PHYSICAL REVIEW D 101, 122004 (2020)

Structure of thunderstorm ground enhancements

A. Chilingarian ,1,2,3 G. Hovsepyan,1 T. Karapetyan ,1 G. Karapetyan,1 L. Kozliner,1 H. Mkrtchyan,1
D. Aslanyan,1 and B. Sargsyan1
1
Alikhanyan National Laboratory (Yerevan Physics Institute), Yerevan 0036, Armenia
2
National Research Nuclear University MEPhI, Moscow 115409, Russia
3
Space Research Institute of RAS, Moscow 117997, Russia

(Received 22 February 2020; accepted 1 June 2020; published 22 June 2020)

After having published the first thunderstorm ground enhancements (TGEs) catalog to explain long-
lasting TGEs, we address here problems pertaining to TGE evolution (shape) and atmospheric conditions
supporting the origination of the relativistic runaway electron avalanches. We also address the question of
radon progeny gamma radiation and its contribution to overall TGE flux. We demonstrate that by using
detectors with different energy thresholds we can identify and reliably separate both mechanisms of TGE
origination. An analysis of measured energy spectra of TGEs reveals contributions of both processes to the
TGE temporal evolution and shape. We also confirm the model of radon progeny radiation during a
thunderstorm.

DOI: 10.1103/PhysRevD.101.122004

I. INTRODUCTION effects of the Rn progeny attachment to aerosols and aerosol

Thunderstorm ground enhancements (TGEs) are inten- charging mechanisms enable the uplift of gamma emitters to
sive and prolonged particle fluxes registered on Earth’s the atmosphere and consequent gamma ray emission, which
surface. TGEs measured by particle detectors are correlated gives a significant contribution to the overall TGE count rate
with occurrences of thunder and high strength of the enhancement in the low-energy domain.
atmospheric electric field. Historically, TGEs on Mount The modification of electron and positron energy spectra
Aragats were measured with detectors having a high energy [modification of electron and positron energy spectra
threshold (>3 MeV). The principal engine initiating TGEs (MOS) process] [7] can also enhance the intensity of the
with energies above 3 MeV was determined to be a cosmic ray electrons by several shares of percent; the MOS
relativistic runaway electron avalanche (RREA) [1–3], process does not multiply the number of electrons but
rather enhances the probability of the bremsstrahlung as the
the most powerful natural electron accelerator operating
energy of electrons increases. However, due to very low
in Earth’s atmosphere, which accelerates and multiplies
intensities, only large size detectors can reliably register
seed electrons from the ambient population of cosmic rays
gamma rays with energies above 50 MeV originated by the
(CRs). Simultaneous measurements of the electron, gamma
MOS process.
ray, and neutron fluxes on Mount Aragats [4], and in situ
A network of large NaI crystals (6 units, 12 × 12 ×
observation of RREAs [5,6], as well as measurements of 24 cm each) can reliably recover energy spectra from 0.3
the energy spectra of the electrons and gamma rays [7], to 50 MeV with resolution (FWHM) ∼50% (minute count
proved that RREA is a robust and realistic mechanism for rate is ∼50; 000). Aragats Solar Neutron Telescope
electron acceleration up to 50 MeV (see the right side of (ASNT) comprises 5 and 60 cm thick plastic scintillators
Fig. 1). However, the high-energy flux duration does not with an area of 4 m2 . ASNT measures energy release
exceed a few minutes, and the recently discovered flux histograms in the energy range 5–100 MeV each 20 s
enhancements that last for hours [8] can be explained by (minute count rate is ∼300; 000). Electronics allows one to
another process in the atmosphere, namely, the radiation of measure intensities of electrons and gamma rays in six
the radon progeny lifted upward by the near-surface electric incident directions and estimate their energy spectrum. A
field (see the right side of Fig. 1; details can be found in gamma spectrometer produced by ORTEC measures
[9,10]). 238U and its first five daughter products are solids gamma ray spectrum in the energy range 0.3–3 MeV with
and remain in the ground; however, the sixth daughter resolution ∼7.7%; the minute count rate is ∼12; 000. The
product, 222Rn, is a monoatomic noble gas with density high resolution of the ORTEC spectrometer allows us to
9.73 kg=m3 , which is ≈10 times heavier than air on Mount resolve the 222Rn progeny radiation lines. All spectrometers
Aragats station altitudes. The half-life of 222Rn (3.82 d) is operate continously and can be cross-checked. Energy
long enough to go out into the atmosphere. The well-known spectra are measured and stored at a sampling interval

2470-0010=2020=101(12)=122004(16) 122004-1 © 2020 American Physical Society

A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

FIG. 1. A schematic view of natural radiation enhancement during thunderstorms.

of 1 s (ORTEC), 20 s (ASNT), and 1 min (NaI network). (MN) charge layer and its mirror image on Earth’s surface
Particle fluxes are registered in coincidence with atmos- started to lift charged aerosols with attached Rn progeny.
pheric discharges registered by electric mill EM-100 from After several minutes the concentration of radon progeny
Boltek (the network of four electric mills also monitors a in the atmosphere, it becomes sufficient to add its over-
near-surface electric field with 50 ms resolution) and by whelming share to low-energy cosmic ray flux (below
antennas attached to high-frequency digital oscilloscopes 1.5 MeV). When an electrified cloud approaches the
(capture length is 1 s, including 0.2 s before triggering the particle detector site and, if the strength and spatial extent
flash and 0.8 after). Thus, particle spectrometers with an of the electric field satisfy the conditions for the RREA
unprecedentedly wide energy range (0.3–100 MeV) and initiation, electrons are accelerated up to tens of MeV and
high energy resolution in the low-energy range (<3 MeV) produce an avalanche. RREA is a threshold process that is
provide observation of spectral lines of 222Rn progeny triggered when the potential drop in the atmosphere reaches
gamma radiation, as well as a continuous spectrum of a threshold value that depends on the air density. When the
gamma rays and electrons of TGE up to 100 MeV. atmospheric electric field exceeds this threshold, CR
One of the possible TGE initiation scenarios (realized electrons become “runaways”; instead of wasting all of
mostly in the spring on Mount Aragats) is shown in Fig. 1 its energy on ionization, runaway electrons produce knock-
and explained in detail in [11]. With an approaching on electrons, bremsstrahlung gamma rays, etc. Avalanches
thundercloud, the electric field between the main negative continue until conditions in the cloud satisfy required

122004-2
STRUCTURE OF THUNDERSTORM GROUND ENHANCEMENTS PHYS. REV. D 101, 122004 (2020)

conditions and constitute a hard core of TGE—a few returns to fair-weather values and, consequently, the radon
minutes of an intense flux of electrons and gamma rays progeny updraft drops, the long-lived isotopes (214Pb minus
with energies up to tens of MeV. If the cloud height is low the half-life ≈ 27 min and 214Bi minus the half-life≈
above Earth’s surface, particle detectors register an abrupt 20 min) continue to emit gamma rays. After 60–90 min,
increase in the time series of a count rate lasting a few TGE finally stops and particle flux intensities return to fair-
minutes. In several circumstances, usually in the spring and weather value. The whole development of TGE included
autumn, the lower positive charged layer (LPCR) in the high-energy and low-energy parts lasting for 3–5 h; some-
bottom of the cloud is formed. times a continuous storm can expand this time span
Embedded LPCR increases the potential drop inside the significantly. In the paper, we will present TGEs observed
cloud and, as a result, the RREA process is intensified, and in 2018 and 2019 and discuss the contribution of the RREA
electron–gamma ray avalanches significantly enlarge the and radon progeny radiation to its temporal evolution, as
intensity and maximal energy of TGE. The LPCR origi- well as approving the model of radon progeny radiation
nation is evidenced by abrupt enhancement of count rate of having roots in the aerosol’s updraft by the near-surface
detectors with high energy threshold and usually by the electric field.
TGE termination by a lightning flash followed by the
elimination of a high-energy portion of the RREA. During II. TGEs OBSERVED IN SPRING 2018
thunderstorms, the concentration of charged aerosols near May 2018 was extremely rich with strong TGEs, in
Earth’s surface is highly enhanced [12]. Radon progeny contrast to 2019 and 2020, when we observed no strong
attached to charged aerosols and lifted by the near-surface TGEs. On the morning of May 4, a moderate storm
electric field enlarges concentration of gamma emitters approached Armenia from the southwest and an isolated
above the gamma spectrometers. Therefore, TGE continues thundercloud “sat” just above Mount Aragats. Over the
with much lower energies originating from radon progeny. ≈3 h of the storm, numerous episodes of particle flux
However, the electric field can rise again, and sometimes enhancements and lightning flashes occurred, in both the
we observe several episodes of high-energy particle appear- positive and negative near-surface electric fields. The
ance and elimination after consequent lightning flashes. majority of them were terminated by lightning flashes;
Normal intracloud flash (ICþ) occurs between the MN and see Table I. We map the storm by registering lightning
main positive charge layers, and the inverted intracloud flashes (see the Fig. 2 inset) with a Boltek lightning tracker.
flash (IC−) occurs between the MN and the LPCR. In Table I we show types and occurrence times of
Negative cloud-to-ground flashes (−CG) occur between detected lightning flashes, and also the simultaneous
MN and the ground. Lightning flashes headed to LPCR can detection by the worldwide lightning location network
be continued to the ground and become −CG [13]. (WWLLN).
Particle fluxes made enough ionization in the lower The worldwide lightning location network currently has
atmosphere to provide a path to the lightning leader and over 60 nodes around the globe and is operated by the
very often lightning flash terminates the TGE [14]. As the management team led by Professor Robert Holzworth
storm ends, the RREA and MOS processes completely of the University of Washington in Seattle [15]. One of
disappear and only low-energy (<3 MeV) particles can be the WWLLN nodes is installed at the headquarters of the
found in the flux. Although the near-surface electric field Cosmic Ray Division at the Yerevan Physics Institute.

TABLE I. Lightning flashes detected at Mount Aragats during a May 4, 2018 thunderstorm.

Time (UT) Lightning type Detection by WWLLN

1 09:41:22.774 Inverted IC TGE-terminating flash
2 09:46:05.326 Hybrid: inverted IC followed by–CG 09:46:05.327
3 09:58:44.548 Inverted IC
4 09:59:37.817 Hybrid: inverted IC followed by–CG
5 10:16:02.715 Inverted IC TGE-terminating flash 09:59:38.176
6 10:17:29.569 Inverted IC TGE-terminating flash
10:17:29.635
10:17:29.636
7 10:19:09.390 Inverted IC TGE-terminating flash
8 10:21:53.173 Inverted IC TGE-terminating flash
10:21:53.203
10:21:53.204
10:21:53.205
9 10:27:04.788 -CG TGE-terminating flash

122004-3
A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

FIG. 2. Time series of disturbances of the near-surface electric field and chart of Armenia with thunderclouds mapped as ICþ (red
pluses), IC − ðblue plusesÞ, and CG − ðblue crossesÞ.

Facilities used for synchronous detection of particle In Fig. 3 we show the time series of the count rate
fluxes and atmospheric discharges on Mount Aragats of a NaI spectrometer with a low energy threshold
are described in detail in [14]. Global Positioning (E > 0.3 MeV) measured during the storm. A detector
System based timing system provides microsecond scale with a low energy threshold shows an ∼4 h long particle
synchronization. flux enhancement, with the largest peak coinciding with

FIG. 3. 1 min time series of a count rate measured with three plastic scintillators with various energy thresholds.

122004-4
STRUCTURE OF THUNDERSTORM GROUND ENHANCEMENTS PHYS. REV. D 101, 122004 (2020)

the excursion of the near-surface electric field in the deep drastically differ from the NaI time series. High-energy
negative domain at ≈10∶35 universal time (UT). The particles are related to the RREA development in the
same TGE was measured with plastic scintillators of a atmosphere above the detectors. The runaway process
STAND3 detector with high energy thresholds using a 3 cm requires a rather stringent condition of the strength and
thick plastic scintillator (blue, E > 10 MeV and, red, extension of the atmospheric electric field. Thus, instead of
E > 20 MeV). STAND3 comprises four plastic scintilla- a continuous enhancement for many hours, in a count rate
tors stacked horizontally (details of detector construction of high-energy particles, we can see several discrepant
episodes of the RREA; the largest ones started at 9:30,
are described in [16]). All scintillators have the same area
10:00, 10:30, and 10:45. The maximal energies of the
of one m2 ; when TGE particles enter the detector in the differential energy spectra measured each minute by the
near-vertical direction, they have to cross upper scintillators NaI spectrometers are shown in the inset of Fig. 3. We can
to reach and be registered by lower ones. Thus, the energy see that maximal energy during the RREAs reaches
threshold is higher for lower scintillators. We show in Fig. 3 ≈40 MeV, whereas, during most of the TGE’s duration
time series of the count rate measured by the STAND3 when particle detectors register Rn progeny radiation only,
detector under the condition that signals appear simulta- maximal energy does not exceed 3 MeV.
neously at the outputs of two upper scintillators (1100 Therefore, the shape of the TGE can be rather sophis-
combinations) and three upper scintillators (1110 combi- ticated. It is controlled by the intracloud and near-surface
nations). The flux intensity is shown in the percent of the electric fields and depends on the particular detector that
flux measured at fair weather just before the beginning monitors particle flux. After the decay of the near-surface
of the TGE. The time series of high-energy particles electric field, the count rate of NaI scintillators does not

FIG. 4. Relation of lightning flashes to particle flux enhancements at 9:30–10:10 (a) and 10:10–10:50 (b); numbers in the hexagons
correspond to those in Table I. TGEs denoted as 1, 3, and 5–8 were terminated by the lightning flash, TGEs 4 and 9 finished smoothly,
and lightning flash 2 is not related to any sizable TGE.

122004-5
A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

FIG. 5. Differential energy spectra recovered at minutes of maximal flux (the third largest RREA episode on May 4, 2018).

immediately stop because of the radiation from the radon In Figs. 4(a) and 4(b), we demonstrate the relation of
progeny with a lengthy half-life time. Such a long tail of the particle fluxes to lightning flashes. The numbers above the
decaying TGE is common for all ∼250 TGEs analyzed in flashes are the same as in Table I. In Fig. 4, we can see that
2017–2019. lightning flashes of both types can abruptly terminate

FIG. 6. 20-sec energy release (deposit) histograms measured by the ASNT detector, terminated by the lightning flash. In the inset 2 s
time series with the same detector count rate.

122004-6
STRUCTURE OF THUNDERSTORM GROUND ENHANCEMENTS PHYS. REV. D 101, 122004 (2020)

particle fluxes. For instance, at 09:41:22 (flash N1 in

Table I) an inverted intracloud lightning flash stopped
the TGE development. At 10:27:04 (flash N9 in Table I),
a negative cloud-to-ground (−CG) flash terminates the
TGE. The largest peak started at ∼10∶30. Flux maximum
containing highest energies occurred from 10∶33 to 10:42
and peaked from 10:36 to 10:37. The long duration of this
episode of the TGE is explained by the absence of flashes
and is rather rare; as can be seen in Table I and Figs. 4(a)
and 4(b), for six episodes the TGEs were terminated by
lightning flashes, and the flux enhancement decayed
smoothly in only two episodes (4 and 9 in Table I and
in Fig. 4).
In Fig. 5, we demonstrate a sequence of 1 min differ-
ential energy spectra measured with the NaI spectrometer at
the highest intensity and highest maximal energy TGE
FIG. 7. NaI spectrometer (N4) covered with lead bricks (4 cm)
episode from 10:32 to 10:39. For all spectra, the minute’s
to prevent registration of the near-vertical RREA.
maximal energy is above 20 MeV, reaching 40 MeV for
4 min of maximal flux. The maximal energy of the other
TGE episodes is also well above 3 MeV. Thus, we can performed using covered by lead and “open” from the top
undoubtedly identify all of these episodes with the RREA NaI spectrometers, we separated both types of TGE drivers.
process in a thundering atmosphere. For prolonged TGEs Observing differential spectra of RREA and spectrum of
with several initiations of RREA, the pattern of TGE radon isotopes, we could confirm this separation by the
measured by detectors with energy threshold larger than expected absence of high-energy particles in the spectra
3 MeV comprises several peaks with a duration of a few measured by the detector with a lead filter on top.
minutes; the detectors with an energy threshold of hundreds In Fig. 8, we show disturbances of the near-surface
of keV demonstrate a more or less continuous pattern, electric field (black curve in the middle), and 1 min count
filling the gaps between RREA episodes with the isotropic rates of NaI spectrometer open to radiation from above
radiation of radon progeny lifted in the atmosphere by the (upper blue curve) and covered by lead bricks (bottom blue
near-surface electric field. curve). Thus, gamma rays and electrons arriving from the
In Fig. 6, we show the energy release spectra measured near-vertical direction are counted by NaI spectrometers
by other particle spectrometers, a 60 cm thick and 4 m2 number 2, and only inclined gamma rays from radon
area plastic scintillator called the ASNT; see the details of progeny radiation can be registered using the NaI spec-
the operation in [17]. The ASNT measures the energy trometer N4.
release of incident particles and stores data as 20-s histo- To better visualize the difference of time series of count
grams. Each slice in the two-dimensional histogram in rates obtained with and without lead filters, we show in
Fig. 6 is such an energy release spectrum. In the figure, we Fig. 9 count rates in percent observed by both detectors for
can see that the lightning flash abruptly terminates the TGE fair-weather values measured just before the TGE start. In
and that the high-energy particles disappear from the flux. Fig. 9, we can see a delay (of ∼10 min) of TGE measured
In the Fig. 6 inset, we can see the abrupt decline of the by a NaI detector covered with lead filters (blue curve)
particle flux the same second as the lightning flash occurs. relative to the start of TGE measured by open spectrometer
The lightning critically decreases the potential drop in the (black curve). The gamma rays from the RREA cascade
atmosphere and the RREA process stops. developed in the thundercloud arrive at Earth’s surface
earlier than the concentration of the radon progeny above
the experimental hall where NaI spectrometers are located
III. TGEs OBSERVED IN SPRING
becomes sufficient to provide significant flux entering the
AND SUMMER 2019
NaI spectrometer from the sides.
To confirm the long-lasting TGE origination by radon The intensity of these “early” gamma rays is low, and the
progeny, we performed in 2019 several experiments expos- energy is comparable to that of the radiation from isotopes
ing a NaI detector covered by thick (4 cm) lead filters to because on the early stage of TGE; only Compton scattered
atmospheric radiation; see Fig. 7. The lead filter prevented gamma rays from remote RREAs enter NaI spectrometers
the registration of near-vertical gamma radiation expected from the side under large zenith angles. When a thunder-
from the RREA process; see the right side of Fig. 1. cloud with an embedded LPCR is above particle detectors,
However, isotropic distributed gamma radiation emitted by both energy and maximal energy are drastically enhanced.
radon progeny entered the NaI detector from open sides of In the insets of Fig. 9, we show the histograms
the NaI spectrometer. In this way, comparing measurements of maximal energies of gamma rays in the whole time

122004-7
A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

FIG. 8. Time series of disturbances of the near-surface electric field (black curve) and one-minute count rate of NaI crystal open to
radiation from above (upper blue curve) and covered by lead bricks (lower blue curve).

interval: in the left inset measured by an open spectrometer, “covered” NaI spectrometer is very small and does not
in the right inset measured by one covered with lead. allow us to recover the energy spectra).
Maximal energies in the two cases are drastically different. The maximal energy measured by the open spectrometer
To emphasize this difference in Figs. 10 and 11, we show reaches 10 MeV, and spectra measured by the spectrometer
energy spectra for the time span indicated in Fig. 9 by two under lead do not show energies above 1 MeV, which is in
red lines (at 15:52–15:55 the TGE flux measured by the good agreement with energies of gamma rays emitted by

FIG. 9. 1 min time series of count rates measured by two NaI detectors with (blue curve) and without (black curve) lead filters. (Right
inset) Histogram of maximal energies of gamma rays measured with a NaI spectrometer with a lead filter on top (inclined gamma rays
from radon progeny decay only). (Left inset) Maximal energy of gamma rays measured with a NaI spectrometer with an open top
(registering near-vertical RREAs as well). 1 min energy spectra depicted in Figs. 10 and 11 corresponding to the time span outlined by
the two red lines.

122004-8
STRUCTURE OF THUNDERSTORM GROUND ENHANCEMENTS PHYS. REV. D 101, 122004 (2020)

FIG. 10. Differential energy spectra measured on August 27, 2019 from 15:52 to 15:59 (spectrometer N2, hardcore of TGE and the
RREA).

most intense isotopes for 214Pb (0.354 MeV) and 214Bi temperature) and differences in values of the near-surface
(0.609 MeV). After the end of the storm at ∼16∶15, the electric field.
particle flux intensity decays according to the lifetime of
these isotopes and completely stops at 17:50. Thus, the IV. CORRELATION OF ABSOLUTE VALUE
shape of TGE from “pure” radon progeny radiation is not OF THE ELECTRIC FIELD AND RADON
symmetric, demonstrating a rather fast rise and slow decay. PROGENY RADIATION
In Fig. 12, we show the decay of TGE measured by a NaI
spectrometer with lead on top for several TGEs observed in The large NaI spectrometers measuring energy spectra
autumn 2019. The histograms of decaying particle flux from 0.3 to 50 MeV cannot resolve spectral lines of radon
intensities are shown after the stop of the storm and return progeny. However, ORTEC’s precise spectrometer that
of the near-surface electric field to fair-weather magnitude. measures energy spectra from 0.3 to 3 MeV provides an
The half-life of TGE decay (18–35 min) fits well with the exact pattern of spectral lines and can be used to measure
half-life of the most abundant gamma emitters from the the enhancement of gamma rays emitted by each isotope
radon chain, namely, 214Pb (with a half-life of 27 min) (see [9] for details). Special electronics allow us to not only
and 214Bi (with a half-life of 20 min). The discrepancy of measure the time series of count rates using an ORTEC
the magnitudes of half-life for different TGE events spectrometer but also obtain histograms of energy releases
can be explained by the changing atmospheric conditions each minute. As usual, the TGE energy spectrum was
(wind velocity, precipitation, atmospheric pressure, and recovered from the difference of energy spectrum registered

FIG. 11. Differential energy spectra measured on August 27, 2019 from 15:56 to 15:59 (spectrometer N4, radon progeny gamma
radiation).

122004-9
A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

FIG. 12. Decay of TGE measured with a NaI spectrometer with a lead filter on top.

during TGE and before TGE at fair weather (see [9] for 1.2 MeV comes from the radon and thoron chain isotopes
details). In Table II, we collect information on the major (75%). In Fig. 13, we demonstrate time series of gamma
gamma emitters for four selected events of enhanced rays originated from Rn progeny only for the same four
particle fluxes observed in autumn 2019. Low-energy selected episodes of particle flux enhancement. By selected
natural gemma radiation (NGR) comprises mostly gamma gamma rays from spectral lines only, we highly suppress
emitters from radon and thoron decay chains, from a possible contamination of low-energy RREA gamma rays,
511 keV positron annihilation peak, and from the low- gamma rays from EAS, and other sources, leaving the only
energy part of extensive air showers (EAS), gigantic contribution of gamma radiation from radon and thoron
avalanches start high in the atmosphere in interactions of progenies. In Figs. 13(a)–13(d), we can see that 10%–20%
galactic or solar cosmic rays with atmospheric atoms. Low- enhancement of isotope radiation coincides with disturb-
energy gamma rays from RREA avalanches also contribute ances of the near-surface electric field of both polarities.
to continuous gamma radiation. NaI spectrometers also Our goal is to demonstrate the influence of the near-
measure additional continuous low energy radiation. surface electric field on the concentration of radon and
Because of the small size of the crystals (5 × 5 cm), gamma thoron progenies in the atmosphere and on the consequent
rays can escape from crystal sides without depositing whole changes of the radiation level with an increase in the near-
energy. Thus, a continuum of lower-energy gamma rays surface electric field of both polarities. For the correlation
(called Compton scattered gamma rays) is concentrated in analysis, we select time spans in the time series of the count
the energy spectrum to the left of the spectral lines emitted rates, which correspond to more or less stable values of the
by radon and thoron progenies. near-surface electric field (both positive and negative). In
In the second column of Table II, we show the total Fig. 13, these time spans are indicated by red lines. We
amount of enhancement registered during the event. The select the most abundant 214Bi (609 keV) isotope and
duration and date of the event are posted in the first column. measure its intensity in selected time spans with an ORTEC
Then we show the absolute enhancement and its share of spectrometer, and we also measure the near-surface electric
important isotopes from radon and thoron chains. In the last field with a BOLTECK EFM-100 electric mill. In Fig. 14,
five columns we summarize the characteristics of the we show the scatterplot corresponding to positive values of
additional NGR during thunderstorms. I3 is the total the near-surface electric field, and in Fig. 15 we show the
number of gamma rays registered by the ORTEC spec- scatterplot corresponding to negative values.
trometer, I1.2 is the number of gamma rays below 1.2 MeV, The rather strong correlation shown in Figs. 14 and 15
and IR is the number of gamma rays from isotopes with (R ≈ 0.8 for both negative and positive fields) demonstrates
energies below 1.2 MeV. Energies of most of the NGR that the near-surface electric field effectively lifts isotopes
emitters measured by the NaI spectrometer are below to the atmosphere. In many studies, it has been observed
1.2 MeV (85%). And most of the gamma radiation below that radon and its progeny are very mobile and readily

122004-10
 TABLE II. Characteristics of the natural gamma ray radiation measured in autumn 2019 during four events of particle flux enhancement.

Total CRþ Intensity until Isotopes

Date intensity (I3) Compton 1.2 MeV(I1.2) only up to Share Share
and time 0.3–3 MeV 214Pb354 keV 511 KeV 214Bi609 keV 214Bi768 keV AC228 911 keV 214Bi1.12 MeV 214Bi1.76 MeV 214Bi2.2 MeV scattered 0.3–1.2 MeV 1.2 MeV (IR) (IR/I1.2) % (I1.2/I3)

2019-09-11, 66384 12975 4334 11838 7709 2827 3614 2367 819 19901 54164 38963 71.9 81.6
start at 12:05,
count rate
(90 min)
% 5.6 19.5 6.5 17.8 11.6 4.3 5.4 3.6 1.2 30.0
2019-09-28, 28524 6058 2531 5096 4587 1427 1992 1231 211 5391 23442 19160 81.7 82.2
start at 09:45,
count rate
(110 min)
% 2.0 21.2 8.9 17.9 16.1 5.0 7.0 4.3 0.7 18.9

122004-11
2019-11-01, 22790 5822 2309 3593 2767 1137 1339 640 510 4673 20929 14658 70.0 91.8
start at 16:30,
count rate
STRUCTURE OF THUNDERSTORM GROUND ENHANCEMENTS

(95 min)
% 1.8 25.5 10.1 15.8 12.1 5.0 5.9 2.8 2.2 20.5
2019-11-27, 16922 4281 1484 3281 1759 618 962 404 301 3832 14403 10901 75.7 85.1
start at 17:07,
count rate
(60 min)
% 2.2 25.3 8.8 19.4 10.4 3.7 5.7 2.4 1.8 22.6 28234.5 20920.5 74.8 85.2
PHYS. REV. D 101, 122004 (2020)
A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

FIG. 13. Near-surface electric field disturbances with multiple lightning flashes and particle enhancements measured with an ORTEC
spectrometer (gamma emitter lines only) for TGEs occurred on September 1, 2019 (a), September 4, 2019 (b), September 11, 2019 (c),
and August 27, 2019 (d).

attach to aerosol surfaces [14]. Emanated radon progeny absolute value of the field, the higher the concentration of
become airborne and immediately attach to the dust radon isotopes in the atmosphere and the more intense the
particles and aerosols existing in the atmosphere. In gamma ray flux measured by the spectrometer. The large
addition, the aerosol particles carry negative or positive correlation apparent in the figures confirms our scenario for
charges and the number of elementary charges of atmos- the TGE origination [11]. Thus, both large positive and
pheric particles is usually above 1 [18]. In a positive near- large negative electric fields effectively lift the aerosols
surface electric field, negative aerosols are lifted by the with attached gamma emitters to the atmosphere. When
near-surface electric field, whereas in a negative electric their concentration in the atmosphere above particle
field, positive ones are lifted. Thus, a large near-surface detectors increases, the NaI spectrometers register enhance-
field of both polarities, as we see in Figs. 13–15, enhances ment of the gamma ray flux. The enhanced flux continues
the gamma ray intensity by 10%–20%. The larger the

FIG. 14. Scatterplot of positive strengths of a near-surface FIG. 15. Scatterplot of negative strengths of a near-surface
electric field with a corresponding intensity of a 214Bi (609 keV) electric field with a corresponding intensity of a 214Bi (609 keV)
isotope measured with an ORTEC spectrometer. isotope measured with an ORTEC spectrometer.

122004-12
STRUCTURE OF THUNDERSTORM GROUND ENHANCEMENTS PHYS. REV. D 101, 122004 (2020)

FIG. 16. Relationship between outside temperature and cloud

base height for TGEs over different seasons.
FIG. 17. Relationship between the intensity of TGE (peak
height is measured by the number of standard deviations Nσ),
60–90 min after the storm stops until all isotopes lifted by cloud base height, and outside temperature for the event “cluster”
the near-surface electric field finally decay. outlined in Fig. 14.

V. STATISTICAL ANALYSIS OF TGE EVENTS

made during those years were the discovery of the origin
REGISTERED ON MOUNT ARAGATS IN 2017–2019
of the low-energy part of the TGE and revelation of the
The natural electron accelerator emerging in the thunder- meteorological conditions supporting the largest TGE
clouds above the Mount Aragats high-altitude research origination. In Fig. 16, we present a scatterplot showing
station in Armenia, operated continuously in 2017–2020, the two-dimensional pattern of outside temperatures and
provided more than 250 TGEs (see also the first catalog of cloud base heights for TGEs registered in 2017–2019 in
the TGE events [19]). Raw data are available via the ADEI different seasons. In the figure, we can see that the harvest
interactive Web platform. The most important discoveries of TGEs peaked in the spring and that we can have also

FIG. 18. Monthly distribution of TGE activity in 2017–2019. (Inset) Mean temperatures in May from 2017 until 2020.

122004-13
A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

TGEs in the winter. The most striking feature of the both larger (2019) and smaller (2020) temperatures lead to
scatterplot is the cluster outlined with an oval in Fig. 16. a decrease of the TGE number.
We found that the largest TGEs fall within this cluster; The distribution of TGEs by the peak significance
see the three-dimensional histogram in Fig. 17. Thus, estimated in the number of standard deviations from the
optimal conditions when a storm generates large TGEs mean count rate value is shown in Fig. 19. As expected,
are an outdoor temperature of −2–2 °C and a corresponding it is a rapidly declining distribution, with the mode at
low height (25–100 m) of the cloud base. The motion of the around 10.
storm relative to Mount Aragats station makes it possible to
investigate meteorological conditions in the lower part of VI. CONCLUSIONS
the cloud. In the spring, the cloud is usually “sitting” on the
We analyzed the TGE development according to the
station, as can be seen in Fig. 16, as well as from the shots main physical processes responsible for TGE origination,
of a panoramic camera located at the station. Temperatures namely, RREA and radon progeny radiation. We explained
around 0 °C isotherm (corresponding to a very low cloud the impact of both processes on the TGE shape and energy
base height of 10–50 m) outline the domain where most of spectrum. We concluded that TGE is a rather complicated
the large TGEs were observed (of course, in the presence of phenomenon having roots in at least three physical proc-
a strong electric field). LPCR average position coincides esses related to thundering atmospheres. These processes
exactly with the 0 °C level [20]. As mentioned in the are controlled by the electric field emerging in the thunder-
Introduction, the origination of the lower dipole signifi- cloud and near Earth’s surface. RREA is a triggered process
cantly enhances the electric field inside the cloud, and the that starts in a thundercloud only when the electric field
initiation of the runaway process becomes possible. The surpasses the threshold value specified for the particular
cloud electron accelerator greatly multiplies the ambient atmospheric density. Gamma radiation of radon and thoron
population of cosmic ray electrons, significantly enlarging origin starts when the updraft of aerosols (with attached
fluxes of gamma rays and neutrons as well. radiated isotopes) provides sufficient concentration of
In Fig. 18, we demonstrate the frequency of TGEs by gamma ray emitters at heights above particle detectors.
year and month. The distribution peaks in May for 2017 RREA radiation is near vertical, whereas the isotope
and 2018 were similar to distributions observed in previous radiation is isotropic (see Fig. 1) and can be registered
years. Thus, the analysis of TGEs observed in 2019 and at large zenith angles. The MOS process contributes to the
2020 demonstrates the importance of the mean monthly TGE by few a shares of a percent for all energies; however,
temperature for TGE origination. As we mentioned pre- due to low intensity, we cannot yet separate it from the two
viously, the necessary condition for the unleashing of large main processes (RREA and radon progeny radiation).
TGEs emerges with a LPCR occurring at the freezing (0 °C) We demonstrated that there are multiple signatures
level. In Figs. 16 and 17, we can see that small (below (tracers, tags) of the RREA occurrences within the long-
þ2 °C) positive temperatures correspond to the maximum lasting TGE:
number of large TGEs. In the inset of Fig. 18, we show the (1) An abrupt surge of particle flux intensity for several
mean temperatures for the last 4 yr, and we can see that minutes.
(2) Presence of gamma rays/electrons with energies
above 3 MeV in the energy spectra.
(3) Detection of the individual electron avalanches by
the distributed surface array.
(4) Abrupt decline of high-energy species (>3 MeV) of
TGE caused by lightning flashes.
(5) Origination of LPCR evidenced by a reversal of
polarity of the near-surface electric field and by
detection of the graupel fall.
(6) Detection of the enhanced fluxes from the near-
vertical direction.
We separate pure radon and thoron progeny radiation
as a continuous part of hours-long TGE. The shape of
TGE time series is rather complicated (see Fig. 3) and is
controlled by the intracloud electric field and near-surface
electric field, and by the decay time of the most frequent
214Pb (0.354 MeV) and 214Bi isotopes of a radon decay

chain. After a fast rise, the TGE continues with a long

FIG. 19. Histogram of TGEs according to intensity of the peak decaying “tail.” Thus, the shape of the TGE can be
measured by NaI no. 1 spectrometer for 2017–2019. separated into three species:

122004-14
STRUCTURE OF THUNDERSTORM GROUND ENHANCEMENTS PHYS. REV. D 101, 122004 (2020)

(1) Induced by relativistic runaway electron avalanches electrons downward in the direction of Earth. Recently, the
in the thundercloud. Large, reaching several hundred tripole structure of the electric field above Mount Aragats
percent peaks above background, lasting a few (including the location of the “snow” and “graupel”
minutes with particle energies reaching tens of dipoles) assumed in this paper was confirmed by radar
MeV; fluxes are usually interrupted by a lightning and satellite measurements [22]. We tried to use for the
flash. Particles come from the near-vertical di- understanding of the dynamics of the electric field in the
rection. cloud a new type of information: particle fluxes and energy
(2) Radon progenies radiation.Low-energy (<3 MeV) spectra as they are measured on Earth’s surface during a
hours of continuous radiation never interrupted by thunderstorm. We directly measured these variables at
lightning; particles become isotropic. rather high precision and connected them to electric field
(3) The decay phase.Decay of radon progeny that are origination between the main negative and its mirror on
still concentrated in the air after the storm finishes. Earth’s surface and between the main negative and LPCR.
The half-life time of TGE decay is consistent with Certainly, there can be other scenarios of TGE develop-
the half-life time of 214Pb (∼300 keV peak) and 214Bi ment. The scenarios of the origination of the downward
(∼600 keV peak) isotopes from the radon chain. electron-accelerating electric field are numerous, and cor-
Certainly, the large peaks and large energies of TGE responding TGEs may vary in intensity and energy spectra,
particles originating from the RREAs cannot be observed if as well as in the fraction of particles reaching Earth’s
the electric field inside the cloud does not exceed runaway surface.
threshold, say, in the winter; in this case, the shape of TGE The data for this article are available by accessing the
is Gaussian-like with a long, exponentially decaying tail. In multivariate visualization software ADEI in numerical and
addition, there can be numerous peaks in TGE develop- graphical type on the website of the Cosmic Ray Division
ment, stopped each time by a lightning flash lasting a few of the Yerevan Physics Institute [23].
hundreds of a millisecond, and again rising to be stopped
by another flash. Owing to the enormous variety of TGE ACKNOWLEDGMENTS
shapes, it is not possible to fit the shape specific to each
particular TGE with one family of parametric functions. We We thank the staff of the Aragats Space Environmental
outline in points 1–3 some common features of the shape Center for the operation of the NaI network on Mount
that we think are useful for analysis and classification Aragats. A. C. thanks E. Williams for the useful, stimu-
purposes. lating discussions and S. Soghomonyan for the valuable
LPCR origination depends on many difficult to measure comments and for providing the data posted in Table I. The
variables: size and shape of the hydrometeors, relative authors thank S. Chilingaryan for his continuous efforts
velocity of the collision objects, water content, etc. at maintaining and improving Web based data analysis
However, based on the analysis of 13 events of TGE facilities for a large stream of data coming online from the
termination by inverted-polarity ICs and hybrid flashes, in Mount Aragats research station. We appreciate the support
[21] we demonstrated for the first time that an electric field of the Russian Science Foundation Grant through Project
between the main negative layer and a LPCR accelerates No. 17-12-01439-P.

[1] A. Gurevich, G. Milikh, and R. Roussel-Dupre, Runaway thunderstorm-correlated fluxes of high-energy electrons,
electron mechanism of air breakdown and preconditioning gamma rays, and neutrons, Phys. Rev. D 82, 043009 (2010).
during a thunderstorm, Phys. Lett. A 165, 463 (1992). [5] A. Chilingarian, G. Hovsepyan, and A. Hovhannisyan,
[2] L. Babich, E. Donskoy, I. M. Kutsyk, A. Yu. Kudryavtsev, Particle bursts from thunderclouds: Natural particle
R. A. Roussel-Dupré, B. N. Shamraev, and E. M. D. accelerators above our heads, Phys. Rev. D 83, 062001
Symbalisty, Comparison of relativistic runaway electron (2011).
avalanche rates obtained from Monte Carlo simulations and [6] A. Chilingarian and G. Hovsepyan, and B. Mailyan, In situ
from kinetic equation solution, IEEE Trans. Plasma Sci. 29, measurements of the runaway breakdown (RB) on Aragats
430 (2001). Mountain, Nucl. Instrum. Methods Phys. Res., Sect. A 874,
[3] J. Dwyer, A fundamental limit on electric fields in air, 19 (2017).
Geophys. Res. Lett. 30, 2055 (2003). [7] A. Chilingarian, B. Mailyan, and L. Vanyan, Recovering
[4] A. Chilingarian, A. Daryan, K. Arakelyan, A. Hovhannisyan, of the energy spectra of electrons and gamma rays
B. Mailyan, L. Melkumyan, G. Hovsepyan, S. Chilingaryan, coming from the thunderclouds, Atmos. Res. 114–115, 1
A. Reymers, and L. Vanyan, Ground-based observations of (2012).

122004-15
A. CHILINGARIAN et al. PHYS. REV. D 101, 122004 (2020)

[8] A. Chilingarian, G. Hovsepyan, S. Soghomonyan, M. [16] A. Chilingarian, L. Vanyan, and B. Mailyan, Observation
Zazyan, and M. Zelenyy, Structures of the intracloud electric of Thunderstorm Ground Enhancements with intense
field supporting origin of long-lasting thunderstorm ground fluxes, of high-energy electrons, Astropart. Phys. 48, 1
enhancements, Phys. Rev. D 98, 082001 (2018). (2013).
[9] A. Chilingarian, A. Avetisyan, G. Hovsepyan, T. Karapetyan, [17] A. Chilingarian, S. Chilingaryan, and G. Hovsepyan,
L. Kozliner, B. Sargsyan, and M. Zazyan, Origin of the low- Calibration of particle detectors for secondary cosmic rays
energy gamma ray flux of the long-lasting thunderstorm using gamma-ray beams from thunderclouds, Astropart.
ground enhancements, Phys. Rev. D 99, 102002 (2019). Phys. 69, 37 (2015).
[10] A. Chilingarian, Reply to Comment on ‘Long lasting low [18] Y. He, Z. Gu, W. Lu, L. Zhang, T. Okuda, K. Fujioka,
energy thunderstorm ground enhancements and possible H. Luo, and C. W. Yu, Atmospheric humidity and particle
222 charging state on agglomeration of aerosol particles, Atmos.
Rn daughter isotopes contamination’, Phys. Rev. D 99,
108102 (2019). Environ. 197, 141 (2019).
[11] A. Chilingarian, G. Hovsepyan, and A. Elbekian, T. [19] A. Chilingarian, H. Mkrtchyan, G. Karapetyan, B.
Karapetyan, L. Kozliner, H. Martoian, and B. Sargsyan, Sargsyan, and A. Arestakesyan, Catalog of 2017 thunder-
Origin of enhanced gamma radiation in thunderclouds, storm ground enhancement (TGE) events observed on
Phys. Rev. Research 1, 033167 (2019). Aragats, Sci. Rep. 9, 6253 (2019).
[12] H.-K. Lee and K.-H. Ahn, Charging effect on the [20] J. Kuettner, The electrical and meteorological conditions
80–200 nm size atmospheric aerosols during a lightning inside thunderclouds, J. Meteorol. 7, 322 (1950).
event, Aerosol Air Qual. Res. 17, 2624 (2017). [21] A. Chilingarian, Y. Khanikyants, V. A. Rakov, and S.
[13] A. Nag and V. Rakov, Some inferences on the role of lower Soghomonyan, Termination of thunderstorm-related bursts
positive charge region in facilitating different types of of energetic radiation and particles by inverted-polarity
lightning, Geophys. Res. Lett. 36, L05815 (2009). intracloud and hybrid lightning discharge, Atmos. Res.
[14] A. Chilingarian, S. Chilingaryan, T. Karapetyan, L. 233, 104713 (2020).
Kozliner, Y. Khanikyants, G. Hovsepyan, D. Pokhsraryan, [22] E. K. Svechnikova, N. V. Ilin, E. A. Mareev et al.,
and S. Soghomonyan, On the initiation of lightning in Characteristic features of the clouds producing thunder-
thunderclouds, Sci. Rep. 7, 1371 (2017). storm ground enhancements (to be published).
[15] See http://wwlln.net/. [23] http://adei.crd.yerphi.am/adei.

122004-16