The prolonged gamma ray enhancement and the short radiation burst events observed in thunderstorms at Tien Shan
A. Shepetov, V. Antonova, O. Kalikulov, O. Kryakunova, A. Karashtin, V. Lutsenko, S. Mamina, K. Mukashev, V. Piscal, M. Ptitsyn, V. Ryabov, T. Sadykov, N. Saduev, N. Salikhov, Yu. Shlyugaev, L. Vildanova, V. Zhukov, A. Gurevich
TThe prolonged gamma ray enhancement and the short radiationburst events observed in thunderstorms at Tien Shan
A. Shepetov a , V. Antonova b , O. Kalikulov c , O. Kryakunova b , A Karashtin d , V. Lutsenko b ,S. Mamina a , K. Mukashev c , V. Piscal a , M. Ptitsyn a , V. Ryabov a , N. Saduev c , T. Sadykov e ,N. Salikhov b , Yu. Shlyugaev f , L. Vil’danova a , V. Zhukov a a P. N. Lebedev Physical Institute of the Russian Academy of Sciences (LPI), Leninsky pr., 53, Moscow, Russia,119991 b Institute of Ionosphere, Kamenskoye plato, Almaty, Republic of Kazakhstan, 050020 c Al-Farabi Kazakh National University, Institute of Experimental and Theoretical Physics, al-Farabi pr., 71, Almaty,Kazakhstan, 050040 d Radiophysical Research Institute, Bolshaya Pechyorskaya, str., 25 / e Satbayev University, Institute of Physics and Technology, Ibragimova str. 11, Almaty, Kazakhstan, 050032 f Institute of Applied Physics of RAS, Ul’yanova str., 46, Nizhny Novgorod, Russia
Abstract
We report the observation results of the hard radiation flashes which accompanied the lightningdischarges above the mountains of Northern Tien Shan. Time series of the counting rate inten-sity, numerical estimations of absolute flux, and energy distribution of accelerated electrons andof (20–2000) keV gamma rays were obtained at the height of 3700 m a. s. l., immediately withinthunderclouds, and in closest vicinity ( (cid:46)
100 m) to discharge region. Two di ff erent kinds of ra-diation emission events are presented here: a relatively prolonged rise of gamma ray intensitywith minute-scale duration (the thunderstorm ground enhancement, TGE) which has preceded anegative field variation, and a short sub-millisecond radiation burst, which accompanied a closelightning discharge in thundercloud. It was revealed also an indication to positron generation inthunderclouds at the time of gamma ray emission, as well as modulation of the neutron countingrate in Tien Shan neutron monitor which was operating at a (1.5–2) km order distance from theregion of lightning development. Keywords: thunderstorm, lightning, runaway electrons breakdown, gamma ray glow, TGE,neutron monitor modulation
1. Introduction
By many experiments which were realizedduring two latter decades in the field of highenergy atmospheric physics it was detected
Email address: (A. Shepetov) the phenomenon of a prolonged excess of theflux of high energy charged particles and localgamma radiation above its usual backgroundwhich takes place in thunderstorm times. Typ-ically, the relative amplitude of the intensityincrease seen during such events varies froma few tens to thousands of percents, and their
Preprint submitted to Elsevier a r X i v : . [ phy s i c s . a o - ph ] S e p uration may last from a few seconds ordertimes up to tens of minutes. The energy rangeof gamma rays registered in the times of en-hanced emission was of about (0.1–10) MeV,and that of the charged particles—up to the(10–100) MeV order. In many cases such ra-diation enhancements were immediately pre-ceding, but not coinciding with, a subsequentlightning, and terminated just at the momentof a nearby lightning discharge. First time ob-served from an airplane by Parks et al. (1981),later on the events of such type were multipletimes detected in upper atmosphere both abovethundercloud tops, e.g. by McCarthy andParks (1985); Kelley et al. (2015); Kochkinet al. (2017); Østgaard et al. (2019), and im-mediately within thunderclouds, with balloon-born detectors such as of Eack et al. (1996,2000). In ground-based experiments these ef-fects were observed at many particles detec-tor installations situated at mountain altitudes,as by Chilingarian et al. (2010, 2013b, 2019);Tsuchiya et al. (2012); Kudela et al. (2017),and at the sea level: Torii et al. (2002, 2011);Tsuchiya et al. (2007, 2013); Bogomolov et al.(2015, 2016); Wada et al. (2019a). In scien-tific literature a somewhat diverse terminol-ogy has been established for these phenomenawhich were referred to as “prolonged radia-tion bursts” in Tsuchiya et al. (2007), “gammaray glows” by Kochkin et al. (2017), “thunder-storm ground enhancements” (TGE) accord-ing to Chilingarian et al. (2010). Hereafterin present publication the latter designationwill be used, mostly because of its convenientshortness, as well as the fact that the subjectof this paper is indeed connected with the dataobtained in a ground-based experiment.Presently it is generally assumed that theappearance of the flow of high energy elec-trons in thunderstorm time is due to the ef-fect of runaway electron breakdown—an ac-celeration of the charged particles in specific conditions when the energy acquired underthe influence of an external electric field bysome particle from the tail of their initial en-ergy distribution (a “runaway”) exceeds thelosses of its interaction with surrounding mat-ter. In turn, emission of gamma radiation inthunderstorm time is due to bremsstrahlungof accelerated electrons, as it was illustratedby many simulations, e. g. , of Torii et al.(2004); Chilingarian et al. (2011); Babich et al.(2013b); Luque et al. (2018). An excessiveneutron flux which has been detected in somemost energetic electric discharge events withinthunderclouds, such as observed by Tsuchiyaet al. (2012); Ishtiaq et al. (2016); Teruakiet al. (2017), may result from the photonu-clear reactions caused by gamma rays: Babichand Roussel-Dupr´e (2007); Chilingarian et al.(2012a); Babich et al. (2013a, 2014); Enotoet al. (2017); Bowers et al. (2017). Crucialrole in formation of TGE events observed inground-based experiments plays the lower re-gion of positive charge distribution in thunder-clouds which sometimes can appear below themain negative charge concentrated in the mid-dle of the cloud, and ensures acceleration ofelectrons in direction to the earth’s surface, asit was discussed in Khaerdinov et al. (2005);Chilingarian et al. (2015).A possibility of appropriate conditions toexist inside thunderclouds for realization ofthe runaway breakdown mechanism has beenpredicted already by Wilson (1925). Lateron, this idea was quantitatively developed ina number of theoretical studies, such as Gure-vich et al. (1992); Gurevich and Roussel-Dupr´e (1996); Lehtinen et al. (1999); Gure-vich and Zybin (2001). Key statement of thistheory is that a runaway electron when beingaccelerated by internal electric field of a thun-dercloud may produce another free electronsby its collisions with atmospheric atoms, andthe latter may accelerate in their turn produc-2ng the next generation of the charged parti-cles. As a result, a single seed electron putinside the region of the electric field with suf-ficiently high strength E > E c gives rise to awhole avalanche of high energy particles, con-sisting both of electrons and bremsstrahlunggamma ray quanta. The critical field E c isequal to 280 kV / m at the sea level and dimin-ishes with height proportionally to air density;this value is almost an order of magnitude be-low the threshold of the conventional dielec-tric breakdown in the air. As many rocket- andballoon-born experiments on direct soundingof thundercloud internals have shown, suchas e. g. those of Marshall et al. (1995a,b);Stolzenburg et al. (2007), generally the light-ning discharges in thunderclouds were initi-ated when the strength of the electric field wasclose to E c value, while the fields essentiallyabove E c were never detected. This experi-mental fact is an indirect evidence in favor ofthe runaway breakdown mechanism.Next obligatory condition for beginningof avalanche process is the presence of thecharged seed particles with su ffi ciently high, (cid:38) (0.1–1) MeV, initial energy inside the vol-ume with critical field. As such, the secondaryelectrons produced by cosmic rays in the at-mosphere and β -electrons stemmed from natu-ral radioactivity were suggested still by Wilson(1925). The need of high-energy (relativistic)seed particles stipulated the commonly usedterm for designation of this model—the rela-tivistic runaway electrons avalanche (RREA).As well, for realization of such process it isnecessary a su ffi ciently large geometrical di-mension of the field region which is limitedfrom below by characteristic size of develop-ing avalanche, and must be of at least a hun-dreds of meters order; this condition is gener-ally satisfied just in thunderclouds.As it was shown by a number of simulations,such as Dwyer (2003); Coleman and Dwyer (2006); Roussel-Dupr´e et al. (2008), energyspectrum of gamma rays from an RREA gen-erally is of an exponential form with the aver-age energy of about 7 MeV. The fact, that thespectra of gamma-emissions detected in thun-derclouds sometimes were consistent with thisresult is another adequacy confirmation of thesuggested theory: Torii et al. (2004); Tsuchiyaet al. (2007); Babich et al. (2010).Di ff erent type of the runaway particlesbased breakdown is the so called thermal, orcold breakdown process which is possible in avery high field, E (cid:38) · kV / m. In thun-derclouds, the fields of such strength can existonly transiently, and its distribution is limitedby closest vicinity of the streamer tips in de-veloping electric discharge leader. As a conse-quence, in contrast to RREA, in this case thedevelopment of electron avalanche takes placeat much smaller distances of a tens of centime-ters or a meters order scale. Another principaldi ff erence of the cold breakdown process fromRREA is the absence of the need in externalsource of seed particles, since any free elec-tron, including the “cold”, or thermal-energyones, can start avalanche acceleration in suchstrong fields. Later on, and under favorableconditions if a su ffi ciently high large-scaledfield is present in the cloud, electrons accel-erated due to the thermal process may play therole of seed particles for “common” RREA.The model of the cold runaway breakdownwas initially suggested by Gurevich (1961)and further on especially developed in appli-cation to thundercloud conditions in publi-cations of Dwyer (2004, 2005); Moss et al.(2006); Celestin and Pasko (2011). Multipleobservations of the short, micro- and millisec-ond long intensive bursts of gamma radiationwhich were immediately coinciding with themoments of lightning discharges, such as re-ported by Moore et al. (2001); Dwyer et al.(2004, 2005); Montany`a et al. (2014), can be3onsidered as an experimental prove of themodel of particle acceleration at the streamertips of developing lightning leaders.Essential modification added by Dwyer(2003) to the runaway breakdown model is therelativistic feedback e ff ect which consists ofa surplus amplification of hard emission dueto high energy quanta of bremsstrahlung ra-diation and to positrons born through the e ± -pair production mechanism. Inside thunder-cloud field the positron component acceleratesin opposite direction in relation to the elec-tron flux, and produces on its way ionizationelectrons which occur to be additional seedparticles for subsequent RREAs which gen-erate another positrons, etc . As a result, themultiplicity of both electrons and positronsgrows exponentially which leads to a signif-icant, up to an order of magnitude, increaseof overall gamma-emission from the dischargeregion comparatively to single RREA. As sug-gested by Dwyer (2008), it is the runawaydischarge process in combination with posi-tive relativistic feedback which lays in the ba-sis of the mostly fast and energetic phenom-ena of the high energy atmospheric physics—terrestrial gamma ray flashes (TGF), such asseen by Franz et al. (1990); Fishman et al.(1994); Smith et al. (2005); Grefenstette et al.(2008) in upper atmosphere, and of analogousdownward TGFs which have been observed ina number of ground-based experiments, suchas of Dwyer et al. (2012a); Tran et al. (2015);Hare et al. (2016); Wada et al. (2019b). Onthe other hand, an equilibrium discharge cur-rent arising because of steady generation ofRREAs amplified by positive feedback maycompensate continuous charge separation inthundercloud and ensure a comparatively pro-longed quasi-static state which reveals itselffor outer observer as a long lasting gamma rayglow or a TGE event.A comprehensive survey of modern progress in the theory of runaway electronsacceleration and its implication to the prob-lems of atmospheric electricity can be foundin the review of Dwyer et al. (2012b).An alternative mechanism which could beresponsible for detection of an additional fluxof high energy particles in thunderstorm times,and primarily for appearance of long-lastingTGE events, is the modification e ff ect of theenergy spectrum (MOS) of charged particlesbackground under the influence of thunder-cloud field. According to this model which hasbeen initially suggested in Chilingarian et al.(2012b, 2013a); Chilingarian (2014), by favor-able orientation of atmospheric electric fieldboth electrons and positrons originated fromcosmic rays interaction can acquire an addi-tional energy which leads to the rise of theirlifetime and attenuation length in atmosphere,and to corresponding intensity increase of de-tected radiation at the observation level. Withaccount to MOS e ff ect an appearance of the at-mospheric electricity connected phenomena ispossible even in those cases when the thunder-cloud field does not exceed the E c limit. As itwas illustrated in Cramer et al. (2017); Chilin-garian et al. (2017, 2018), detecting of TGEevents in combination with an intensive sim-ulation of particles interaction in atmosphericfields can be a mean to probe the structure ofelectric field distribution in thunderclouds.One more source of the excessive flux ofhigh energy charged particles and gamma raysin thunderstorm times may by the decay of ra-dioactive nuclei, primarily the radon Ra iso-tope and its daughter products whose concen-tration increases in near-earth atmosphere be-cause of precipitations which usually accom-pany the periods of thunderstorm activity. Asit is often stated, e. g. by Chilingarian (2018),such a temporary rise of the local radioactivebackground can be accountable for the mostprolonged (tens of minutes) and low-energy4below 1 MeV) part of TGE events.Thus, thanks to intensive e ff orts undertakenin the course of last decades it was found avariety of unexpected phenomena in such amanifold investigation field as the high-energyatmospheric physics, and a number of theo-retical models was created for their explana-tion. In spite of this progress the general pic-ture of atmospheric electric discharge, and pri-marily the processes which lay in the basis oflightning initiation, still remain not completelyclear. This circumstance necessitates further insitu observations and collection of as much aspossible new experimental data on radiationswhich accompany electric discharges in thun-derclouds.An appropriate place for such investigationsis Tien Shan Mountain Cosmic Ray Station ofLebedev Physical Institute. At multiple stud-ies executed here in 2000s a number of ob-servations were made of the intensive gammarays and high energy electrons bursts at thun-derstorm time, some of them being similar towhat was designated later on as TGE and TGFin literature: Chubenko et al. (2000, 2003,2009); Gurevich et al. (2004, 2009, 2011,2013). Since then a special complex of detec-tor facilities was created at Tien Shan stationwhich is especially aimed to experimental in-vestigations in the range of high energy atmo-spheric physics Gurevich et al. (2016, 2018).A key feature of this complex is its abilityto detect the soft gamma radiation and accel-erated charged particles immediately withinthunderclouds, at a small distance to develop-ment region of lightning discharges. For thispurpose a special high altitude detector pointwas created in the neighborhood of Tien Shanstation which is placed on the top of a moun-tain ridge, at an altitude of 3700 m above thesea level, and ∼
400 m above the average heightof the station. In the time of thunderstormsthis point frequently occurs to be deeply im- mersed within thundercloud, so detection ofvarious radiations is possible at a few tens andhundreds of meters order distance from the re-gion of their generation in close lightning dis-charges. Such disposition permits to avoid anysignificant influence on the part of the scatter-ing and absorption processes, and to expandthe range of detected radiations into the low-energy part of their spectrum. As a result,the lowest energy threshold in the Tien Shanexperiment now is of about (20–30) keV forgamma rays, and about (1–2) MeV for accel-erated electrons. This is an essential di ff er-ence from, correspondingly, a few hundreds ofkeV and tens of MeV detection thresholds ofgamma rays and electrons which were typicalfor the measurements made previously in otherground-based experiments, at a kilometer or-der distances from thundercloud.The subject of current publication is topresent the results obtained by observation atthe high altitude detector point of Tien Shanstation of the two close events observed in thetimes of thunderstorm activity: a TGE typeprolonged gamma ray glow which preceded aclose atmospheric discharge, and a hard radi-ation burst from a nearby lightning. In bothcases the distances between the measurementpoint of radiation intensity and the region oflightning development occurred of the (cid:46)
100 morder only.
2. Instrumentation
For registration of the soft gamma rayswhich accompany the lightning discharges atTien Shan station it is used a scintillation de-tector on the basis of a cylindrical (cid:31) ×
110 mm NaI(Tl) crystal coupled with a pho-tomultiplier tube (PMT). Since the amplitudeof any particular electric pulse at PMT outputis proportional to the energy of corresponding5amma ray quantum absorbed in scintillator(supposing the conversion linearity of scintil-lation flash into electric signal), an amplitudeanalysis of output pulses permits to recoverthe energy spectrum of detected radiation. Forthis purpose the signals from the detector out-put come to a set of analog pulse discrimi-nators with consecutively increasing operationthresholds. Every time when the amplitude ofscintillation light exceeds the given threshold,the corresponding discriminator channel gen-erates a single standard pulse signal which canby counted by a digital scaler scheme. Fixedlength of these signals is set to 10 µ s which iscompatible with the own decay time of usedscintillator.An absolute energy calibration of the wholeamplitude measurement system was madewith a set of radioactive gamma sources.Concrete values of the radiation detectionthresholds were somewhat di ff erent in thecourse of considered experiment. In the mea-surement season of the year 2018, when it wasdetected the TGE event which will be pre-sented below, there were 14 amplitude chan-nels with their thresholds distributed between ∼
20 keV and 2.5 MeV. At the time of the shortradiation burst which happened in summer2017 there were only 12 channels with an up-per energy threshold of 1 MeV.The registration probability of gamma ra-diation in dependence on its energy, i. e. thee ffi ciency of the considered detector was ob-tained through a Geant4 based simulation ofthe gamma ray interaction processes with in-ternal material of scintillator. In this calcula-tion the absorption probability ε of a gammaray quantum within the scintillator was de-fined for a number of the fixed values of ra-diation energy E γ . Since neither the probabil-ity of subsequent scintillation light emission,nor the e ffi ciency of PMT photocathode weretaken into account by this simulation, the val- GAMMA RAY ENERGY, E γ , keV D E T E C T I O N P R O B A B I L I T Y , ε ( E γ ) , % Figure 1: Probability of gamma quantum registration(detection e ffi ciency) of the gamma radiation detector. ues ε thus defined should be considered in thesense of a possible upper limit of detection ef-ficiency. These estimations are presented inFigure 1 in dependence on corresponding E γ values. The detection of electrons accelerated in theelectric field of thundercloud in Tien Shan ex-periment is realized by a coincidence telescopemade of a pair of large sized 1 × × charge particles detectors. Each detector con-sists of a set of the square shaped plates ofmolded plastic scintillator interlaced with thelight conducting optic fibers, as explained inBritvich et al. (2006). According to the detec-tor testing results presented in that publication,the detection probability of relativistic chargedparticles by a single scintillator of such a kindis of about 0.95–0.99, while that for the softgamma rays is quite negligible ( (cid:46) . µ s.In the telescope set-up used at the high al-titude point of Tien Shan station both scin-tillators were installed horizontally one abovethe other inside a wooden cabin with theroof thickness of about 1 g / cm , and with a40 g / cm thick layer of additional absorbingmaterial placed between them. Hence, the sig-nals registered from the upper scintillator ofsuch detector correspond to relativistic elec-trons with the energy of at least 2–3 MeV, andobservation of coincidence between the pulsesfrom the upper and lower scintillators meansnearly vertical passage of an electron with theenergy above 80–100 MeV, in dependence onthe slope of its trajectory. To avoid unwanted absorption of radiationson their way from generation region in thun-dercloud to detector, both the gamma rays andthe charged particles detectors used in the dis-cussed experiment were installed at a high al-titude point which is situated just on a moun-tain top, 400 m above the common level ofTien Shan station. Quite often during thun-derstorms this point occurs to be deep in theclouds, and there were some cases when thelightning was developing at a distance of abouta few tens of meters only from the detectors.Special measures were taken to ensure sta-ble operation of electronic hardware placedin such unordinary conditions. All necessarycomponents both of the detector and of thedata acquisition (DAQ) system installed therewere made as compact as possible to have min-imum length of all connecting wires. At thun-derstorm time any external cable lines werephysically disconnected from the high altitudepoint, so the powering of its equipment wasperformed completely from an in-build batterysource. This battery, the whole DAQ elec-tronics, the data registration computer, and the gamma ray detector were installed togetherwithin a same 0 . × . × Faraday cagewelded of a solid, 1 mm thick iron sheet, andby such a way that neither internal wire con-nection between them exceeded the length of(0.2–0.3) m. Two scintillation signals of thecharged particles telescope were connected tothis DAQ system by a pair of shielded, 0.5 mlong coaxial cables.Intensity registration of the digital pulse sig-nals in the high altitude point is realized bya compact DAQ system with reduced powerconsumption which is built on the basis of aSTM32 type microcontroller Shepetov et al.(2017). There are 17 separate signals con-nected to this system: 14 outputs of the gammadetector amplitude discriminator, the outputsof the upper and lower scintillator plates ofthe charged particles telescope, and the coinci-dence pulse between the latter two signals. Forall these signals the microcontroller driver pro-gram ensures the measurement of their count-ing rates simultaneously in two modes. Theseare the regular monitoring mode of the aver-age levels of signal intensity with periodicityof 1 s, and the detailed registration of the timehistories of signal development with two tem-poral resolutions of 160 µ s and 800 µ s.The high resolution time series of signal in-tensity registered by the microcontroller DAQsystem may be strictly bound to some specificevent which in the case of the considered ex-periment typically coincides with the momentof lightning discharge. For this purpose it isused a special control signal—the trigger. Thistrigger can be either generated by a hardwaresensor of the fast jump in the strength of lo-cal electric field in detector vicinity, or it canbe elaborated internally by the microcontrollerdriver program itself. In latter variant the pro-gram looks continuously after behaviour of allinput signals, and in the case if some logicalcondition is fulfilled ( e. g. if the current sum of7he pulse numbers stored over a specified timeperiod exceeds a predefined threshold limit) itfires an internal trigger signal which initiatesrecording of the next event. For the high al-titude detector point where any outside con-nections are strictly prohibited at thunderstormtime, it is the internal logical type of triggerwhich is applied for synchronization of the de-tected gamma ray and electron flux series withthe lightning moment.Regardless of synchronization type,the DAQ system operates always in pre-trigger / post-trigger registration mode: thecurrent multiplicity values for all input sig-nals measured with 160 µ s or 800 µ s timeresolution are kept continuously in cyclicbu ff er within microcontroller memory, sothat whenever any trigger comes, the wholehistory of signal development is available bothbefore and after its moment. In the time ofthe discussed experiment the duration of boththe pre- and post-trigger time series in everyregistered lightning event was accepted to beequal to 1 s. The data on the behaviour of the neutronflux intensity in thunderstorm times which willbe discussed further on were obtained at theneutron detector site of the Tien Shan station,400 m below the thunderstorm radiations de-tector system of the high altitude point. Thestraight line-of-view distance between the lat-ter and the neutron detectors is of about (1.5–2) km.There are two detector kinds for the neu-tron flux measurements at Tien Shan MountainStation. The first is a NM64 type neutron su-permonitor, such as described by Hatton andTomlinson (1968); Hatton (1971), which op-erates here continuously during many decadesZusmanovich et al. (2009). Generally, the neu-tron monitors of such type can be used for a high precision registration of the intensity ofenergetic ( (cid:38) (cid:46) (0.5-1)% only: Clem and Dorman (2000); Shibataet al. (2001); Abunin et al. (2011).Together with cosmic ray hadrons, the neu-tron monitor may detect high energy muonswhich possibility was particularly analyzed byShepetov et al. (2019). This e ff ect is due tothe high energy bremsstrahlung radiation emit-ted at muon passages, with subsequent neu-tron production in photonuclear reaction of re-sulting gamma rays, to direct nuclear interac-tions of the muonic cosmic ray component,and to the µ − -capture mechanism. As well, thephotonuclear interaction channel ensures de-tection of the high energy ( (cid:38)
10 MeV) gammarays, electrons and positrons by the moni-tor, thou with a probability ∼ (10–30) timesbelow the e ffi ciency of hadron registration:Tsuchiya et al. (2012); Babich and Roussel-Dupr´e (2007); Babich et al. (2013a, 2014);Teruaki et al. (2017); Babich (2019).Tien Shan neutron monitor consists of three2 × units each of which includes six big( (cid:31) × ) gas discharge counters. Thecounters are sensitive to low-energy (thermal)neutrons because of their special filling—theenriched BF gas. To ensure the fast energyloss by evaporation neutrons originating fromnuclear reactions down to thermal values, theneutron moderator layers of a light hydrogenreach material were included into monitor set-up. Registration of the shaped pulse signalsfrom the anode wires of these counters is real-ized by a STM32 microcontroller DAQ systemof the same type as what is used at the high8 − − NEUTRON KINETIC ENERGY, E n , eV D E T E C T I O N P R O BAB I L I T Y , ε ( E n ) , % LOW THRESH.MONITOR
Figure 2: Detection e ffi ciency of the NM64 neutron su-permonitor and of the low-threshold neutron detectorconsisting of 6 gas discharge neutron counters. altitude point. The periodicity of the neutronintensity measurements accepted at Tien Shanmonitor is one minute, which is a standard inthe world wide net of the cosmic ray intensitymonitoring.Second kind of neutron measurements atTien Shan station is fulfilled with the low-threshold detectors consisting of a set of“nude” neutron-sensitive counters which areused without any surrounding heavy target nei-ther moderator material. Because of such con-figuration these counters are sensitive mostlyto the flux of thermal neutrons born under theinfluence of cosmic rays in outer environmentaround the detector.The neutron registration e ffi ciency in the ex-periments of Tien Shan station was defined bya Geant4 based model simulation which tookinto account characteristic features both of theneutron detector internals, and of the outerenvironment typical for the station. In Fig-ure 2 it is presented the principal result of thesesimulations—the dependence of detection e ffi -ciency on the energy of incident neutron whichwas obtained for both detector types: for the NM64 neutron supermonitor and for the low-threshold neutron detector. Similar data on thissubject were reported by Clem and Dorman(2000) and Shibata et al. (2001). Temporal history of atmospheric dischargein the lightning events of Tien Shan experi-ment can be precisely traced by the records ofelectromagnetic emission which generally ac-companies development of the electric stream-ers in thundercloud. At the time of consid-ered experiment the signals of electromagneticemission were detected simultaneously in twofrequency ranges: MF / HF (0.1–10 MHz), andVLF (1.5–11.5 kHz). For this purpose a pairof corresponding antenna sets together withsubsequent receiver electronics is now in useat Tien Shan station. The analog output sig-nals of both receivers were operated by twoamplitude-to-digital converter (ADC) systems.The digitization was made with a 12-bit ac-curacy, and with periodicity of 0.16 µ s (in theMF / HF range) and 200 µ s (for the VLF signal).The total duration of the time series records forboth signals in detected events was of about(2.5–3) s.Along with waveform recording, analogoutput of the MF / HF receiver is used for elab-oration of the hardware lightning trigger. Forthis purpose the output signal comes, in par-allel with ADC, to an amplitude discrimina-tor which generates the trigger pulse each timewhen the detected amplitude of electromag-netic emission occurs above some predefinedthreshold. This trigger provokes recording ofthe next data series at both receivers of electro-magnetic radiation, as well as at other detectorsystems it is connected to.The strength of the local electric field inthunderstorm time was measured by a “field-mill” kind sensor, similar to what is described9n Boltek Lightning Detection Systems (2016).This sensor is installed at the main territory ofTien Shan Mountain Station, ∼ ∼
400 m below the high altitude detectorpoint. The strength of detected field is encodedin analog form by the voltage of output signalof the field sensor; so that any level above zeroat this output means the presence of a positiveelectric charge in the spatial region just overthe sensor, as well as possibility of electron ac-celeration in downward direction between themain negative charge region in the middle ofthundercloud and its lower positive charge dis-tribution area, in the manner of as how it wasdescribed by Chilingarian et al. (2011, 2015).At measurement time, the analog signal ofthe electric field sensor was digitized by a 12-bit ADC unit integrated into an STM32 micro-controller. The recording of these ADC datawas realized similarly to the case of particledetectors: there were two parallel datasets, themonitoring data of the slow field variation withone second periodicity, and the fast time se-ries of the field strength behaviour fixed witha 200 µ s resolution. The high resolution se-ries were strictly bound to the lightning triggergenerated by the discriminator of the electro-magnetic emission signal.The distance from the detector system to theregion of electric discharge in recorded eventscan be roughly estimated by the time delaybetween the lightning flash and the arrival ofthunder sound. For this purpose a microphonewas installed at the high altitude point whoseanalog signal was digitized with another ADCunit, quite in the same way as what was ap-plied for the electric sensor.
3. Experimental results
During the measurements made in thunder-storm time of the date of 12 August 2018 it wasdetected an extremely large rise of gamma ra-diation which had preceded a close lightningdischarge. The monitoring records of radia-tion intensity taken around this moment (Au-gust 12, 2018, 05:04–05:05 UT) are presentedin the plots of Figure 3. Here, the gamma rayintensity is expressed in the units of the signalpulse numbers N p registered every second withvarious energy thresholds of the high altitudegamma detector. As it was mentioned in Intro-duction, in literature similar events were com-monly designated as the thunderstorm groundenhancements—TGE. The distance from thedetector point to discharge region estimatedby the time delay of thunder sound relative tolightning flash in this event was of about (100–200) m.As it follows from Figure 3, at the timeof TGE event the relative excess of radiationintensity in all energy ranges was 5–6 timesabove its background level, and the total dura-tion of the TGE was of about 1.5 min. The riseof radiation intensity has started significantly before a close lightning discharge which hascaused a large negative jump of the local elec-tric field, and terminated just with beginningof the latter. The peak amplitude of the wholefield variation at the end of TGE event was ofabout (100–150) kV / m.In the high resolution data plots of the leftpanel in Figure 4 the time history of the light-ning discharge which has terminated the TGEcan be traced precisely by the waveforms ofits MF / HF and VLF electromagnetic emission.The time series of these signals were detectedwith high temporal resolution and strictly syn-chronized by the lightning trigger with the mo-10 k V / m ELECTRIC FIELD
GAMMA >20keV
GAMMA >100keV
GAMMA >200keV p u l s e c o un t i n g r a t e , N p / s GAMMA >500keV
GAMMA >700keV
GAMMA >1300keV : : : : : : : : Time of the date 12 Aug 2018, UT
GAMMA >2000keV
12 Aug 2018 -1500150 k V / m ELECTRIC FIELD
GAMMA >20keV
GAMMA >100keV
GAMMA >200keV p u l s e c o un t i n g r a t e , N p / s GAMMA >500keV
GAMMA >700keV
GAMMA >1300keV : : : : : : : : : : : : Time of the date 12 Aug 2018, UT
GAMMA >2000keV
12 Aug 2018
Figure 3: The data of the gamma ray intensity monitoring made during the day of August 12, 2018. The timescaleof the right frame is stretched around the moment of 05:05 UT. Ordinate axes are graduated in the number of detectorpulses N p obtained in a 1 s long time interval. -40040 A D C MF/HF -808 A D C VLF k V / m ELECTRIC FIELD
12 Aug 2018 05:04:53 -40040 A D C MF/HF
GAMMA >20keV
GAMMA >100keV
GAMMA >200keV c o un t i n g r a t e , p u l s e nu m b e r i n s GAMMA >400keV
GAMMA >500keV
GAMMA >700keV GAMMA >1300keV
300 200 100 0 100 200 300 400TIME RELATIVE TO TRIGGER, x0.001 s48
GAMMA >2000keV
12 Aug 2018 05:04:53
Figure 4: Left: the local field variation and synchronous waveforms of the MF / HF and VLF electromagnetic emission(expressed in arbitrary ADC codes) after the moment of TGE termination. Right: high resolution records of thegamma radiation counting rates around the lightning discharge which has terminated the TGE. Zero points of abscissaaxes correspond to the moment of the lightning trigger caused by the discharge. k V / m ELECTRIC FIELD
UPPER c o un t i n g r a t e , N p / s LOWER : : : : : : : : : : : : Time of the date 12 Aug 2018, UT
COINC
12 Aug 2018 -40040 A D C MF/HF -808 A D C VLF
UPPER p u l s e nu m b e r i n s LOWER
300 200 100 0 100 200 300 400TIME RELATIVE TO TRIGGER, x0.001 s48
COINC
12 Aug 2018 05:04:53
Figure 5: The counting rates in the channels of the charge particles telescope (see the text), together with the recordsof the electric field and of the MF / HF and VLF electromagnetic emission at the time of TGE event. Left plot—themonitoring data with a 1 s time resolution; right plot—the 800 µ s resolution series synchronized with the lightningdischarge trigger (which coincides with zero point of the time axis). ment of discharge beginning (see Instrumenta-tion section above). The values of the elec-tromagnetic signal amplitude in these plots areexpressed in the arbitrary ADC code units.In the electric field panels of Figures 3 and 4it is clearly seen that during all the time of theTGE event the field measured by the electricsensor was oscillating at a moderate positivelevel of about + (10–30) kV / m, which agreeswith acceleration of an electron flux in direc-tion to the earth surface. Seemingly, such in-dication can be explained by rather prolongedexistence of a positively charged region at thebottom of the thundercloud above the sensor.Thus, all the TGE time the sensor remainedbeing screened from the influence of the mainnegative charge layer of the cloud, and the gen-eral field configuration was appropriate for ac-celeration of electrons in downward direction.Just at beginning moment of the terminat-ing lightning discharge the polarity of detected field had quickly changed to opposite (i.e. neg-ative) and fell down to the level of − (120–130)kV / m. This jump agrees again with suddendissipation of the lower positive charge causedby lightning current, and with subsequent re-sponse to the main negatively charged layerof thundercloud from the side of the field sen-sor. Total duration of the TGE terminating dis-charge was of about (0.3–0.5) s only, but thecomplete relaxation of the electric field afterthe jump to its initial level in vicinity to zerohas taken an order of magnitude longer time.The high resolution records of the intensityof gamma radiation signal measured just be-fore and after the terminating discharge are il-lustrated by the right panel plots of Figure 4. Itis seen there a rather abrupt disappearance ofany gamma radiation signal just at the momentof final lightning. More on this subject followsbelow.Time series of the charged particles signal12 A D C MF/HF
GAMMA >20keV p u l s e nu m b e r i n s GAMMA >1300keV
UPPER (ELECTRON >2 MeV)
12 Aug 2018 05:04:53
Figure 6: The radiation counting rate records zoomed around the TGE termination moment. Time resolution is 160 µ s.Vertical bars correspond to statistical error of the intensity measurement. registered by the high altitude telescope detec-tor in the time of the TGE event are presentedin Figure 5. As it was explained above, the sig-nals of the upper scintillator layer of this tele-scope correspond to relativistic electrons withthe energy of a few MeV order. Accordingto the UPPER designated panels of Figure 5,the peak amplitude of the intensity increaseof such particles during the TGE is of ∼ (10–12) times as much as its usual background.The rise of the charged particles intensity ter-minates just before the moment of the finallightning discharge, simultaneously with TGEgamma radiation.In principle, the signal from the electroncomponent of the thunderstorm activity causedradiations could be imitated by the chargedproducts of gamma ray interaction, such asCompton scattering and the like. Relative ad-mixture caused by this e ff ect into sum signalof our plastic scintillator detector can be evalu-ated by comparison of the UPPER an LOWER labelled panels in Figure 5: while the formerdemonstrates an order of magnitude high riseat the maximum of TGE event, the peak am-plitude of the latter above its background was of about (50–100)% only. Because of a thickabsorber layer between these two scintillators,the lower one should be sensitive mostly, in-deed, to products of gamma ray interactionboth within the absorber and in outer environ-ment (the soil beneath, etc ), while any excessover its signal detected by the upper scintilla-tor corresponds to the pure deposit from theside of charged particles in thundercloud com-ing from above.Most significant is the null response in
CO-INC panels of Figure 5. Since, as it waspointed out in Instrumentation section, theprobability of gamma radiation detection bya single plastic scintillator is essentially be-low 1%, it remains only negligible probabil-ity for any particular gamma ray quantum tointeract twice in the upper and lower detec-tors of the telescope set-up with producing ofcharged particles and generation of scintilla-tion pulses in both layers within the same gatetime of (2–3) µ s. Hence, the detection of co-incidence pulses between these two layers, ifany, would be a passage sign of a high-energy, (cid:38) (80–100) MeV, charged particle, and the lackof such signals in Figure 5 means the absence13f energetic electrons acceleration in the timeof considered TGE event.The sharpness of the radiation decay at ter-mination moment of the TGE event is illus-trated once more by Figure 6. As it fol-lows from the plots presented here, any surplusemission above the usual background, both forthe gamma ray and electron components, hadextincted very quickly just at the beginning ofthe final discharge. Evidently, the characteris-tic time of this disappearance at any rate wasbelow 160 µ s length of a single interval of in-tensity measurement. Another kind of thunderstorm activityevents detected at Tien Shan Mountain Sta-tion is presented by a short time outburst ofthe intensity of gamma rays and acceleratedelectrons which occurred simultaneously witha close lightning discharge. As it is shownin Figure 7, at the moment of June 13, 2017,10:30:05 UT a sudden jump of field tensionbetween the limits from
E ≈ −
130 kV / m to E (cid:38) +
100 kV / m was accompanied by inten-sive electromagnetic emission from an atmo-spheric discharge which has started just at be-ginning of the field variation. According tothe delay of thunder sound, the discharge dis-tance from the high altitude detector point inthis event was below 100 m.In contrast to the TGE event consideredabove, from the high resolution time seriesof the gamma ray counting rates presented inFigure 8 it follows that the intensive radia-tion flash in present case just coincided withthe initiation moment of the electric discharge.The whole duration of the radiation increasewas of a sub-millisecond order or shorter: as itis seen in the right panel plot of Figure 8, mostpart of the flash goes in a single 160 µ s longinterval of the intensity measurements, and it does terminate completely up to the end of thenext interval.Together with gamma-radiation, syn-chronous outbursts both of low-energy( (cid:62) (cid:38)
100 MeV)electrons are seen in the plots of Figure 8.These data correspond to the
UPPER and
COINC channels of the charged particlestelescope detector.Relative amplitude of the excessive radi-ation peak calculated as ( I peak − I bckgr ) / I bckgr over the data of Figure 8 (where I bckgr is thebackground counting rate) varies in the limitsof (1500–2000)% for gamma rays, 1000% for (cid:62) ff er from theproperties of the former TGE event. On ac-count of an order of magnitude higher rela-tive amplitude, the presence of high energyelectrons, and its generally short-term timeprofile, the observed radiation excess can beattributed to a quite another class of the at-mospheric electricity phenomena. Evidently,this is a representative of short time intensivebursts of hard radiation coinciding with light-ning. As it was discussed in Introduction, sim-ilar events were multiple times reported byvarious experimental groups, and presently itis supposed that such transient bursts originatefrom stepped leaders which arise in the courseof lightning development. As it was pointedout in Dwyer et al. (2012b) and in literaturecited therein, in the basis of these phenomenait may be the “cold” runaway breakdown reac-tions which take place in vicinity to the tops ofthe electric discharge streamers in developinglightning leader.With application of streamer model the ori-gin of the burst event detected on June 13,14 : : : : : : Time of the date 13 Jun 2017, UT -1000100 k V / m ELECTRIC FIELD
13 Jun 2017 -50050 A D C MF/HF -808 A D C VLF k V / m ELECTRIC FIELD
13 Jun 2017 10:30:05
Figure 7: Left: the behaviour of local electric field around the time of the June 13, 2017, 10:30:05 UT radiation burstevent; the moment of the burst is marked with a vertical line. Right: the variation of electric field and the records ofelectromagnetic radiation from the atmospheric discharge. Zero point of abscissa axis in the right panel correspondsto the moment of discharge initiation notified by a lightning trigger signal. -40040 A D C RADIO (MF/HF)
GAMMA >25keV
GAMMA >100keV
GAMMA >200keV
GAMMA >300keV c o un t i n g r a t e , p u l s e nu m b e r i n s GAMMA >600keV GAMMA >1000keV ELECTRON >2 MeV
ELECTRON >100 MeV13 Jun 2017 10:30:05
GAMMA >25keV
GAMMA >100keV
GAMMA >200keV
GAMMA >300keV c o un t i n g r a t e , p u l s e nu m b e r i n s GAMMA >600keV GAMMA >1000keV ELECTRON >2 MeV
ELECTRON >100 MeV13 Jun 2017 10:30:05
Figure 8: High resolution time series of the gamma ray and charged particles counting rates at the time of a closeradiation burst. Time axis in the right frame is stretched around the moment of the lightning trigger. Time granularityof signal intensity measurements is 0.8 ms in the left plot, and 0.16 ms in the right, zero point of the time axescorresponds to initiation moment of the lightning. MF / HF and VLF labelled panelsof Figure 7, but none of them was accompa-nied by any detectable flash of hard radiation.The rarity of the burst observations even atclose lightnings may arise because of narrowangular distribution and fast absorption of the high energy emissions from electron-photonavalanche developing in thunderclouds.It should be stressed that because of a ratherprolonged dead time of the applied detectorswhich was of the order of a few microseconds,in the “trigger” (zero-point) time interval ofFigure 8 we have to deal most probably withan overlap of signals from many elementarystreamers which were developing successivelyin lightning leader, so all conclusions made be-low on energy spectra, intensity of particlesfluxes, etc relate mostly to average streamercharacteristics.Some more considerations on possible ori-gin of the June 13, 2017 event follow in theend of next section 4.1.
4. Discussion
The monitoring type data presented in Fig-ure 3 permit to calculate the momentary en-ergy spectra of detected gamma rays, each ofwhich corresponds to some particular periodof time preceding the moment of TGE termi-nation. A set of such partial distributions givesan opportunity to trace any possible evolutionof the integral spectrum N ( (cid:62) E γ ) of gammaradiation in the course of the TGE event de-velopment.A number of such energy spectra is pre-sented in the left frame plot of Figure 9. Bytheir calculation it was taken into account thetotal sensitive area of scintillator crystal usedin the gamma detector (570 cm ), as well as thedistribution of gamma ray detection e ffi ciencyfor di ff erent E γ from Figure 1.The amplitude of the leading points in allspectra of Figure 1 can be somewhat under-estimated because of attenuation of the softgamma rays within the 1 mm thick iron wallsof the shielding box which surrounds thegamma detector from outside; nevertheless, as16 GAMMA-RAY ENERGY, E γ , keV − N ( ≥ E γ ) , s − · c m − -1-10-20-30-40-60+1 TGE
12 Aug 2018 GAMMA-RAY ENERGY, E γ , keV − N ( ≥ E γ ) , s − · c m − BURSTBCKGR
13 Jun 2017
Figure 9: Left: integral spectra of TGE radiation. The numbers beside curves mean the time delay (in seconds) relativeto the final discharge moment. Right: the spectrum of the short radiation burst (triangles). Background spectra aremarked with circles. The smooth continuous lines correspond to analytic approximation of the experimental datapoints (see text).
17t can be calculated by subtracting the back-ground intensity from the intensity of corre-sponding spectra points in the left panel plotof this figure, just before the termination mo-ment of the TGE an absolute increase of thelow-energy radiation flux E γ (cid:38)
30 keV was ofabout (30–35) s − cm − order. In the energyrange of (100–300) keV the peak intensity ofthe TGE connected radiation excess occurs of(15–18) s − cm − , and around 1 MeV it was of ∼ (5–7) s − cm − only.As it is seen in the left panel plot of Fig-ure 9, generally the gamma ray spectrum hadretained untouched its shape during the wholeTGE event, and the only observable changeof its form was a gradual rise of the absoluteradiation intensity. Just after the final light-ning discharge this intensity falls down againto the level of the background spectrum (whichis shown by circles in Figure 9). Similar be-haviour of the gamma ray spectra detected dur-ing TGE times was mentioned in Chilingarianet al. (2019).The TGE radiation spectra in Figure 9have a rather complicated form but inthe range above a few hundreds of keVthey can be approximated by a function f ( E γ ) ∼ E − αγ · exp( − E γ /ε )); analogously to ashow it was done by Wada et al. (2019a) forthe gamma rays originating from an electron-photon avalanche, and with account to theCompton scattering process. In the presentcase the best fit parameters are α ≈ .
68 and ε ≈ . α falls well into therange of power indices reported for the TGEevents which were observed by Aragats group, e. g. in Chilingarian et al. (2014). As it is ex-plained in that publication, such power spectramay be accounted for by the MOS type pro-cess of particles interaction in thundercloudswith a comparatively small electric field which remains below the critical energy threshold ofthe large-scale RREA discharge. To the sameconclusion leads as well the value of ε param-eter above which occurs much below the meanenergy of RREA particles ( (cid:39) e. g. by Arab-shahi et al. (2015), in that case there doesnot exist any typical average energy of result-ing gamma emission because of strong depen-dency of avalanche development on particularsof specific configuration of the local electricfield in thundercloud. Nevertheless, it seemsvery unlikely that the field with any strengthnecessary for initiation of the “cold” dischargecould sustain some noticeable time in the air,so the MOS model remains to be consideredas an only probable mechanism for explana-tion of our prolonged radiation event.The energy spectrum of the gamma ray flashdetected at the moment of a close lightningdischarge on June 13, 2017 is presented inthe right panel plot of Figure 9. This spec-trum was calculated according to the summedpulse count values in three consecutive 160 µ slong intervals, the first of which coincideswith zero point of abscissa axis in the highresolution data series in Figure 8 ( i. e. onlythose signals participate in calculation whichhave come during a (cid:39) µ s long time pe-riod just after the lightning trigger). Unfor-tunately, at that time in the DAQ system ofthe gamma detector there were not anticipatedany data channels with registration thresholdhigh enough to analyze the signal from hard18amma rays, so the measured spectrum ter-minates at the point of 1 MeV. Nevertheless,it is seen in Figure 8 that at the energies of E γ (cid:38) (100 − ∼
50 s − cm − in the range of (100–300) keV,and of ∼
20 s − cm − at 1 MeV. For soft gammarays with E γ (cid:38)
30 keV an absolute excess ofthe burst radiation above the background oc-curs of the order of ∼
100 s − cm − .Besides, the spectrum of the short-timeburst event seems to be somewhat harder thenthat of the TGE: above a few hundreds of keVit generally corresponds to a power approx-imation of the ∼ E − . γ type (which is shownby a continuous sloped line in the right frameplot of Figure 9), instead of a fast exponen-tial fall down in the case of TGE. A similarpower shape of the integral radiation spectra,and with close value of its slope index hasbeen just detected in the energy range of (0.1–10) MeV in former measurements at Tien Shanstation, such as those presented in Gurevichet al. (2016). Di ff erential energy spectra of apower shape dI / dE γ ∼ E − αγ with α within thelimits of 2.0–2.5 were reported also among theresults of other experimental groups, e. g. byChilingarian et al. (2015, 2010, 2011), for anumber of gamma ray flashes detected at thetimes of lightning discharges. These data arealso compatible with our present result.Absolute deposit on the part of thunder-storm activity into the electron component ofdetected signals can be estimated immediatelyby the monitoring type records from Figure 5for the TGE event, and as a sum of pulsecounts in corresponding time series of Figure 8for the short time radiation burst. (In lattercase the numbers of the charged particles de-tector pulses detected in three 160 µ s intervals just after the lightning trigger were includedinto this summing, quite in the same manneras how it was done above by calculation of thegamma burst spectrum). In both cases the cor-responding background count rates were sub-tracted from the intensity of thunderstorm con-nected signals, and the remaining di ff erenceswere normalized to the 1 m sensitive area ofthe charge particles detector, and to duration ofthe pulse measurements period ( i. e. to 480 µ sin the case of the short radiation burst). Theresults of this procedure are of about F e (cid:39) − cm − and F e (cid:39) − cm − for the electronfluxes which were detected at the moment ofthe short radiation burst with energy thresh-olds, correspondingly, of E e (cid:38) E e (cid:38)
100 MeV. As for the TGE event, the peakflux of low-energy electrons detected duringthe last second just before its termination wasat a comparable level of 1.3 s − cm − .The electron flux estimate for the June 13,2017 burst event permits to make an additionalconclusion on its possible origin. Based on theabove F e values, accepting ∼
100 m order dis-tance to the source of detected emission, andignoring attenuation of the electron flux on itsway to detector one can deduce a lower pos-sible limit for the total electrons number as N e ∼ (10 − ), in supposition of the outersize of emitting region of about (30 − N e is in reason-able agreement with typical multiplicities ofelectron production by stepped leaders whichwere reported by many experimental groupsand can be found e. g. in the review of Dwyeret al. (2012b). This result confirms once moreour initial assumption on a stepped leader pro-cess in close lightning as a source of observedhard radiation in this case. The question on presence of the positroncomponent among radiations generated in19 GAMMA-RAY ENERGY, E γ , keV − − − ∆ N / ∆ E γ , k e V − · s − · c m − TGEBURST
Figure 10: Di ff erential energy spectra of gamma radiation detected at the time of the TGE event (below, squares) andof the short-time burst (above, triangles). Background is subtracted. The thick continuous curves indicate the depositfrom the annihilation line. thundercloud takes a special place by ex-perimental investigations in the field of at-mospheric electricity. Possible source ofpositrons at thunderstorm time can be ei-ther electromagnetic interaction of acceleratedrunaway particles, as suggested by Gurevichet al. (2000), or β + decay of short living ra-dioactive nuclei which arise under influenceof gamma rays from the developing electron-photon avalanche due to the mechanism ofphotonuclear interaction, as it was discussed e.g. in Enoto et al. (2017); Babich (2019).Hence, investigation of the energy and tem-poral characteristics of positron flux helps toshed light on various features of atmosphericelectric discharge.A convenient marker of positron produc-tion is observation of the e ± annihilation lineat 511 keV in the spectrum of gamma emis-sion. To disclose most prominently the pres- ence of this line in the events consideredin present publication the di ff erential energyspectra dI / dE γ were built on the basis of in-tegral data from Figure 9. For this purpose itwas used the next to last ( -1s labelled) integralspectrum on the left hand plot of this figure,which has just preceded the terminating dis-charge at the time of TGE event, and the spec-trum of the gamma ray burst in the right panel.As before, the di ff erential spectra dI / dE γ ineach point were normalized to the total sen-sitive area of the scintillation detector crystal,and to the e ffi ciency of gamma radiation de-tection, as the latter is given by Figure 1 forcorresponding E γ . Together with normaliza-tion, the levels of the background radiation in-tensity (those shown with circles in Figure 9)were subtracted from both spectra. The di ff er-ential energy spectra thus calculated are pre-sented in Figure 10. In such form these spectra20gree rather well, both by their shape and ab-solute intensity, with the di ff erential spectrumof TGE emission reported by Tsuchiya et al.(2011).As it follows from Figure 10, a statisticallysignificant peak can be found indeed withinthe energy range of (400–600) keV in gammaray spectra of both the TGE event and of theshort-time radiation burst. An excess of bothenhancements over the extrapolated levels of“regular” power spectrum shown with dottedlines is of about (cid:39) − cm − for the TGE, andof (cid:39)
10 s − cm − in the case of the burst (by lat-ter estimations the width of the energy spec-trum bins accepted at the measurements time,correspondingly that of 100 keV and 150 keV,was taken into account). If to interpret thepeaks as originated from annihilation line,these two values correspond to an absolutecontribution of the 511 keV gamma rays emit-ted due to e ± reactions within a (50–100) mneighbourhood of discharge region. Since itcannot be excluded an admixture of other ra-diation sources into this spectrum range bothestimates should better be meant in the senseof an upper possible limit.Besides the supposed annihilation peak, thedi ff erential spectrum detected at the time ofthe TGE event demonstrates another irregular-ities and amplitude enhancements around theenergy ranges of about ∼
300 keV and (1500–2000) keV. Similar features were reported inthe works of Bogomolov et al. (2015) and Bo-gomolov et al. (2016) which were aimed toprecision study of radiation spectra of TGEsobserved, correspondingly, at the ground level(in Moscow region), and at Aragats mountain.In these publications it is claimed that such de-viations of TGE spectra from uniform powerbehaviour correspond to the gamma ray linesoriginating from radioactive decays of
Rnnuclei and their daughter products. Since therewas a many hours long rainy period during the beginning of the day of August 12, 2018which has immediately preceded the TGE, andthe radon concentration in the near-earth at-mosphere is known to increase considerably atprecipitation time such explanation seems tobe quite plausible in our case.
Together with the data on gamma radiation,the measurements of the intensity of neutronbackground in thunderstorm time are availableat Tien Shan station as well. As it was ex-plained in Instrumentation section, two typesof neutron detectors are used for this purposein Tien Shan experiments: the neutron moni-tor for detection of evaporation neutrons pro-duced in interaction of high energy cosmic rayparticles, and the low-threshold detector forregistration of the environmental backgroundof thermal neutrons. In particular, the count-ing rates of neutron signals were registered atthe time of the TGE event considered above,and the results of these measurements are pre-sented in Figure 11. Before to be plotted herethe original neutron counts which have beenrecorded continuously every minute at both in-stallations were corrected to variation of at-mospheric pressure, and the moving averagefilter algorithm was applied to them to stressmore distinctly any systematic intensity varia-tion against the background of random fluctu-ations. The length of the filter kernel acceptedin latter procedure was equivalent to a 3 minlong dataset, which is comparable with totalduration of the TGE event.As it is seen in the left panel of Figure 11,in spite of a rather rough time resolution ofneutron data a prominent intensity excess doesexist in the record of the neutron monitor sig-nals made in the day of August 12, 2018, withits position being superimposed on duration ofthe observed TGE event. The relative ampli-21ude of this transient increase above the back-ground is of about 2%, while its highest peakvalue just precedes the negative jump of thelocal electric field which was detected at themoment of TGE termination. Just after TGE a(0.5–1)% deep depression is seen in the moni-tor intensity record. In contrary, simultaneousdata on the thermal neutron background pre-sented in the right plot window of Figure 11do not demonstrate any irregularities in TGEtime which could exceed the level of usual sta-tistical fluctuations.An absolute amplitude of the peak excess ofthe neutron counting rate above its usual back-ground R at the time of TGE occurs to be ofabout (1300–1400) min − , or, with account ofthe sum sensitive area of the neutron monitor(18 m ), R ≈ − s − cm − . Seemingly, thisestimation can give a hint on the origin of thispeak.According to Babich (2019), presently it isbelieved that the most probable source of neu-tron generation at thunderstorm times is con-nected with photonuclear interaction of thehard gamma rays with E γ (cid:38) e. g. in atmosphere, or inthe soil around the monitor, an excessive sig-nal count in the latter must be caused by theflux of low-energy (thermal) neutron compo-nent, since the neutrons born in photonuclearreactions loose their initial MeV-order energyrather quickly on their way to the detector site.With account of the ∼
1% registration proba-bility of thermal neutrons by NM64 type su-permonitor (see Figure 2), an agreement withthe above R value can be achieved only if theneutron flux at the time of TGE was of about I n ∼ .
01 s − cm − . This value is an order ofmagnitude above the typical background levelregistered by the low-threshold neutron detec-tor ( I bckgr ≈ .
002 s − cm − ), and consequentlysuch an excess should be prominently seen in the counting rate record of the latter, but ac-cording to the right plot of Figure 11 this isnot the case. Hence, the supposition on exter-nal neutron generation in the space outside themonitor fails.Alternatively, it is possible to suppose af-ter Tsuchiya (2014) the origin of neutronsfrom photonuclear reactions which take placenot outside but immediately within the mon-itor volume. In this variant the intensityof the E γ (cid:38) I γ ∼ . − cm − , with account of R value and geometrical area of the monitor, aswell as of the energy threshold and e ffi ciencyof neutron production by gamma ray quantagiven e. g. in Tsuchiya et al. (2012). Atthe same time, an extrapolation into this en-ergy range of the f ( E γ ) approximation func-tion which was introduced by discussion ofthe left panel plot in Figure 9 results in esti-mation of (cid:39) − cm − only, i. e. it is twoorders of magnitude below the necessary I γ value, even in the case of the most intensivespectrum which has immediately preceded theterminating discharge.Hence, the photonuclear interaction can notexplain the increase of the neutron monitorcounting rate which was observed at the timeof the considered TGE event.Another possible mechanism of neutrongeneration just within the monitor is connectedwith the muonic component of cosmic rays,as it was discussed by Muraki et al. (2004);Alexeenko et al. (2004); Dorman and Dorman(2005), and primarily with the process of µ − capture by atomic nuclei. According to the re-sults of a Geant4 simulation made by Shepetovet al. (2019), the peak of the average neutronproduction multiplicity in monitor because ofthis e ff ect is of about ¯ ν (cid:39) (0 . − .
1) neu-trons per an incident muon, and this maxi-mum is reached in the muon energy range of ∼ (100–500) MeV. With such ¯ ν , an additional22 : : : : : : : : : Time of the date 12 Aug 2018, UT N e u t r o n c o un t i n g r a t e , m i n − NEUTRON MONITOR E l e c t r i c f i e l d , k V / m : : : : : : : : : Time of the date 12 Aug 2018, UT N e u t r o n c o un t i n g r a t e , m i n − LOW THRESH. E l e c t r i c f i e l d , k V / m Figure 11: Time series of the counting rate of neutron signals in the Tien Shan neutron monitor, and in the low-threshold neutron detector at the time of the August 12, 2018 TGE event. Upper curve in the plots indicates simulta-neous measurements of the local electric field. ∼ − cm − order flux of negative muonsis quite su ffi cient to explain the observed am-plitude R of the neutron intensity increase.The latter estimation is a small value in com-parison with the total background flux of the (cid:38)
100 MeV muons at Tien Shan station (whichis of about (cid:39) − cm − according to the inte-gral energy spectra presented in Shepetov et al.(2019)), and such variation can be convenient-ly accounted for by modulation influence ofthe local electric field. Indeed, as it is seen inFigure 11, in the time of the neutron intensityincrease the near-earth field sensor detected apositive field of the order of + (25–30) kV / m.As it has been just noticed above by discus-sion of the TGE event, this field may be ac-counted for acceleration of negatively chargedparticles, both electrons and muons, in down-ward direction to the earth surface. In such acase the growth of the neutron counting rate atthe TGE time can be explained by favorabledisposition of the di ff erent domains of elec-tric field just above the neutron monitor, quiteanalogously to situation with the gamma rayand electron TGE components. Then, a deepgap in the record of the neutron monitor count-ing rate after TGE may be a consequence of µ − deceleration caused by a strong opposed fieldwhich has appeared after its reversal in the mo-ment of terminating discharge. Such obser-vations demonstrate a very fine sensitivity ofmuon signal as a messenger on the structureof electric fields in thundercloud which e ff ectcan be used in further investigations of thun-derstorm phenomena.
5. Conclusion
Two di ff erent kinds of the hard radiationflashes connected with thunderstorm activitywere detected at Tien Shan Mountain CosmicRay Station near the moments of close light-ning discharges: a prolonged TGE type eventpreceding a lightning, and a short-time radi-ation burst emitted just in the moment of alightning discharge. The measurements of ra-diation fluxes were made in immediate vicin-ity ( (cid:46)
100 m) to spatial region of their genera-tion in thunderclouds. Both events di ff er sig-nificantly by their general time profiles, abso-lute intensity, and energy of emitted radiation.In spite of this di ff erence the signs were no-ticed in both cases among the energy spectraof detected gamma radiation of the presence of23 able 1: Peak intensity (in the units of s − cm − ) of the thunderstorm activity connected emissions which weredetected at the time of two considered atmospheric discharge events. γ (cid:38)
30 keV γ (cid:38)
300 keV γ (cid:38) e − (cid:38) e − (cid:38)
100 MeV µ − TGE 30 15 5 (cid:54) (cid:38) (cid:54) ∼ ∼ Acknowledgement
This work was supported by “AppliedSpace Research” program, the project
References
A. A. Abunin, E. V. Pletnikov, A. L. Shchep-etov, and V. G. Yanke. E ffi ciency of detec-tion for neutron detectors with di ff erent geometries. Bull. Russ. Acad. Sci. Phys. , 75(6):866–868, 2011.doi:10.3103 / S1062873811060037.V. V Alexeenko, N. S. Khaerdinov, and V. B.Petkov. Transient variations of secondary cos-mic rays due to atmospheric electric field and evi-dence for pre-lightning particle acceleration.
Phys.Lett. A , 301:299–306, 8 2004. doi:10.1016 / S0375-9601(02)00981-7.S. Arabshahi, J. R. Dwyer, E. S. Cramer, J. E. Grove,C. Gwon, J. D. Hill, and et al. The energy spec-trum of x-rays from rocket-triggered lightning.
J.Geophys. Res. Atm. , 120:10951–10963, 10 2015.doi:10.1002 / Physics-Uspekhi , 62(10):976–999, oct 2019.doi:10.3367 / ufne.2018.12.038501.L. P. Babich and R. A. Roussel-Dupr´e. Origin ofneutron flux increases observed in correlation withlightning. J. Geophys. Res. , 112:D13303, 2007.doi:10.1134 / S0021364006180020.L. P. Babich, E. I. Bochkov, E. N. Donskoi, and I. M.Kutsyk. Source of prolonged bursts of high-energygamma rays detected in thunderstorm atmosphere inJapan at the coastal area of the Sea of Japan and onhigh mountaintop.
J. Geophys. Res. Space Phys. ,115:A09317, 9 2010. doi:10.1029 / Jetp Lett. , 97:291–296, 2013a.doi:10.1134 / S0021364013060027.L. P. Babich, I. M. Kutsyk, E. N. Donskoy, and J. R.Dwyer. Analysis of the experiment on registration ofX-rays from the stepped leader of a cloud-to-ground ightning discharge. J. Geophys. Res.: Space Phys. ,118:2573–2582, 5 2013b. doi:10.1002 / jgra.50236.L. P. Babich, E. I. Bochkov, I. M. Kutsyk, andH. K. Rassoul. Analysis of fundamental inter-actions capable of producing neutrons in thun-derstorms. Phys. Rev. , D89(9):093010, 2014.doi:10.1103 / PhysRevD.89.093010.V. Bogomolov, A. Chilingaryan, G. Garipov, and et al.Results of TGE study in 0.03–10 MeV energy rangein ground experiments near Moscow and Aragats.In
Proc. of Int. Symp. TEPA 2016: Thunderstormsand Elementary Particle Acceleration , page 50, Nor-Amberd, Armenia, 2016.V. V. Bogomolov, S. I. Svertilov, I. A. Maximov, andet al. Study of TGEs and gamma-flashes from thun-derstorms in 20–3000 keV energy range with SINPMSU gamma-ray spectrometers. In
Proc. of Int.Symp. TEPA 2015: Thunderstorms and ElementaryParticle Acceleration , page 120, Nor-Amberd, Ar-menia, 2015.Boltek Lightning Detection Systems. EFM-100 elec-tric field mill, 2016. .G. S. Bowers, D. M. Smith, G. F. Martinez-McKinney,M. Kamogawa, and et al. Gamma ray signa-tures of neutrons from a terrestrial gamma rayflash.
Geophys. Res. Lett. , 44:10063–10070, 8 2017.doi:10.1002 / Nucl. Instrum. Methods A , 564(1):225–234, 8 2006.doi:10.1016 / j.nima.2006.03.042.S. Celestin and V. P. Pasko. Energy and fluxesof thermal runaway electrons produced by ex-ponential growth of streamers during the step-ping of lightning leaders and in transient luminousevents. J. Geophys. Res. Atm. , 116:A03315, 3 2011.doi:10.1029 / J.Atmos. Sol.-Terr. Phys , 107:68–76, 11 2014.doi:10.1016 / j.jastp.2013.11.004.A. Chilingarian. Long lasting low energy thunderstormground enhancements and possible Rn-222 daughterisotopes contamination. Phys. Rev. D , 98:022007, 72018. doi:10.1103 / PhysRevD.98.022007.A. Chilingarian, A. Daryan, K. Arakelyan, and etal. Ground-based observations of thunderstorm-correlated fluxes of high-energy electrons, gammarays, and neutrons.
Phys. Rev. D , 82(8):043009, 8 2010. doi:10.1103 / PhysRevD.82.043009.A. Chilingarian, G. Hovsepyan, and A. Hovhan-nisyan. Particle bursts from thunderclouds:Natural particle accelerators above our heads.
Phys. Rev. D , 83(3):062001–11, 3 2011.doi:10.1103 / PhysRevD.83.062001.A. Chilingarian, N. Bostanjyan, and L. Vanyan. Neutronbursts associated with thunderstorms.
Phys. Rev. D ,85:085017, 2012a.A. Chilingarian, B. Mailyan, and L. Vanyan.Recovering of the energy spectra of elec-trons and gamma rays coming from thunder-clouds.
Atm. Res. , 114–115:1–16, 5 2012b.doi:0.1016 / j.atmosres.2012.05.008.A. Chilingarian, G. Hovsepyan, and L. Kozliner. Thun-derstorm ground enhancements: Gamma ray di ff er-ential energy spectra. Phys. Rev. D , 88:073001, 102013a. doi:10.1103 / PhysRevD.88.073001.A. Chilingarian, T. Karapetyan, and L. Melkumyan.Statistical analysis of the thunderstorm ground en-hancements (TGEs) detected on Mt. Aragats.
Ad-vances in Space Research , 52:1178–1192, 9 2013b.doi:10.1016 / j.asr.2013.06.004.A. Chilingarian, G. Hovsepyan, and L. Vanyan. On theorigin of the particle fluxes from the thunderclouds:Energy spectra analysis. Europ. Phys. Lett. , 106:59001, 2014. doi:10.1209 / / / J.Geophys. Res.: Space Phys. , 120:1–9, 2015.doi:10.1002 / Sci. Rep. , 7(1):1371–1381, 5 2017. doi:10.1038 / s41598-017-01288-0.A. Chilingarian, G. Hovsepyan, H. Mkrtchyan,S. Soghomonyan, S. Chilingaryan, B. Zazyan, andM. Zelenyy. Structures of the intracloud electricfield supporting origin of long-lasting thunderstormground enhancements. Phys. Rev. D , 98:082001, 82018. doi:10.1103 / PhysRevD.98.082001.A. Chilingarian, H. Mkrtchyan, G. Karapetyan,S. Chilingaryan, B. Sargsyan, and A. Arestakesyan.Catalog of 2017 thunderstorm ground enhancement(TGE) events observed on Aragats.
Sci. Rep. , 9:6253,4 2019. doi:10.1038 / s41598-019-42786-7.A. P. Chubenko, V. P. Antonova, S. V. Kryukov, V. V.Piscal, M. O. Ptitsyn, A. L. Shepetov, L. I. Vil-danova, K. P. Zybin, and A. V. Gurevich. Intensive -ray emission bursts during thunderstorms. Phys.Lett. A , 275:90–100, 1 2000. doi:10.1016 / S0375-9601(00)00502-8.A. P. Chubenko, I. V. Amurina, V. P. Antonova, S. V.Kryukov, K. M. Mukhashev, R. A. Nam, N. M. Nes-terova, V. V. Oskomov, V. V. Piscal, M. O. Ptit-syn, T. Kh. Sadykov, A. L. Shepetov, L. I. Vil-danova, K. P. Zybin, and A. V. Gurevich. E ff ectivegrowth of a number of cosmic ray electrons insidethundercloud. Phys. Lett. A , 309:90–102, 1 2003.doi:10.1016 / S0375-9601(03)00062-8.A. P. Chubenko, A. N. Karashtin, V. A. Ryabov, andet al. Energy spectrum of lightning gamma emis-sion.
Phys. Lett. A , 373(39):2953–2958, 6 2009.doi:10.1016 / j.physleta.2009.06.031.J. M. Clem and L. I. Dorman. Neutron monitor responsefunctions. Space Sci. Rev. , 93:335–359, 2000. ISSN0038–6308. doi:10.1023 / A:1026508915269.L. M. Coleman and J. R. Dwyer. Propagation speed ofrunaway electron avalanches.
Geophys. Res. Lett. , 33(11):L11810, 2006. doi:10.1029 / J.Geophys. Res. Atm. , 122:4763–4772, 4 2017.doi:10.1002 / Advances in Space Research , 35:476–483, 6 2005.doi:10.1016 / j.asr.2005.01.013.J. R. Dwyer. A fundamental limit on electric fieldsin air. Geophys. Res. Lett. , 30:2055, 10 2003.doi:10.1029 / Geophys. Res. Lett. , 31(12):L12102, 2004.doi:10.1029 / Geophys. Res. Lett. , 32:L20808, 122005. doi:10.1029 / J. Geophys. Res. Atm. , 113:D10103, 12008. doi:10.1029 / Geophys. Res. Lett. , 31(3):L05118, 2004.doi:10.1029 / Geophys. Res. Lett. , 32:L01803, 12 2005.doi:10.1029 / J. Geophys. Res. , 117:A10303, 10 2012a.doi:10.1029 / Space Science Re-views , 173(1):133–196, 2012b. doi:10.1007 / s11214-012-9894-0.K. Eack, W. Beasley, W. Rust, T. Marshall, andM. Stolzenburg. Initial results from simultaneousobservation of X-rays and electric fields in a thun-derstorm. J. Geophys. Res. , 101(D23):29637–29640,1996. doi:10.1029 / Geophys. Res. Lett. , 27(2):185–188, 1 2000. doi:10.1029 / Nature ,551:481–484, 11 2017. doi:10.1038 / nature24630.G. J. Fishman, P. N. Bhat, R. Malozzi, and et al.Discovery of intense gamma ray flashes of atmo-spheric origin. Science , 264(5163):1313–1316, 51994. doi:10.1126 / science.264.5163.1313.R. C. Franz, R. J. Nemzek, and J. R. Winckler. Tele-vision image of a large upward electrical dischargeabove a thunderstorm system. Science , 249:48–51,1990.B. W. Grefenstette, D. M. Smith, J. R. Dwyer, and G. J.Fishman. Time evolution of terrestrial gamma rayflashes.
Geophys. Res. Lett. , 35:L06802, 2008.A. V. Gurevich. On the theory of runaway electrons.
Soviet Physics JETP , 12(5):904–912, 5 1961.A. V. Gurevich and R. Roussel-Dupr´e. On run-away breakdown and upward propagating dis-charges.
J. Geophys. Res. , 101(A2):2297–2311,1996. doi:10.1029 / Physics-Uspekhi , 44(11):1119–1140, 112001. doi:10.1070 / PU2001v044n11ABEH000939.A. V. Gurevich, G. A. Milikh, and R. Roussel-Dupr´e.Runaway electrons mechanism of the air break-down and preconditioning during thunderstorm.
Phys. Lett. A , 165:463, 1992. doi:10.1016 / . P. Zybin. Generation of electronpositron pairs inrunaway breakdown. Phys. Lett. A , 275:101–108, 12000. doi:10.1016 / S0375-9601(00)00558-2.A. V. Gurevich, A. N. Karashtin, A. P. Chubenko,L. M. Duncan, V. A. Ryabov, A. L. Shepetov, V. P.Antonova, S. V. Kryukov, V. V. Piscal, M. O. Ptit-syn, Yu. V. Shlyugaev, and K. P. Zybin. Experi-mental evidence of giant electron-gamma bursts gen-erated by extensive atmospheric showers in thun-derclouds.
Phys. Lett. A , 325:389–402, 5 2004.doi:10.1016 / j.physleta.2004.03.074.A. V. Gurevich, G. G. Mitko, V. P. Antonova, and et al.An intracloud discharge caused by extensive atmo-spheric shower. Phys. Lett. A , 373(39):3550–3553, 92009. doi:10.1016 / j.physleta.2009.07.085.A. V. Gurevich, A. P. Chubenko, A. N. Karashtin, andet al. Gamma-ray emission from thunderstorm dis-charges. Phys. Lett. A , 375(15):1619–1625, 2011.doi:10.1016 / j.physleta.2011.03.005.A. V. Gurevich, V. P. Antonova, A. P. Chubenko,A. N. Karashtin, and et al. Correlation ofradio and gamma emissions in lightning initia-tion. Phys. Rev. Lett. , 111:165001, 10 2013.doi:10.1103 / PhysRevLett.108.125001.A. V. Gurevich, A. P. Chubenko, A. N. Karashtin, andet al. Observations of high-energy radiation dur-ing thunderstorms at Tien-Shan.
Phys. Rev. D , 94:023003–9, 2016. doi:10.1103 / PhysRevD.94.023003.A. V. Gurevich, A. P. Chubenko, A. N. Karashtin, and etal. Simultaneous observation of lightning emission indi ff erent wave ranges of electromagnetic spectrum inTien Shan mountains. Atmos. Res. , 211:73–84, 2018.doi:10.1016 / j.atmosres.2018.04.018.B. M. Hare, M. A. Uman, J. R. Dwyer, and et al.Ground-level observation of a terrestrial gamma rayflash initiated by a triggered lightning. J. Geo-phys. Res. Space Phys. , 121(11):6511–6533, 2016.doi:10.1029 / Progress in elementaryparticle and cosmic ray physics , volume 10, pages 3–97. North-Holland Publishing Company, 1971. ISBN9780720414103.C. J. Hatton and E. V. Tomlinson. The lifetime of neu-trons in cosmic-ray neutron monitors.
Il Nuovo Ci-mento , 53:63–72, 1 1968. doi:10.1007 / BF02710961.P. M. Ishtiaq, S. Mufti, M. A. Darzi, T. A. Mir, andG. N. Shah. Observation of 2.45 MeV neutrons cor-related with natural atmospheric lightning dischargesby Lead-Free Gulmarg Neutron Monitor.
J. Geo-phys. Res. Atm. , 121(2):692–703, 2016. ISSN 2169- 8996. doi:10.1002 / Nature Communication ,6:7845, 8 2015. doi:10.1038 / ncomms8845.N. S. Khaerdinov, A. S. Lidvansky, and V. B. Petkov.Cosmic rays and the electric field of thunderclouds:Evidence for acceleration of particles (runaway elec-trons). Atmospheric Res. , 76(1-4):346–354, 7-82005. doi:10.1016 / j.atmosres.2004.11.012.P. Kochkin, A. P. J. van Deursen, M. Marisaldi,A. Ursi, A. I. de Boer, M. Bardet, C. Allasia, J.-F. Boissin, F. Flouren, and N. Østgaard. In-flightobservation of gamma ray glows by ILDAS. J.Geophys. Res. Atm. , 122:12801–12811, 12 2017.doi:10.1002 / J. Geophys. Res. Atm. , 122:1–11, 9 2017.doi:10.1002 / J. Geophys. Res. , 104(A11):24699–24712, 1999. doi:10.1029 / Geosci-entific Model Development , 11:4515–4535, 11 2018.doi:10.5194 / gmd-11-4515-2018.T. Marshall, M. McCarthy, and W. Rust. Electricfield magnitudes and lightning initiation in thunder-storms. J. Geophys. Res. , D100:7097–7103, 1995a.doi:10.1029 / J. Geophys. Res. , D100:20815–20828,1995b. doi:10.1029 / Geophys. Res. Lett. ,12:393–396, 1985. doi:10.1029 / GL012i006p00393.J. Montany`a, F. Fabr´o, O. van der Velde, D. Romero,G. Sol`a, J. A. Hermoso, S. Soula, E. R. Williams,and N. Pineda. Registration of X-rays at 2500 m al-titude in association with lightning flashes and thun-derstorms.
J. Geophys. Res.: Atm. , 119:1492–1503,11 2014. doi:10.1002 / . B. Moore, K. B. Eack, G. D. Aulich, and W. Ri-son. Energetic radiation associated with lightningstepped-leaders. Geophys. Res. Lett. , 28:2141–2144,11 2001. doi:10.1029 / J.Geophys. Res. Space Phys. , 111:A02307, 2 2006.doi:10.1029 / ff ects of atmospheric elec-tric fields on cosmic rays. Phys. Rev. D , 69:123010–123022, 6 2004. doi:10.1103 / PhysRevD.69.123010.N. Østgaard, H. J. Christian, J. E. Grove, D. Sar-ria, A. Mezentsev, P. Kochkin, and et al. Gammaray glow observations at 20-km altitude.
J.Geophys. Res. Atm. , 124:7236–7254, 7 2019.doi:10.1029 / Geophys. Res. Lett. , 8:1176–1179,11 1981. doi:10.1029 / GL008i011p01176.R. Roussel-Dupr´e, J. J. Colman, E. Symbalisty, D. Sent-man, and V. P. Pasko. Physical processes related todischarges in planetary atmospheres.
Space Sci. Rev. ,137:51–82, 6 2008. doi:10.1007 / s11214-008-9385-5.A. Shepetov, A. Chubenko, O. Kryakunova, andet al. The STM32 microcontroller basedpulse intensity registration system for the neu-tron monitor. EPJ Web of Conf. , 145:19002,2017. doi:10.1051 / epjconf / J. Phys.:Conf. Ser. , 1181:012017, 2019. doi:10.1088 / / / / ffi ciency of a neutron monitor calibrated byan accelerator neutron beam. Nucl. Instrum. Meth-ods A , 463:316–320, 05 2001. doi:10.1016 / S0168-9002(01)00280-7.D. M. Smith, L. I. Lopez, R. P. Lin, and C. P. Barring-ton Leigh. Terrestrial gamma flashes observed up to20 MeV.
Science , 307(5712):1085–1088, 2005.M. Stolzenburg, T. C. Marshall, W. D. Rust, andE. Bruning. Electric field values observed nearlightning flash initiations.
Geophys. Res. Lett. , 34:L04804, 2 2007. doi:10.1029 / Na-ture , 551:481–484, 2017. doi:10.1038 / J. Geophys. Res. Atm. , 107:4324, 9 2002.doi:10.1029 / Geophys. Res. Lett. , 31:L05113, 2004.doi:10.1029 / Geophys. Res. Lett. , 38(24), 2011.doi:https: // doi.org / / Journal of Atmo-spheric and Solar-Terrestrial Physics , 136, 10 2015.doi:10.1016 / j.jastp.2015.10.010.H. Tsuchiya. Surrounding material e ff ect onmeasurement of thunderstorm-related neu-trons. Astropart. Phys. , 57(5):33–38, 2014.doi:10.1016 / j.astropartphys.2014.03.009.arXiv:1403.7597.H. Tsuchiya, T. Enoto, S. Yamada, T. Yuasa,M. Kawaharada, and et al. Detection ofhigh-energy gamma rays from winter thunder-clouds. Phys. Rev. Lett. , 99:165002, 10 2007.doi:10.1103 / PhysRevLett.99.165002.H. Tsuchiya, T. Enoto, S. Yamada, T. Yuasa,K. Nakazawa, T. Kitaguchi, M. Kawaharada,M. Kokubun, H. Kato, M. Okano, and K. Makishima.Long-duration γ ray emissions from 2007 and2008 winter thunderstorms. J. Geophys. Res. Atm. ,116:D09113, 5 2011. doi:10.1029 / Phys. Rev. D , 85(9):092006, 2012. doi:10.1103 / PhysRevD.85.092006.arXiv:1204.2578.H. Tsuchiya, T. Enoto, K. Iwata, S. Yamada,T. Yuasa, and et al. Hardening and termi-nation of long-duration γ rays detected priorto lightning. Phys. Rev. Lett. , 111:015001,7 2013. doi:10.1103 / PhysRevLett.111.015001.arXiv:1306.2388.Y. Wada, T. Enoto, Y. Nakamura, and et al. Gamma-ray glow preceding downward terrestrial gamma- ay flash. Communications Physics , 2:67, 6 2019a.doi:10.1038 / s42005-019-0168-y.Y. Wada, T. Enoto, K. Nakazawa, and et al. Down-ward terrestrial gamma-ray flash observed in a win-ter thunderstorm. Phys. Rev. Lett. , 2:67, 6 2019b.doi:10.1038 / s42005-019-0168-y.C. T. R. Wilson. The acceleration of β -particlesin strong electric fields such as those of thunder-clouds. Proc. Camb. Philos. Soc. , 22:534–538, 31925. doi:10.1017 / S0305004100003236.A. G. Zusmanovich, O. N. Kryakunova, and A. L.Shepetov. Tien-Shan Mountain Cosmic Ray Sta-tion of Ionosphere Institute of Kazakhstan Repub-lic.
Adv. Space Res. , 44(10):1194–1199, 2009.doi:10.1016 / j.asr.2008.11.030.j.asr.2008.11.030.