Plain language summary
Many species of elementary particles are born in the terrestrial atmosphere by high-energy protons and fully-stripped nuclei accelerated at exotic galactic sources. During thunderstorms, in addition to this more-or-less constant flux, electrons and gamma rays are produced by the most powerful natural electron accelerator operated in the electrifying atmosphere. Huge fluxes of electrons and gamma rays can exceed the background up to 100 times and pose yet not estimated influence on the climate. More than 2,000 thunderstorms are active throughout the world at a given moment, producing nearly 100 flashes per second. The overall surface of the thunderous atmosphere each moment can be estimated as ≈ 200,000 km2, and according to our estimates, ≈ 1.3*1016 gamma rays with energies above 300 keV are hitting the earth’s surface each second. The long-term effects of this radiation on humans should be thoroughly examined.
Introduction
In spite of many experimental and theoretical studies the relationships between storm dynamics, severe weather, and lightning activity have been least understood (Pawar, et al., 2010). The end of storm oscillations (EOSO) observed at many locations worldwide are closely related to the cloud charge structure, height of the site, and storm dynamics demonstrating different sequences of the field polarities as storm decays. Dependent on the sign of the lowest charge the sequence of the changing near-surface (NS)electric field polarities can be different at different locations and for different storms. However, adding to the NS electric field measurements, the TGE particle registration, and graupel fall observation the EOSO and the electric field structure in the lower part of the atmosphere can be characterized with more details.
It is widely accepted, that the cloud charge structure for a typical thunderstorm contains an upper positive charge region consisting of ice crystals, a main negative charge region consisting of both graupel and ice crystals, and a lower positive charge region consisting of graupel (Kuettener, 1950). The electric charge of graupel is positive at temperatures warmer than -10° C, and negative at temperatures cooler than -10° C (Takahashi, 1978, Wada et al., 2021). In review (Williams, 1989) was stated that the tripolar structure of thunderstorms is supported by a wide variety of observations and that temperature appears to be the most important single parameter in controlling the polarity of charge acquired by the precipitation particles. When graupel falls into the region warmer than ≈ -10° C, a charge reversal will occur in the central part of the storm, and the graupel population will change the charge from negative to positive. Large and dense graupel population either suspended in the middle of the thunderstorm cloud or falling toward the earth’s surface constitutes a “moving” lower positive charge region (LPCR). The dipole formed by the LPCR and relatively stable main negative (MN) charge region significantly intensify the electric field of the dipole formed by the MN and its mirror image in the ground (MN-MIRR, first scenario of RREA initiation, see Fig.1 in Chilingarian et al., 2020, 2021a). A free electron entering the strong and extended electric field accelerates and unleashes the relativistic runaway electron avalanches (RREA, Gurevich et al., 1992). The RREA is a threshold process, which occurred only if the electric field exceeds the critical value in a region of the vertical extent of about 1–2 km. When the second scenario of the RREA origination (MN-MIRR plus MN-LPCR) is realized the electric field in the cloud frequently surpasses the critical value and an intense RREA ends up in an extreme thunderstorm ground enhancement (TGE, Chilingarian et al., 2010, 2011) sometimes exceeding the background level of gamma rays and electrons up to hundred times (Chum et al., 2021). After the graupel fall, the surface electric field again is controlled by the main negative charge region only.
In this letter we discuss the observation of the 24 May 2021 storm on Aragats, during which the NS electric field beneath the decaying thunderstorm makes several characteristic polarity changes over a period of ≈ 60 min; this behavior is called the end-of-storm oscillation (EOSO, Stolzenburg, et al., 2008, Marshall et all., 2009). We analyze the evolution of the EOSO invoking information on TGE electron and gamma ray energy spectra and on electron-to-gamma ray ratio. Thus, demonstrating lowering and consequent decaying of LPCR. Detection of the graupel fall during TGE confirms our inference on LPCR dissipation.
Comparative analysis of 23 - 25 May 2021 thunderstorms
The time series of count rates of electron and gamma ray fluxes, as well, as the energy release histograms, are measured by Aragats solar neutron telescope (ASNT, see detector description in Chilingarian et al., 2016 and Chilinarian et al., 2017a). By the estimation of the spectrometer response function with GEANT4 code, we recover differential energy spectra of both charged and neutral fluxes if a sufficient number of electrons and gamma rays reach the earth’s surface. The lightning identification and distance to lightning flash estimation are done by monitoring of disturbances of the near-surface (NS) electric field with the network of EFF-100 electric mills of BOLTEK company and with Worldwide lightning location network (WWLLN), one of the nodes of which is located at CRD headquarters premises. Meteorological measurements are made with the DAVIS weather station. Panoramic cameras are used for the monitoring of skies above Aragats.
In this letter, we present results of the multivariate analysis of the measurements performed during 3 storms that occurred at Aragats in the end of May 2021. On May 23 storm duration was ≈6 hours with more than a hundred registered lightning flashes. In Fig 1, we present a 16-minute period of the thunderstorm with 2 TGEs terminated by the lightning flashes distance to which are denoted by red lines (1.6 and 5.4 km). The abrupt termination of the TGE can be followed by both time series of the particle detector count rates (black), and – by disturbances of the NS electric field (blue). TGEs were terminated on the initial stage of development, duration of each was ≈20 sec; the NS electric field was in the negative domain, the amplitude of the NS field surge caused by 2 terminating lightning flashes was ≈ 50 kV/m. The electric field recovery after lightning strikes were very fast (a few seconds). In the first part of the storm numerous attempts to start TGE were registered by the ASNT spectrometer. It is interesting to note that a new TGE started just after lightning terminates the previous one during the electric field recovering stage. This is evidence of the largely electrified atmosphere when lowering of the potential drop (voltage) by lightning flash did not quench fully electrostatic field and the field very fast returned to the high values exceeding the critical threshold for starting a new runaway process. There is some kind of interplay between lightning activity and TGE development. When an electric field is very large above the station, multiple RREAs started, however, the combination of the very strong electric field and intense ionization made by the RREA electrons leads to an early stop of RREA by the lightning flash. The electron flux of started RREA opens an ionization path to the lightning leader as was discussed in (Chilingarian et al., 2017b). Numerous examples of the TGEs preceding lightning flashes are shown in the Mendeley dataset (Soghomonyan et al., 2021a). This dataset and other publications unambiguously show that MeV energy particles are not produced by the lightning bolt, but are multiplied and accelerated in the strong electric fields by the RREA process. For many years we perform monitoring of lightning flashes and particle detector signals synchronized with nanosecond accuracy. During these years we did not register any coincidence of thousands of nearby lightning flashes with particle bursts in scintillators, in NaI crystals, and in proportional chambers of neutron monitor (Chilingarian et al., 2019).