First, we provide an overview of physicochemical anomalies observed during the Kobe earthquake, which might be related to the mysterious EQL phenomena. During the Kobe earthquake, the Rokko-Awajishima fault zone was activated. Especially, the fault on Awaji Island showed a large lateral slip of about 2 m and a vertical displacement of about 1 m, but a less remarkable surface seismic fault was observed than that in Kobe. Nevertheless, most of the city areas were shaken with seismic intensity VII on the Japan Meteorological Agency (JMA) scale, and serious damage was concentrated along the fault zone in Kobe City. In this area, a variety of physicochemical phenomena were observed at the locations noted in Fig. 1.
The EQL appeared near the Rokko-Awajishima fault. Tsukuda5 estimated the features, position, and size of three luminous bodies, #2–#4, shown in Fig. 1, floating in the lower sky above Kobe by interviewed eyewitnesses. These bodies were 150–200 m above sea level. The color of luminous body #2 was blue, while that of #3 and #4 was orange. Note that the duration time was as short as several seconds or some more, unlike the long-lived luminous cloud #1 in Fig. 2.
Anomalous increases of radon in association with the Kobe earthquake were observed in the groundwater6 (C in Fig. 1) and in the atmosphere7 (D in Fig. 1) near the eastern end of the Rokko-Awajishima fault zone. It is worth noting that Yasuoka et al.,7 who observed the radon anomaly at Kobe Pharmaceutical University located at point D in Fig. 1, showed an interesting analytical result: the anomalous 222Rn yearly variation, which they observed, could be interpreted as linked to the main stresses that directly led to the Kobe earthquake. Pulinets8 confirmed that the radon anomaly before the Kobe earthquake is a perfect example satisfying the formal seismological determination of an earthquake precursor.
In addition, anomalies in electromagnetic wave emissions were also observed over a wide frequency range from extremely low to high frequencies.9 Among them, the 22.2-MHz electromagnetic noise signals observed at the Nishi-Harima Observatory, shown in the inset of Fig. 1, showed that it occurred immediately before and after the earthquake and that the direction of the noise wave came from the epicenter area.10
As mentioned above, the Kobe earthquake caused great damage in Kobe City and in the northern part of Awaji Island along the Rokko-Awajishima fault zone, but not as much damage in the Uemachi fault zone. However, strong seismic motion with a seismic intensity of VII on the JMA scale occurred not only in the Kobe area and at northern Awaji Island along the Rokko-Awajishima fault zone, but also in some areas on the west side of the Uemachi fault zone, as shown by the darkened areas identified as “VII” along Uemachi fault zone in Fig. 1.11
Beneath this plain is Neogene granite, which is buried in the fault about 1 km lower at the west side than at the east side, and has a two-layer structure in which a relatively light and soft sedimentary layer formed millions of years ago covers the bedrock of granite.11 It is highly probable that the cause of the localized violent shaking of seismic intensity VII was attributed to the focusing effect due to the refraction of seismic lines and total internal reflection related to the area’s complicated structure, such as the deep underground structure, surface ground composition, and topography.11
In light of these facts, we hypothesized that the possible cause of luminous cloud formation was radon emanating from land reclaimed from swamp near the Uemachi fault zone, which is adjacent to the stratum where the violent seismic intensity VII was localized. We assume that the gas bubbles containing radon and carrier gases such as methane and water vapor present in the sediment/soil on the granite basement, which is rich in radioactive species, continued to uplift toward the ground surface from the beginning of the unstable rupture just before the earthquake to the aftershock.
Even though high-density radon (ρr = 9.73) emanates from the ground and migrates in air with a density of ∼1.29, our preliminary numerical simulation showed that it is impossible for radon to migrate to an altitude above ∼140 m unless we assume a very high blow-out velocity, greater than 10 m/s, from the ground surface. However, recent studies suggest that: Water vapor is capable of capturing radon12. Water clusters exist in gas phases, and the sizes of clusters in the gas phase increase with the relative humidity13. The water molecules rearrange around the guest molecules such as radon or methane to form a hydrate structure14,15, where the surrounding water molecules are connected to each other by hydrogen bonds14. Water vapor and methane, lighter than the humid air, should be considered as carriers of radon to the atmosphere12.
With reference to these previous studies, we further assume that radon and/or radon/methane bearing bubbles migrate up to the ground surface. The bubbles will then pop and generate water vapor cluster. The mixed clusters in gas phases composed of radon and/or radon/methane water vapor migrates from the ground surface into the atmosphere.
To investigate how the concentration of radon and/or radon/methane bearing water vapor spatiotemporally changes as it emanates from underground into the atmosphere, two-dimensional convection-diffusion transport equations were numerically solved using the open-source software “OpenFORM® v2106,” with the SST k-ω model employed as the turbulence model. To consider the weather conditions on the day of the Kobe earthquake, we set the relative humidity of the atmosphere as 80% and the temperature as 3°C, with no influence due to crosswinds. In this case, the effective density ρha of the 80% humid air is obtained as 1.27 kg/m3 from the psychrometric chart. The effective density of radon-bearing water vapor ρrw should be in the range of 0.598 (density of water vapor at 3°C) < ρrw ≤ 1.27 (density of air with 80% humidity) because the buoyancy does not work on radon-bearing water vapor when ρrw > ρha. The buoyancy due to the density difference between the radon and/or radon/metane bearing water vapor and the humid air is expressed by the Boussinesq approximation. We then numerically calculated the parameters of ρrw at 1.270 (ρrw = ρha), 1.100, 0.800, 0.600, and 0.598 (without radon), of which the expansion coefficients β = (ρrw - ρha)/ρha are 0, 0.134, 0.370, 0.528, and 0.529 kg/m3, respectively. We further set the initial upward velocity of radon-bearing water vapor v as 0.5 or 0.1 m/s.
As a result, we confirmed that the radon-bearing water vapor can reach an altitude of nearly 140 m. The typical results with ρrw of 0.800 kg/m3 and v of 0.1 m/s are shown in Fig. 4a-c for t = 0.5 and every 30 s from t = 0 (start of migration into air), where the relative spatiotemporal distribution of the concentration of the radon-bearing water vapor is displayed in color and the condensation of radon and/or radon/methane bearing gas is set at 1 at t = 0 s. The leading vortex of radon-bearing water vapor grows gradually over time and with increasing altitude, ultimately reaching an altitude of 100–150 m at around t = 180 s. Figure 5d-f shows the results at v = 0.1m/s with ρrw of 1.27 kg/m3,1.10 kg/m3, 0.800 kg/m3,0.600 kg/m3, and 0.598 kg/m3, respectively. When the density ρrw is 1.10 kg/m3 or less, the leading vortex reaches 100–150 m within a short period of 150–240 seconds, but the concentration there dilutes to about ~ 1/100. On the other hand, when the density ρrw is the same as the humid air density of 1.27 kg/m3, it will take a considerable amount of time to reach an altitude of 100 m.
Now we focus on radioactivity of radon. The radon isotope, 222Rn decays and emits alpha particles and gamma rays as
222Rn→218Po་0γ+4He(α).
As a result, air and water molecules are impacted by high-energy alpha rays, causing excitation and ionization of the molecules. Then, the ions act as condensed nuclei for water vapor to form clouds. Then, radioluminescence during transition and recombination of excited and ionized molecules also occurs.16,17,18 Previous work on radioluminescence has involved theoretical studies16,17 and laboratory investigation.18 Thus, if the present hypothesis is valid, the photographs in Fig. 2 might provide the first evidence of radioluminescence observed in the field.
Evidence for the validity of this hypothesis is the existence of clouds that extend linearly from the luminous clouds. Because γ-rays are a type of electromagnetic wave, they can rapidly reach the distant atmosphere. Also, because of their high energy, electrons are ejected from air molecules in the γ-ray path, and so aggregated water molecules can allow visualization of the γ-ray path. The electrons and ions generated by γ-rays create electric currents in the Earth’s electric field, the magnetic field induced by the electric currents causes “kink instability,” and then clouds are forced to twist, resulting in the formation of mottled clouds. In fact, we can observe such a phenomenon in the laboratory using a Wilson cloud chamber.
In addition, the tracks of radiation extending from the luminous clouds seen in Fig. 2 are not isotropic but are biased to the left side of the photograph, which is the direction toward the sea in Osaka Bay, while the right side is the direction toward the land. This anisotropic phenomenon might be due to clouds being formed in water-saturated air above the relatively warmer sea, rather than colder land areas.
Another reason why the present radioluminescence hypothesis is plausible is based on previous research. Specifically, according to Boyarchuk et al.,16 radon gas is a source of positively and negatively charged heavy particles formed by the ionization of radioactive products that resulted in the formation of complex chemically active structures of ion radicals such as H3O+∙(H2O)m, NO3−∙HNO3∙H2O, and NO3−∙(H2O)n. As a result of the association of such hydrated ion radicals, a neutral cluster is formed that is a dipole and a quasi-molecule. Rotation-rotation transitions of the corresponding dipoles then can be a source of megahertz-order electromagnetic radiation.
In fact, as mentioned above, unusual pulsed radio emissions at 22.2 MHz, as shown in Fig. 5,10 were observed with the radio interferometer at Nishi-Harima Astronomical Observatory, located about 77 km from the epicenter (inset in Fig. 1), both before and after the Kobe earthquake.10 The direction of the noise wave came from the epicenter area, and this noise source can be attributed to any seismic physicochemical phenomenon, such as EQL. Even if high-frequency electromagnetic waves are generated in the deep underground focal zone, they cannot reach the ground surface due to the skin effect, so the source should exist at ground level. As the duration of illuminous bodies #2–#4, witnessed in Kobe City, was as short as several seconds,5 and therefore it could not be a candidate source for the long-lasting 22.2-MHz electromagnetic noise. No other possible causes have not found around the area along the Rokko Awajishima fault zone. The long-lived EQL #1 is therefore the most likely candidate for the electromagnetic noise at 22.2 MHz, as it continued for about 30 min after the earthquake, which is consistent with Yokota’s testimony that the luminous cloud in Fig. 2 might have lasted for about 20–30 min. Note that the altitude of Nishi-Harima Observatory is 435m above sea level, and the distance to Izumiotsu City is 114km. Therefore, it is possible to propagate the 22.2 MHz electromagnetic wave emitted from the luminous clouds at an altitude of 140m to the Nishi-Harima Astronomic Observatory in line of sight.
Similar electromagnetic noise appeared before the earthquake for about 20 min from 05:06 (Fig. 4). In relation to this, it is noted that in Nishinari, Osaka City (E in Fig. 2), about 16 km away from Izumiotsu City, a witness testified that “the outside was as bright as before sunrise” at 4:30, about 1 hour before the earthquake.1 In the photograph in Fig. 2, some streaky clouds that spread over time can be seen, as noted by dotted lines in Fig. 3a. These are likely to have occurred before the earthquake. In other words, it can be inferred that the luminous clouds also appeared just before the earthquake, causing electromagnetic noise at 22.2 MHz and temporarily brightening the dark sky.
Next, we discuss why luminous cloud #1 in Fig. 1 had a long life, while the luminous bodies witnessed over the Rokko-Awajishima fault zone (cf. #2–#4) had a short life. All of the luminous clouds/bodies #1–#4 might be related to the emanation of radon from the fault, but the fault in Izumiotsu City has shear-crushed strata with weak gravel, where radon-bearing gas bubbles could continue to be emanated for a considerable time period. Another possibility is that numerous ions and electrons generated by alpha-ray impacts formed intense currents with induced magnetic fields, which resulted in the formation of long-lived “plasmoids.”19
Finally, we discuss why the color of the streaky clouds seen in Fig. 2 changed from pale blue to white and red depending on the elapsed time and altitude. As already noted above, among the R, B, and G images in Figs. 2i-k, the R image is predominant. Because red sunlight was not present during the time when the luminous clouds were photographed, we suggest that this red color may be attributed to the light-emission band due to the transition from the excited level of a huge cluster containing water molecules, as described by Boyarchuk et al.16 As illustrated in Fig. 6a, the light emitted from luminous clouds was selectively scattered by mottled clouds consisting of water molecule complexes with different sizes. The size of cloud particles formed when γ-ray tracks reaching higher altitudes remained small, possibly as small as about 10 nm or less, because the concentration of water vapor is lower at higher altitudes. Under these circumstances, Rayleigh scattering may have come into play, and the scattered light became predominantly blue. When the cloud particle size in the middle/lower altitude grew to several tens of nanometers, the scattered light took on a white appearance due to Mie scattering. Furthermore, due to the higher water vapor concentration in the lower sky, huge cluster molecules could grow to sizes on the order of 1 µm or more, after which Mie scattering with a red color emitted from luminous clouds became predominant. This discussion is in good agreement with the computer graphic images in chronological order that Sakaguchi2 created, which are included in Fig. 6b.
The present study is summarized as follows. Mysterious EQL consisting of luminous clouds and mottled/colored clouds extending from them were surprising and frightening for residents when they witnessed the unusually bright and colorful luminous phenomenon in the dark sky. These clouds can be explained by the hypothesis that the cause is radioluminescence due to radon emanating from an active fault shear zone. This hypothesis is consistent with various physicochemical anomalies, such as a radon increase in water and the atmosphere and megahertz electromagnetic noise, observed during the Kobe earthquake. Even if the mysterious EQL is attributed to the formation of radon-induced plasmoids, there are still remaining issues as to how much radon concentration in air would be needed to result in such a critical state to generate the photographed EQL. Further investigation of such radioluminescence is therefore required.