756–0820 19–20d 6 Mu18 20 7 1961 Mu 1645–1721 20–21d 3 Mu19 28 9 1961 Missed20 21 9 1963 Mu 0704–0758 21d 2 Mu, Mi21 26 9 1963 Mi 0742–0807 26d 1 Mi22 7 7 1966 Mi 0758–0842 7d 1 Mi23 28 8 1966 Missed24 2 9 1966 Mu 0753–0844 2d 6 Mu, Mi25–26* 28 1 1967 Mu 1236–1326 28–31d 9 Mu, Mi27 23 5 1967 Missed28 28 5 1967 Mu 0856–0937 28–29d 3 Mu29 9 6 1968 Mu 10d0834–0911 10d 2 Mu30 29 9 1968 Missed31 18 11 1968 Mu 1238–1311 18d 2 Mu32 25 2 1969 Mu 1250–1331 25d 2 Mu33 26 2 1969 Mu 0656–0722 26–28d 8 Mu, Mi34 30 3 1969 Mu 0756–0833 30–04d 27 Mu, Mi35 11 4 1969 Mu 12d0819–0827 12–15d 16 Mu, Mi36 2 11 1969 Mu 2300 02–03d 4 Mu, Mi37 29 3 1970 Mu 0539–0619 29–31d 20 Mu, Mi38 24 1 1971 Mi 25d0810 25–27d 21 Mu, Mi39 1 9 1971 Mi 02d0820 02–04d 17 Mu, Mi40 4 8 1972 Mi 0758–0832 4–7d 30 Mu, Mi41 7 8 1972 Mi 8d0404–0429 8d 8 Mu, Mi42 29 4 1973 Mi 30d0830 30–01d 5 Mu, Mi43 23 9 1974 Mu 24d0658–0958 24–26d 13 Mu44 30 4 1976 Mi 01d0820 01d 1 Mi45 19 9 1977 Mu 1101–1150 19–20d 16 Mu, Mi46 24 9 1977 Mu 0747–0839 24–25d 19 Mu, Mi47 22 11 1977 Mi 1620–1716 22–24d 7 Mi48 28 4 1978 Mu 29d0630–0714 29–30d 12 Mu, Mi49 7 5 1978 Mu 0613–0702 7–8d 18 Mu, Mi50 23 9 1978 Mu 1206–1335 23–25d 11 Mu, Mi51 18 8 1979 Mi 20d0355–0407 20–21d 7 Mu, Mi52 21 8 1979 Mi 0823–0853 21d 5 Mu, Mi53 14 9 1979 Mu 16d0703–0715 16–20d 16 Mu, Mi54 1 4 1981 Mu 0655–0709 1d 2 Mu55 4 4 1981 Mu 0728–0747 4d 6 Mu, Mi56 10 4 1981 Mu 1820–1856 10–11d 10 Mu, Mi57 30 4 1981 Mi 0859–0925 30–01d 8 Mu, Mi58 10 5 1981 Mi 0849–1058 10–11d 8 Mu, Mi59 7 10 1981 Mu 08d0031–0045 08–11d 39 Mu, Mi60 12 10 1981 Mu 0647–0706 12–15d 48 Mu, Mi61 12 7 1982 Mu 13d0757–0806 13–15d 4 Mu62 26 11 1982 Mu 0615–0724 26d 7 Mu, Mi63 8 12 1982 Mi 0730–0820 08–09d 10 Mu, Mi


Table 1 (continued)

# Day Month Year Site of the firstobservation

Time of the firstobservation, UT

Whole time ofobservations inthe stratosphere

Number ofballoon launches

Sites ofobservations

64 17 12 1982 Mu 18d0623–0705 18d 4 Mu65 16 2 1984 Mu 0709–0715 16–19d 5 Mu, Mi66 14 3 1984 Mu 0608–0703 14d 3 Mu, Mi67 25 4 1984 Mu 26d0627–0657 26–28d 11 Mu, Mi68 14 12 1988 Mu 15d0728–0842 15–16d 4 Mu69 16 12 1988 Mi 1059–1156 16–18d 12 Mu, Mi70 25 7 1989 Missed71 12 8 1989 Mu 13d0512–0605 13–15d 15 Mu, Mi72 16 8 1989 Mu 0641–0805 16d 6 Mu, Mi73 17 8 1989 Mu 0309–0435 17–19d 8 Mu, Mi74 29 9 1989 Mu 1521–1708 29–02d 28 Mu75 19 10 1989 Mu 1651–1741 19–22d 31 Mu, Mi76 22 10 1989 Mu 1814–1832 22–23d 12 Mu, Mi77 24 10 1989 Mu 1734–1843 24–27d 15 Mu, Mi78 15 11 1989 Mi 0726–0844 15d 4 Mu, Mi79 21 5 1990 Mi 22d0536–0632 22–23d 9 Mu, Mi80 24 5 1990 Mu 25d0601–0710 25–26d 9 Mu, Mi81 26 5 1990 Mi 27d0349–0425 27d 2 Mi82 28 5 1990 Mi 0744–0830 28d 2 Mi83 23 3 1991 Mi 1236–1248 23–24d 5 Mi84 4 6 1991 Mi 6d0745–0921 6–8d 11 Mu, Mi85 11 6 1991 Mu 0643–0744 11–12d 7 Mu, Mi86 15 6 1991 Mu 2226–2337 15–17d 4 Mu87 25 6 1992 Missed88 2 11 1992 Mu 0634–0734 2–3d 6 Mu, Mi89 4 11 1997 Missed90 6 11 1997 Mo 1317–1448 Mo91 20 4 1998 Mi 21d0635–0702 21–22d 3 Mi92 2 5 1998 Mi 3d0740–0804 3d 3 Mi93 6 5 1998 Mu 1119–1123 6d 1 Mu94 24 8 1998 Mi 25d0758–0916 25–31d 7 Mu, Mi95 14 11 1998 Mi 0709–0833 14–16d 6 Mu, Mi96 14 7 2000 Mu 1450–1650 14–15d 4 Mu97 12 9 2000 Mu 13d0619–0849 13d 3 Mu98 8 11 2000 Mi 9d0858–1006 9–10d 9 Mu, Mi99 2 4 2001 Mi 3d0845–1012 3–4d 4 Mu, Mi100 15 4 2001 Mi 16d0648–0750 16–17d 3 Mi101 18 4 2001 Mu 0636–0847 18–19d 6 Mu, Mi102 16 8 2001 Mi 0735–0823 16–18d 4 Mu, Mi103 24 9 2001 Mu 1801–1843 24d 1 Mu104 4 11 2001 Mi 5d0726–0823 5–9d 8 Mu, Mi105 22 11 2001 Mi 23d0747–0838 23d 3 Mu, Mi106 26 12 2001 Mu 0710–0845 26–28d 5 Mu, Mi107 21 4 2002 Mu 22d0842–1018 22d 2 Mu108 24 8 2002 Mi 0706–0807 24d 4 Mu, Mi109 26 10 2003 Mu 27d0946–0949 27d 1 Mu110 28 10 2003 Mu 1637–1740 28–29d 5 Mu111 29 10 2003 Mu 2251–2415 30–31d 2 Mu112 2 11 2003 Mu 3d0053–0132 3–4d 6 Mu, Mi113 4 11 2003 Mi 5d0558–0704 5d 1 Mi114 17 1 2005 Mi 18d0554–0606 18d 2 Mu, Mi115 20 1 2005 Mu 0732–0838 20–21d 11 Mu116 7 9 2005 Mi 8d0901–0951 8–9d 6 Mi117 5 12 2006 Mi 6d0922–0957 6d 1 Mi118 6 12 2006 Mu 8d0814–0852 8d 1 Mu119 13 12 2006 Mu 0824–0912 13–14d 7 Mu, Mi120 14 12 2006 Mi 16d0841–1150 16d 1 Mi

Doubtful events17 3 1958 Mu 1018–1049 17d 2 Mu3 10 1958 Mu 0815–0903 3d 1 Mu12 5 1960 Mu 0818–0827 12d 1 Mu9 5 1961 Mu 0811–087 5d 2 Mu

* #26 and 27 are two events occurred on 28.01.1968 according to the GLE list (Cliver et al., 1982).


1.E-02

1.E-01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1955 1965 1975 1985 1995 2005

Year

J(>1

00 M

eV),

1/(c

m2 s

sr)

0

50

100

150

200

Yea

rly

aver

aged

su

nsp

ot n

um

ber

Fig. 4. SEP events recorded in the stratosphere during solar cycles 19–23. Vertical bars denote the flux values corresponding to the maximum of intensity–time profiles. Gray curve presents the yearly averaged sunspot data taken from (ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/SUNSPOT_NUMBERS/).


between the cumulative numbers of balloon events andGLEs.

Upper panel of Fig. 6 presents the summary of cumulativeSEP event numbers for the solar cycles 19–23. While Nc forthe cycles 20–23 behave more or less alike, the cycle 19 looksdifferent than the others because of the late start of balloon

Cycle 20

0

5

10

15

20

25

30

35

0 50 100 150Month after solar cycle min

Cu

mu

lati

ve e

ven

t n

um

ber

0

50

100

150

200

250

0 50 100 150

Rz

Cu

mu

lati

ve e

ven

t

Cycle 22

0

5

10

15

20

25

30

35

0 50 100 150

Month after solar cycle min

Cu

mu

lati

ve e

ven

t n

um

ber

0

50

100

150

200

250

Rz

Cu

mu

lati

ve e

ven

t

Fig. 5. Sunspot monthly (gray line) and smoothed (black line) values alongsidevents and thin step-like curve for GLEs) versus time elapsed from the solar cySOLAR_DATA/SUNSPOT_NUMBERS/).

events and retarding growth of the GLE number. It shouldcertainly be expected that several SEP events occurred beforethe month 40 from the beginning of the cycle 19 when the bal-loon observations started. The corresponding numbers ofevents in the month 40 from the cycle start were 7, 9, 11,and 7 for the solar cycles 20–23. The average <Nc> was

Cycle 21

0

5

10

15

20

25

30

35

0 50 100 150Month after solar cycle min

nu

mb

er

0

50

100

150

200

250

0 50 100 150

Rz

GLE

BAL

Rz

Cycle 23

0

5

10

15

20

25

30

35

0 50 100 150


nu

mb

er

0

50

100

150

200

250

Rz

GLE

BAL

Rz

e with cumulative SEP event number Nc (thick step-like curve for ballooncle beginning. Sunspot data are taken from (ftp://ftp.ngdc.noaa.gov/STP/

Table 2Characteristics of the sunspot and SEP event production cycles: time from the beginning of a solar cycle till the first SEP event recording (Tb); duration ofSEP event occurrence inside a solar cycle (Tin); time from the last SEP event recording till the end of the solar cycle (Ta); total number of SEP eventsrecorded (Nct).

Rz Bal GLE

Cycle Start/end (YYYYMM) Solar cycle duration Tb Tin Ta Nct Tb Tin Ta Nct

Months Months

19 195404/196410 127 13 21(30)* 22 66 39 10(16)*

20 196411/197605 139 20 118 1 23 20 118 1 1321 197606/198609 124 15 80 29 23 15 78 31 1222 198610/199604 115 26 48 41 21 33 41 41 1523 199605/200812** 152 18 110 24 32 18 110 24 16

* The restored number of SEP events in the cycle 19 is given in parentheses (see text).** Assumed time of the 23rd solar cycle end, which is not known exactly at the moment of the paper writing.


8.5. It means that from 7 to 11 SEP events may actually bemissed in the cycle 19 by balloon observations and the totalNct in case of full coverage by observations would be about30 instead of 21 actually observed.

Fig. 7a shows regression between the Nc of balloonevents and GLEs, i.e., Nc for balloon events in a given

Balloon events (E>100 MeV)

0

5

10

15

20

25

30

35

0 50 100 150 200


Cu

mu

lati

ve e

ven

t n

um

ber

0

5

10

15

20

Cu

mu

lati

ve e

ven

t n

um

ber

Balloon events (E>100 MeV)

0

5

10

15

20

25

30

35

0 50 100 150 200Month after solar cycle min

Cu

mu

lati

ve e

ven

t n

um

ber

0

5

10

15

20

Cu

mu

lati

ve e

ven

t n

um

ber

Fig. 6. Upper panels: cumulative SEP event number for balloon events (left) anand thin black step-like curves denote cycles 20–23. Thick black step-like curvwith “restored cumulative SEP event number” for the cycle 19 (see text).

month of a solar cycle is plotted versus Nc for GLEs inthe same month. It is seen that regression does not changestrongly from one cycle to another during the cycles 20–23.In the cycle 19, the first event detected both by neutronmonitors and LPI balloons was that of 16.07.1959 (the64th month from the beginning of the cycle 19). It is

GLEs

0 50 100 150 200Month after solar cycle min

Cycle 19

Cycle 20

Cycle 21

Cycle 23

Cycle 22

GLEs

0 50 100 150 200


d for GLEs (right) versus time elapsed from the solar cycle beginning. Graye marks the cycle 19. Bottom panels: the same as on the upper panels but

Table 3Correlation coefficients between the characteristics of the sunspot and SEPevent production cycles. Bold numbers mark the coefficient exceeding twoerror values.

Rz_max Rz_dur Number ofcycles used

GLE_Nct �0.27 ± 0.54 0.36 ± 0.50 4bal_Nct �0.48 ± 0.45 0.84 ± 0.17 4GLE_Tb 0.19 ± 0.48 �0.57 ± 0.39 5bal_Tb 0.07 ± 0.57 �0.45 ± 0.46 4GLE_Tin �0.75 ± 0.22 0.89 ± 0.10 5bal_Tin �0.87 ± 0.14 0.89 ± 0.12 4GLE_Tin/Rz_dur �0.70 ± 0.25 0.79 ± 0.19 5Bal_Tin /Rz_dur �0.80 ± 0.21 0.77 ± 0.24 4GLE_Ta 0.81 ± 0.17 �0.62 ± 0.31 5bal_Ta 0.26 ± 0.47 �0.50 ± 0.37 5Rz_dur �0.66 ± 0.28 5

0

5

10

15

20

25

30

35

0 5 10 15 20NGLE

Nb

al

0

Cycle 19

Cycle 20

Cycle 21

Cycle 23

Cycle 22

0

5

10

15

20

25

30

35

0 5 10 15 20

NGLE

Nb

al

a

b

Fig. 7. (a) Regression between the cumulative numbers of SEP eventsobserved on balloons and GLEs for cycles 19–23. (b) Same as in (a) butcorrected for the partial balloon observation coverage of the cycle 19(dashed gray line) and with “cycle 19 restored” by adding of probablemissed SEP events to both balloon and GLE data.


remarkable that the regression slope for the cycle 19 since16.07.1959 is akin to that of other cycles. It proves thatrelation between the GLEs and balloon events after16.07.1959 was similar to that in cycles 20–23. However,addition of 9 balloon events presumably missed by balloonobservations, distorts the regression between balloonevents and GLEs as it is shown in Fig. 7b by the graydashed line. It is clear that Nc for GLEs has also to beincreased in order to adjust to the regression of the cycles20–23. The first simultaneous observation of a SEP eventby balloons and neutron monitors occurred in July 1959(month 64 from the cycle start). The averaged regressionbetween Nc of balloon events and GLEs in the month 64for the cycles 20–23 requires for GLEs Nc = 9 instead ofNc = 3 recorded. This does not contradict to the averageNc in the month 64 for the cycles 20–23 (<Nc> = 8.75)(see right upper panel of Fig. 6). Thus, 6 GLEs might beadded between August 1956 and July 1959. Then, the prob-able total number of GLEs in the cycle 19 was Nct = 16instead of Nct = 10 actually observed. The regressionbetween “corrected” numbers of balloon events and GLEsis also given in Fig. 7b by the thick black line with light tri-angles. It is seen that the solar cycle 19 may really be ratherabundant, but not extraordinary, in the SEP event produc-tion. However, this is only an estimation of the total num-ber of SEP events in the cycle 19.

5. Correlation between the cycles in sunspots and SEP events

It is interesting to consider the correlation between thecharacteristics of solar cycles and SEP event productioncycles. Table 3 gives the corresponding correlation coeffi-cients that have rather big errors since the number of cyclesunder consideration is small – 4 or 5 (where cycle 19 couldbe used). Bold numbers mark the coefficients exceedingtwo error values. The statements that may be derived fromthe Table 3 should be considered as preliminary because ofvery poor statistics.

Total number of SEP events in a solar cycle (Nct) is inde-pendent of either Rz maximum or duration of a solar cyclewith one exception for high correlation between the balloonNct and Rz duration, which may be accidental. There is noconnection of the time elapsing between the solar cyclebeginning with the start of the SEP event production (Tb).

Duration of SEP event production (Tin) is shorter in thesolar cycles with higher maximum Rz and longer in thelonger solar cycles. The same is true for the fraction of asolar cycle covered by SEP event production (i.e. Tin

divided by duration of the corresponding sunspot solarcycle). That means nonlinear connection between the solarcycle features and duration of the SEP production cycle.

The time elapsing between the end of the SEP event pro-duction and the end of the corresponding solar cycle (Ta)positively correlates with Rz maximum and negatively (ifany), with the solar cycle duration. There is an exceptionfor the Ta of balloon events and Rz maximum (insignifi-cant correlation), which is connected with the short Tafor balloon events in the cycle 19. Elimination of the cycle19 leads to increasing of the correlation coefficient up0.86 ± 0.15. There is an unusually large difference betweenthe Ta values for GLEs and balloon events in the cycle 19(see Table 2).

The opposite correlation of the SEP event productioncharacteristics with Rz maximum and the solar cycle dura-tion is a consequence of negative correlation between thetwo latter (see the last line of Table 3) and, in general, ofthe so-called Waldmeier relations connecting the amplitude


and the duration of different phases of a solar cycle (Uso-skin and Mursula, 2003). No underlying physical processesresponsible for the Waldmeier relations have been pro-posed as yet. However, the relationship between the sun-spot and SEP event production cycles, if confirmed, maybe used for the space weather prognosis.

6. Summary

The Lebedev Physical Institute series of SEP eventsobserved on balloons in the atmosphere refers to �100–500 MeV solar protons and covers 1958–2006. The homo-geneity of the data series is provided by the careful calibra-tion of detectors. The SEP intensities and energy spectrainferred from the Lebedev Physical Institute observationsare consistent with the results taken in the adjacent energyintervals by the spacecraft and neutron monitors. Reliabil-ity of the observational data is also confirmed by consis-tency with the results of GEANT4 simulation.

The number of SEP events as observed by balloons inthe cycles 20–23 was 23, 23, 21, and 32, respectively.Although the 19th solar cycle was not fully covered bythe Lebedev Physical Institute observation, comparisonwith the cycles 20–23 allowed us to derive the total balloonSEP event number as �30. Because of close correlationbetween the balloon SEP events and GLEs, this suggestedthat about six GLEs might be missed between August 1956and July 1959. In this case the total number of GLEs in thecycle 19 may be �16 instead of 10 actually observed.

The number of SEP events recorded in a solar cycle doesnot depend on the sunspot cycle amplitude and, probably,on the cycle length. The duration of SEP event productionnegatively correlates with the maximum sunspot numberand positively correlates with the sunspot cycle length.The same is true for the fraction of the solar cycle lengthcovered by the SEP production. The relationship betweenthe sunspot and SEP event production cycles, if confirmed,may be used for the space weather prognosis. However, thecorrelations are obtained with very poor statistics andshould be considered as preliminary.

Acknowledgements

This work is partly supported by the Russian Founda-tion for Basic Research (grants 08-02-08357-z, 08-02-00054, 08-02-91006, 09-02-10018, 07-02-01019) and theProgram of Presidium of RAS “Neutrino physics andastrophysics”. We thank the reviewers for the useful com-ments leading to significant improvement of the paper.G.A.B. thanks COSPAR for the financial support forattending the 37th COSPAR Scientific Assembly.

References

Bazilevskaya, G.A., Makhmutov, V.S. Derivation of absolute fluxes of>100 MeV solar protons from balloon and neutron monitor measure-ments. Geomagnet. i Aeronomiya 23 (3), 373–377 (In Russian), 1983.

Bazilevskaya, G.A., Svirzhevskaya, A.K. On the stratospheric measure-ments of cosmic rays. Space Sci. Rev. 85 (3/4), 431–521, 1998.

Bazilevskaya, G.A., Makhmutov, V.S., Svirzhevskaya, A.K., Svirzhevsky,N.S., Stozhkov, Y.I., Vashenyuk, E.V. Electron precipitation into theEarth’s polar atmosphere during solar proton events. Izv. RAN, Ser.Fiz. 67 (4), 489–491 (In Russian), 2003.

Bazilevskaya, G.A., Usoskin, I.G., Fluckiger, E.O., Harrison, R.G.,Desorgher, L., Butikofer, R., Krainev, M.B., Makhmutov, V.S.,Stozhkov, Y.I., Svirzhevskaya, A.K., Svirzhevsky, N.S., Kovaltsov,G.A. Cosmic ray induced ion production in the atmosphere. Space Sci.Rev. 137 (1), 149–173, 2008.

Charakhch’ian, A.N., Charakhch’ian, T.N. Measurement of cosmic-rayintensity in the stratosphere at various altitudes and latitudes. Sov.Phys. JETP 35 (8), 761–771, No. 5, 1959.

Charakhch’yan, A.N. Investigation of stratosphere cosmic ray intensityfluctuations induced by processes on the Sun. Sov. Phys. Uspekhi 7 (3),358–374, 1964.

Cliver, E.W., Kahler, S.W., Shea, M.A., Smart, D.F. Injection onsets of�2 GeV protons, �1 MeV electrons, and �100 keV electrons in solarcosmic ray flares. Astrophys. J. 260, 362–370, 1982.

Duggal, S.P. Relativistic solar cosmic rays. Rev. Geophys. Space Phys. 17(5), 1021–1058, 1979.

ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/SUNSPOT_NUMBERS/.http://pgia.ru/CosmicRay/.http://spidr.ngdc.noaa.gov/spidr.http://www.env.sci.ibaraki.ac.jp/database/html/WDCCR/data_e.html.Makhmutov, V.S., Desorgher, L., Bazilevskaya, G.A., Fluckiger, E.,

Raulin, J.P. Evaluation of solar proton spectra using balloon cosmicray observations and Monte Carlo simulation results. Adv. Space Res.39 (9), 1460–1463, 2007.

Rymko, N.P., Tulinov, V.F., Charakhch’yan, A.N. A case of a sharpincrease in cosmic-ray intensity in the stratosphere. Sov. Phys. JETP 36(9), 1202–1203, No. 6, 1959.

Shea, M.A., Smart, D.F. A summary of major solar proton events. Sol.Phys. 127, 297–320, 1990.

Shea, M.A., Smart, D.F. Significant solar proton events for five solarcycles (1954–2007), in: Proceedings of the 30th International CosmicRay Conference, vol. 1 (SH), Merida, Mexico, pp. 261–264, 2008.

Smart, D.F., Shea, M.A. World grid of calculated cosmic ray verticalcutoff rigidities for epoch 2000.0, in: Proceedings of the 30thInternational Cosmic Ray Conference, vol. 1 (SH), Merida, Mexico,pp. 737–740, 2008.

Stozhkov, Y.I. The role of cosmic rays in the atmospheric processes. J.Phys. G 29 (5), 913–923, 2003.

Stozhkov, Yu.I., Svirzhevsky, N.S., Bazilevskaya, G.A., Kvashnin, A.N.,Makhmutov, V.S., Svirzhevskaya, A.K. Long-term (50 years) mea-surements of cosmic ray fluxes in the atmosphere. Adv. Space Res. 44(10), 1124–1137, 2009.

Usoskin, I.G., Mursula, K. Long-term solar cycle evolution: review ofrecent developments. Sol. Phys. 218, 319–343, 2003.

Vashenyuk, E.V., Bazilevskaya, G.A., Balabin, Y.V., Gvozdevsky, B.B.,Makhmutov, V.S., Stozhkov, Y.I, Svirzhevsky, N.S., Svirzhevskaya,A.K., Schur, L.I. The GLE of December 13, 2006 according to theground level and balloon observations, in: Proceedings of the 30thInternational Cosmic Ray Conference, vol. 1 (SH), Merida, Mexico,pp. 221–224, 2008.

http://pgia.ru/CosmicRay/

http://spidr.ngdc.noaa.gov/spidr

http://www.env.sci.ibaraki.ac.jp/database/html/WDCCR/data_e.html

solar proton events recorded in the stratosphere during cosmic ray balloon observations in...

Documents