3.1. Local weather on May 17th, 2020
The weather one hour before the TGG was rather typical weather in May with no significant weather, air temperature + 9°C, relative humidity 53%, pressure 996 mbar, visibility 50km, wind speed 3 m/s and direction 200° and cloud base at 1140m above the ground and the tropopause height was approximately at 7600m above the ground. At 09:53 the temperature started to drop from 9.4°C and wind speed had increased to 5.5 m/s and the direction at 180°. Light rain started at 10:07 and the wind speed 4.4 m/s with a direction of 170°, the visibility had dropped to 8 km. At 10:16 light rain had already turned into heavy rain and the wind speed was at 8 m/s and the direction had turned almost to the opposite direction to 350°. At 10:19 the heavy rain turned into wet snow and the temperature had dropped to 2.9°C ie. 7°C drop than half an hour before and the visibility was down to 800 m. At 10:23 UTC at the time of the first TGG event the cloud base had dropped to 640 m above the ground, temperature at 2.6°C with medium intensity of wet snow with a wind speed of 6.7 m/s and with a direction of 320° and relative humidity was at 89%. At 10:26 − 10:27 UTC, during the second event, the weather parameters were the same except the wet snow had turned into rain. During the third event the weather parameters were the same except the cloud base had dropped down to 520 m above the ground level. After the TGG events the poor weather conditions continued for about an hour. The weather parameters are plotted in the Fig. 1.
Figure 1 Weather parameters from the Helsinki-Vantaa airport from 09:00 UTC to 12:00 UTC. The panel A shows the barometric pressure (grey) and relative humidity (black). The panel B shows cloud base in meters above the ground (grey) and the snow cover (black). The panel C shows the visibility (black) and the precipitation intensity (grey). The panel D shows the ground temperature (grey) and the 2m air temperature (black)
Time series of the global radiation at the FMI weather station at Helsinki-Vantaa airport from May 17th, 2020, is shown in Fig. 2. Before the TGG events, the Global Radiation varied largely between 150–1100 W/m2 depending on the cloud cover. However, from 10:00 UTC to about 10:30 UTC, or 13:20 − 13:30 local time, the level of the global radiation decreased from 1068 W/m2 to 7.9 W/m2 within half an hour. The cloud cover was, for a moment, optically so thick that it absorbed over 99% of solar radiation.
This means that the cloud emitting γ radiation also contained large amounts of moisture and cold air making the cloud in this storm front exceptional in many ways.
Figure 2 Global radiation measured at FMI weather station during May 17th, 2020. During the passing of the storm front the value dropped from 1068 W/m2 to 7.9 W/m2 indicating a very opaque cloud
3.2. Timing of the γ radiation enhancements
As shown in Chap. 3.1., during the measurements on May 17th, 2020, between 10:20 and 10:30 UTC, an exceptional weather front by or over the location of the detector systems allowing a detection of a TGG.
Figure 3. shows the location of the lightnings observed during 10:16 and 10:31 UTC recorded by the FMI’s lightning observation system in the municipality of Vantaa. The blue circles indicate the locations of the lightning that hit the ground and the red circles indicate intracloud lightnings. The red triangle indicates the location of the detectors. From Fig. 3 it can be seen that the storm front was quite well positioned above the detector system aligning roughly west by southwest and east by northeast and moving towards northeast.
Figure 3 Location of lightning strikes (circles) and the detector (triangle). Lightning strikes from cloud to ground are marked with blue and intracloud lightnings with red. The numbers refer to lightning numbering in Table 2. Basemap data sources: Esri, Intermap, NLS, NMA, USGS; National land Survey of Finland, TomTom, Garmin, Foursquare, METI/NASA
Table 1
The observed γ radiation enhancements.
| 1st enhancement* [%] | duration** [s] | break [s] | 2nd enhancement [%] | duration [s] | break [s] | 3rd enhancement [%] | duration [s] |
Det 1 | 12% | 52 | 86 | 40–50 | 102 | | | |
Det 2 | 5% | - | - | 30 | 104 | 100 | 7 | ~ 20 |
Det 3 | 7% | ~ 50 | 104 | 20 | 90 | | | |
Hisadomi et al. | 4 x | 30 | 90 | 10 x | 70 | | | |
*The radiation increases were calculated from the spectra with 10 second collection time. |
** The duration and the break were estimated from the spectra with 2 second collection time except the third event duration was estimated from the spectra with 10 second collection time. |
Table 2
Finnish Meteorological Institute’s lightning data before and after the two TGG events. The lightnings corresponding to the end of the TGG events are marked in bold letters.
Lightning ID | longitude [°] | latitude [°] | Time [UTC] | Cloud indicator | Multiplicity | Peak current [kA] |
1 | 24,669 | 60,2955 | 10.16.52 | 1 | 1 | 3 |
2 | 24,8846 | 60,2965 | 10.16.52 | 1 | 1 | 9 |
3 | 24,9329 | 60,3165 | 10.16.52 | 1 | 1 | 12 |
4 | 24,8405 | 60,276 | 10.16.52 | 0 | 2 | 26 |
5 | 24,902 | 60,2936 | 10.16.52 | 1 | 1 | 7 |
6 | 25,0056 | 60,3556 | 10.16.52 | 0 | 1 | -3 |
7 | 25,1967 | 60,3626 | 10.16.52 | 1 | 1 | -3 |
8 | 25,1966 | 60,3677 | 10.16.52 | 1 | 1 | -4 |
9 | 24,7263 | 60,2407 | 10.16.52 | 0 | 0 | -4 |
10 | 24,9721 | 60,3079 | 10.18.55 | 1 | 1 | 18 |
11 | 24,9346 | 60,3168 | 10.18.55 | 1 | 1 | 5 |
12 | 24,9554 | 60,3022 | 10.18.55 | 0 | 1 | 35 |
13 | 24,9724 | 60,3107 | 10.21.52 | 1 | 1 | 22 |
14 | 24,7441 | 60,1963 | 10.21.52 | 1 | 1 | -3 |
15 | 24,9538 | 60,3064 | 10.21.52 | 0 | 2 | 15 |
16 | 24,9393 | 60,3122 | 10.21.52 | 0 | 0 | 14 |
17 | 25,1776 | 60,2601 | 10.23.59 | 1 | 1 | 9 |
18 | 25,0117 | 60,2619 | 10.23.59 | 1 | 1 | 3 |
19 | 24,9582 | 60,3249 | 10.23.59 | 1 | 1 | 3 |
20 | 24,7374 | 60,286 | 10.23.59 | 1 | 1 | 3 |
21 | 24,9989 | 60,3373 | 10.23.59 | 1 | 1 | 4 |
22 | 25,025 | 60,3173 | 10.27.10 | 1 | 1 | 29 |
23 | 24,9171 | 60,2056 | 10.27.10 | 1 | 1 | 3 |
24 | 25,0258 | 60,3181 | 10.27.10 | 1 | 1 | 16 |
25 | 25,0324 | 60,3731 | 10.27.10 | 1 | 1 | 5 |
26 | 25,2309 | 60,3715 | 10.27.10 | 1 | 1 | -3 |
27 | 25,2293 | 60,3813 | 10.27.10 | 1 | 1 | -3 |
28 | 25,0545 | 60,3597 | 10.31.13 | 1 | 1 | -6 |
29 | 25,0569 | 60,3371 | 10.31.13 | 1 | 1 | -3 |
30 | 25,0448 | 60,3557 | 10.31.13 | 1 | 1 | 6 |
31 | 25,0395 | 60,2409 | 10.31.13 | 1 | 1 | 3 |
32 | 25,0636 | 60,3662 | 10.31.13 | 0 | 1 | 18 |
33 | 25,003 | 60,3286 | 10.31.13 | 1 | 1 | 6 |
34 | 25,2396 | 60,271 | 10.31.13 | 1 | 1 | -3 |
From the locations of the lightning strikes it can be estimated that the cloud emitting TGGs was very close, if not directly above the detector system at some point. As the lightning strikes, it discharges the cloud and the TGG event is terminated. Figure 4 shows the time series of the total number of counts in the 10 second spectra collected in all three detectors. Unfortunately, detector number 1 suffered from stability issues where the baseline jumped up and down during the collection time, and also sporadic spikes in the count rate were produced. The count rates in detectors 2 and 3 were stable before and after the assumed TGG event. In detectors 2 and 3 the count rate baseline slightly increases after the TGG event which most likely is caused by the washout of short-lived radon progeny from air to ground (Paatero and Hatakka, 1999).
In the present work, the total number of counts in the spectra represent the amount of radiation and are plotted in time series. The other alternative would be to use dose rates but the conversion from a spectrum to dose rate in the presence of high energy gamma-rays may not work properly. Hence, we prefer to use the total number of counts here.
Figure 4 Time series of the total number of counts in the spectra measured with 10 second intervals
As shown in Fig. 4 the normal background varied between the detectors. In detector 1 the background count rate was between 16 000–17 000 counts / 10 seconds while in detector 2 the background count rate was about 25 000 counts / 10 seconds and in detector 3 the background level was around 20 500 counts / 10 seconds. The count rate was depended on the detector location. The count rates were quite high due to the large volume of the detectors and due to the fact that the detectors were unshielded. By visual inspection on May 17th, 2020, after 10:20 UTC anomalously high count rates was observed to last for few minutes. Before the clear jumps in the count rates of the detectors 2 and 3 showed very stable behavior where the count rate fluctuated only slightly. The count rate in the detector 1 varied quite significantly due to the stability issues which produced sporadic jumps and peaks in the count rate time series. It should be noted that the detectors 2 and 3 show very stable count rate also after the observed TGG events.
Figure 5 shows the same time series as in the Fig. 4 but the time window has been zoomed to time period 10:17 − 10:33 UTC when the two or three γ radiation enhancements were observed. The detectors 1 and 3 both observed two events, while in detector 2, there are possibly three events. The first event was visible in all three spectra where the starting time varies slightly. This is most likely since the detectors were placed some hundred meters apart and the cloud emitting TGG triggered the detectors at slightly different times. In the first event in detector 1 the total counts in 10 second spectra increased from 16 500 to 18 500 counts making the increase in γ radiation count rate by 12%, in det 2 the increase was from 25 100 to 26300 corresponding to an increase of 5% and in the det 3 the increase was from 20 600 to 22 000 with an increase of 7%. After the first event, the count rates in all detectors settled to normal values for about 80-90s. After the break a much larger increase was observed in all three detectors. In the det 1 the total count increased from 16 000–17 000 counts/10s to 24 000 counts/ 10s. corresponding to 41–50% increase. In detector 2 the second event raised the total count number from 25 000 to 33 000 corresponding to an increase of 31%, and in detector 3 the total count number increased from 20 700 to 25 000 corresponding to an increase of 21%. These relative increases would suggest that the detector 1 was closest to the second enhancement event and detector 3 the furthest. In detector 2 there is possibly a small third event after a break of about 100s. which is not seen by the other detectors. This third event was the weakest where total count number increased from 25 300 to 27 000 corresponding to an increase of 7%. The information about the observed events were collected to the Table 1. The study by Hisadomi et al. (2021) was inserted to the table for comparison. Interestingly, the γ-ray count rate enhancements observed in this study has a very similar pattern as the γ-ray count rate enhancements observed by Hisadomi et al. (2021). Hisadomi et al. (2021) observed first a smaller enhancement which lasted ~ 30s. with sudden termination, then a break for ~ 90 s., and the second and much larger enhancement lasting ~ 70s. In this study, the first enhancement lasted for ~ 50 s. with sudden termination, then a break for 80–90 s., and then the second and stronger event lasting ~ 100-130s. as shown in the Table 1. Similar pattern for a TGG observation was also done by Torii et al. (2011). This raises a question if this type of pattern, where the 1) weaker TGG event occurs then 2) a break for several tens of seconds and 3) much stronger TGG event lasting several tens of seconds with sudden termination, is really an indicative for TGG events ?
Figure 5 Time series of total counts in ten second spectra in all three detectors zoomed to time period 10:17:00 and 10:33:00 UTC
Figure 6 panel A shows the time series of the total number of counts above the background level in all three detectors with two second collection time. The background was estimated for the detector 1 as 3340 counts/2s., 5030 counts/2s. for detector 2 and 4130 counts/2s. for detector 3. These background counts rates were then subtracted from the time series to obtain the Fig. 6A. Unfortunately, in two-second measurements there were a lot of missing data points due to problems with the data storage. Despite of that from these spectra we can extract more accurate information on timing and the duration of the events. These two-second measurements are still too long for an accurate detection of possible TGFs where the duration is of the order of tens of milliseconds. The first event is fully visible only in detector 1 where the starting time for it was 10:23:10 UTC and the end time 10:24:02 UTC, making the total duration 52 s. During the first event the detector 2 had only recorded one two-second spectrum and hence it must be omitted from this analysis. In detector 3 the starting time can only be estimated since the data begins at 10:23:24 UTC when the count rate has already risen slightly above the background. The start time in det 3 was estimated to be 10:23:10 UTC and the termination time was at 10:24:00 UTC making the total duration of the event ~ 50 s. Interestingly, the first event was terminated with rather sharp cut off at 10:24:00 UTC in both detectors. The second and the larger event started about 1 minute and 15 seconds after the first event. This event was significantly bigger, or the charge center was possibly much closer to the detectors than in the first event. This event was recorded fully by all three detectors. In detector 1 the starting time of the second event was 10:25:26 and the end time 10:27:08 UTC while in detector 2 the corresponding start and end times were 10:25:26 and 10:27:10 UTC and in detector 3 10:25:40 and 10:27:10 UTC, respectively. Based on detector 1 and 2 start and end times, the total duration of the second event was ~ 100 s making it twice as long as the first event. The differences in starting time of the event is impacted by the different cloud to detector distances (detection efficiency decreases with the distance). Since the event termination is a fast process this time difference illustrates more the level of time synchronization between the different detectors. The information about the enhancements is collected to the Table 1.
In general, the temporal structures of the TGG events are interesting. The first event shows gradual increase where the count rates rises modestly and then a sudden termination. The second event shows similar pattern in detectors 2 and 3 while detector 1 is different. A relatively gradual increase in count rate in detectors 2 and 3 then a flat peak and very sharp cut off which ends the event. The time structure including the cut-off was different in detector 1. The rise time of the second event varied from 32 s. in detector 1, 66 s. in detector 2 and 44 s. in detector 3. The second event began at the same time in detectors 1 and 2 but in detector 1 the increase to the peak count rate was faster. In detector 1 the count rate began to decrease already at 10:26:10 UTC. The timing differences can most likely to be attributed to the different locations of the detectors and the movement of the cloud with respect to the detectors. If this was caused by the movement of the charge center it can be estimated that the charge center was not very big in size, of the order of distance between the detectors, possibly some hundreds meters.
Figure 6. Panel A: Time series of the background subtracted total counts in two-second spectra from detectors 1, 2 and 3 drawn to common time scale starting from 10:22:00 to 10:30:00 UTC. Detector 1 data shown in red, detector 2 data shown in black and detector 3 data shown in blue. Panel B: Time series of the high energy γ rays above 3.5 MeV.
In the second TGG event the γ-ray intensities began to rise in slightly different times. The total count rate in detectors 1 and 2 started to increase almost at the same time while in detectors 3 the count rate began to increase about 14 seconds later. This may have been caused by the movement of the charge center in the thundercloud where detectors 1 and 2 might have been closer compared to detector 3. After rising to the peak count rate, the total counts began to drop earlier in detector 1 at 10:26:12 UTC, about 44 s. earlier compared to the detector 2. The detectors 2 and 3 reached a brief steady phase where count rates were roughly stable before the sudden and sharp termination at 10:27:10 UTC. The steady count rate phase in detector 2 lasted 44 s. and in detector 3 42 s.
Figure 6 panel B shows the high energy γ rays during the TGG events. In this study high energy γ rays are considered those γ rays which have energy above 3500 keV. The high energy γ rays are the best proof of a TGG. The high energy γ rays are scatter in forward direction in low angles and the angular spread from the cloud to ground is small. The temporal pattern in Fig. 6B is similar to that of Fig. 6A except the peaks in the detectors 1 and 3 during the second TGG are more narrow. In Fig. 6B the stable phase in detectors 2 and 3 lasts roughly from 10:26:10 to 10:27:00 lasting about 50 s. We added data to the detector 2 time series after 10:28:04 UTC by taking the 10 s. measurements and diving the data by 5 to get an equivalent count rate with 2s. measurements. This was done to see if there is a increase in the high energy γ rays due to the third enhancement. Unfortunately, no clear signal was seen and the third enhancement remains as a speculative.
The Table 2 shows the FMI lightning observations with one second time resolution. The table shows the calculated coordinates from the lightning sensor, multiplicity ie. how many strikes there were in one flash, the calculated peak current in kiloamperes (kA) and the cloud indicator which shows if the lightning struck from the cloud to the ground (0) or was an intracloud (1) lightning. As the table shows the lightning flashes marked in bold corresponded to the termination times of the TGG events. Mäkelä et al. (2017) reported that the median peak current of the flashes in 2016 was about 10 kA but the most powerful ones exceed 100 kA.
A TGG occurs when the electric potential inside a thundercloud is high enough to accelerate free electrons to relativistic energies and cause RREAs. The RREAs collide with atmospheric atomic nuclei exciting them and also loosing energy via bremsstrahlung when passing atmospheric atoms. The original seed electrons are provided by cosmic ray induced ionization. This process continues as long as the potential difference inside the cloud remains high enough. Lighting discharges the potential difference and terminates the acceleration and the RREAs. This stop can be seen as a sudden termination of the TGG and γ-ray enhancement observed by the detectors.
From the Table 2 and Fig. 3 we can see that there were lightnings around the study area. In the lightning data there were lightnings before and after the observed TGG events but no indications of
enhancements in the γ radiation from these lightnings were detected as shown in Fig. 4. It is possible that there were other clouds producing TGGs but they were not close enough to our detectors in order to be detected. Interestingly, in the Table 2 there are three lighting strikes registered at 10:23:59 UTC which correspond to the exact time of the termination of the first TGG event and there are another set of intense lightning flashes occurring at 10:27:10 UTC at the exact time of the termination of the second TGG event. The time of the lightning flashes correspond very accurately to the sharp terminations of the observed γ radiation enhancements. In both cases the lightning flashes occurred as intracloud flashes which was also the case with Torii et al. (2011) and Hisadomi et al. (2021). The lightning that ended the second event had a higher peak current whereas the lightning that ended the first event has a relatively low peak current. The flash that ended the second TGG was significantly stronger flash than a usual flash detected in Finland. Unfortunately, the no matching lightning for the third event could be found.
The Fig. 7 shows the comparison with normal conditions and the third event detected only by the detector 2. The panel 7A shows the low energy part of the γ spectra where reference conditions are drawn in black and the peak of the third event in red. There is a minor enhancement in the energy range of 70–400 keV which is the region where the radiation from the bremsstrahlung is expected. In the panel 7C there are higher number of counts in the high energy part of the spectrum compared to normal conditions shown in panel 7B. The evidence in the γ spectra would suggest that there was a third and a brief TGG event which moved away from the detectors so only detector 2 was able to see it for a brief moment. Another possibility is a TGF which was relatively far away. A TGF would last only a fraction of a second while the collection time was 10 seconds ie. the background data collected before and after the TGF could significantly mask the fast TGF signal. A TGF, and also at some point TGG, would involve a lightninig. Suitable candidate was not registered by the FMI’s lightning sensors. However, it is also possible that this lightning was simply gone undetected by the lightning detection system.
Figure 7 The third event detected by detector 2. Panel A shows the full energy spectrum, panel B the high energy part of the spectrum in normal conditions and panel C at the height of the third event at 10:28:54 UTC
3.4. γ-ray spectrum analysis
The detector systems recorded γ-ray spectra with an energy range from 130 keV to 8900 keV and with time information. The MCAs used 2048 channels in energy scale with second degree polynomial to convert channels to energy. On average this meant that one channel corresponded to 4.5 keV/channel. The NaI(Tl) detectors are widely used in field measurements since they do not require cooling and can operate in a wide range of temperatures and hence are easily transported. The detectors used in this study were large (4”x4”x16”) which meant high efficiency throughout a wide energy range. Especially good efficiencies at high energies are interesting since above 3000 keV there are typically very low number of environmental γ rays. However, there is a cost since the resolution of the NaI(Tl) detector is poor compared to e.g. CdZnTe or HPGe detectors. Therefore, the peaks in the spectrum are broad which means that higher number of counts is needed to form a visible γ-ray peak.
The Fig. 8 shows a spectrum with a ten second measurement time measured with a large NaI(Tl) detector with a typical energy range used in environmental measurements. This spectrum was recorded with the detector 1 at 09:00:50 UTC, well before the γ-ray enhancements, when the count rate in the spectrum were at reference level. In the spectrum three main structures can be seen, at low energies the continuum caused by the cosmic ray muons and peaks at 1461 kev and at 2615 keV which were related to 40K and 208Tl. All these peaks are natural and more or less always present in the γ-ray spectra.
Figure 8 Example of a γ-ray spectrum measured with a large NaI(Tl) detector
The Fig. 9 shows a comparison of five 10 s. spectra from detector 1 where both TGG events were observed. We selected three time periods for comparison, first at 09:00:00 UTC when the detector was in reference condition as shown in Fig. 4, the second spectrum corresponds to the maximum of the 1st enhancement at 10:23:40 UTC and the third at the maximum of the 2nd enhancement at 10:25:50. The panel 9A shows the comparison of the low energy part of the spectrum where the 1st and 2nd enhancements caused a rather continuous increase in the 70–300 keV energy range. This is believed to be bremsstrahlung from the RREAs (Babich et al., 2004). The panel 9B shows the comparison in the typical energy range of 0 to 3000 keV for environmental γ spectrometry. The enhancement can be seen in the low energies, and it continues until to ~ 800 keV where the intensity of γ radiation is at the same level as in the normal conditions measured at 09:00:00 UTC. The panels 9C, 9D and 9E show the high energy part of the γ spectra from 3000 keV to about 8900 keV. In normal conditions this part contains very few counts as shown in panel 9C where there are some counts caused by the cosmic-ray muons. The cosmic-ray flux can be considered as constant in the time scales used in this study. The panel 9D shows the high energy part of the spectrum during the 1st enhancement where the number of high energy γ rays was increased compared to the normal condition shown in panel 9C. In panel 9E the high energy part of the spectrum during 2nd event is shown. The number of hard γ rays has increased further and was significantly higher than during normal condition. No clear peaks were visible and this can be attributed partly to the poor resolution of the NaI(Tl) detectors. The observed γ-ray energies go up to the maximum energy of 8900 keV. It should be noted that in NaI(Tl) detectors the detection efficiency is the highest at the low energies and the detection efficiency goes down as a function of energy. This enhances the detection of low energy γ rays compared to the detection of high energy γ rays. The formation of hard γ rays is complex where the bremsstrahlung from the RREAs is the main source for high energy γ rays. The relativistic electrons can also cause nuclear reactions e.g. in 14N and in 16O. 14N and 16O have rather complex nuclear structures where excitation would lead to emission of multiple hard γ rays. The hard γ rays can also cause photonuclear reactions where a γ ray knocks out a neutron from atmospheric 14N or 16O, forming radioactive 13N and 15O isotopes which decays via positron emission (β+). This causes an increase in 511 keV γ-ray emissions also known as the annihilation peak. Similar observations of a TGG was also done by Helmerich et al. (2024) in a balloon flight which detected a clear increase in low energies, an increase of 511 keV peak and also hard γ rays with energy up to 6000 keV.
Figure 9 Ten second γ spectra from detector 1. Spectrum taken at 09:00:00 UTC, drawn with black, represents a normal spectrum while spectrum taken at 10:23:40 UTC, drawn with red, represents the peak of the first TGG event. The spectrum taken at 10:25:50 represents the peak of the second event. The panel A shows the best view for low energy part of the spectra, panel B shows the middle energy part of the spectra, panel C the high energy part of a normal spectrum, panel D high energy part of the peak of the first event and panel E the high energy part of the peak of the second event
We also performed a background subtraction for the second TGG event. In total 11 ten second spectra, were summed up to form one spectrum. For the background, or reference conditions, we summed up 11 spectra from 09:00 to 09:02 UTC time period. These two sum spectra were then subtracted to study thenet spectrum caused by the second TGG event shown in Fig. 10. The TGG affected the whole energy range starting from the lowest energies. The majority of the TGG radiation comes from the bremsstrahlung emitted by the slowing down of the electrons in RREAs. There is a minor enhancement of the 511 keV annihilation peak in the detectors 2 and 3 which could possibly originate from the pair production caused by high energy photons and/or from β + decay of 13N or 15O. In high energy γ rays there are no visible peaks but rather steady continuum. There are cases where slightly higher number of counts accumulated into same channels in high energies above 2700 keV but this is most likely statistical fluctuation. NaI(Tl) detectors typically have poor resolution at these energies and forming a clearly visible peak would require much higher statistics. In detector 3 there are peaks from the naturally occurring 40K, 208Tl and 214Bi where the background was slightly higher during the TGG compared to the reference conditions. The interesting fact is that the shape of the spectra shown in Fig. 10 resembles closely the observations by Wada et al. (2019) and Helmerich et al. (2024). Berge and Celestin (2019) calculated the theoretical TGF photon spectrum created by a RREAs and that resembles closely the spectrum shape observed in this study. The data in Fig. 9 has been plotted into log-log scale, after 200 keV the data can be fitted a straight line meaning that the data presents a power law behavior. However, in order to resolve the true be distribution of γ energies from RREA the spectrum would have to corrected for detection efficiency, but this could not be done since the detector is not calibrated for high energies.
Figure 10 Background subtracted gamma spectra from all detectors. The spectra produced by the second TGG looks very similar to the predicted photon spectrum from RREAs