Analysis of Long-Term Monitoring of Radon Levels in a Low-Ventilated, Semi-Underground Laboratory—Dose Estimation and Exploration of Potential Earthquake Precursors

Abstract

This study involves continuous radon monitoring during the academic year 2023/2024. An Airthings Corentium Home radon detector placed in the basement laboratory of a faculty building in Kosovska Mitrovica (N 42.897°, E 20.867°) was used for continuous measurements. The average radon concentration was 303 Bq/m³, and a seasonal pattern during the measuring period was observed. For the first time, the results were grouped by week, excluding non-working days, to present a real case scenario with the aim of assessing the radon exposure of students, teachers, and employed persons. The inhalation dose from radon (1.54 mSv) was very high considering that exposure occurred in both semesters. Another aspect of continuous radon monitoring was to explore the relationship between indoor radon measurements and the occurrences of earthquakes in the Balkan region. Daily variations in radon (peaks and differences) were analyzed at the monitoring site by using both empirical laws and taking into account the earthquake data set provided by the Seismological Survey of Serbia. The analysis revealed that the events chosen to confirm a clear association between earthquake occurrence and enhanced radon activities in the air(as a precursor of seismic activities) did not meet the required criteria but most likely reflected external meteorological conditions.

1. Introduction

Rocks, soil, and water in the Earth’s crust contain radionuclides from radioactive series of uranium and thorium, which are constantly decaying and producing radium as the immediate parent of radon. Radon is a noble radioactive gas (odorless, colorless, and tasteless) and the main contributor to the annual dose that humans receive from natural sources of radiation on Earth [1]. Due to it being much more mobile than radium, radon easily rises from the ground to the air above by diffusion and convection. Its mobility is affected by the extent of rock fracture and soil permeability. Radon travels more slowly through clay (impermeable soils) than coarse sand and gravel (permeable soils). High radon levels may be detected in indoor environments depending on the concentrations of parent radionuclides, atmospheric and soil conditions, and anthropogenic features (type of building, ventilation, heating, occupant behavior, etc.). Generally, the main radon supplier is bedrock, particularly in areas with high geogenic radon potential [2]. Radon can directly travel from shallow soil through cracks in buildings, leading to radon accumulation primarily in low-ventilated underground floors and basements. If a steady-state radon source (radium, whose half-life is as long as 1600 years) is present, the distribution and behavior of radon in the atmosphere are mainly governed by meteorological processes.

Among the three radon isotopes, namely radon (²²²Rn), thoron (²²⁰Rn), and actinon (²¹⁹Rn), from a radioprotection point of view, radon is the most significant due to its relatively long half-life of 3.82 d [3]; it can travel relatively far from its radium source in the soil, carried by fluids such as soil-gas and water or CO₂ and methane. Radon and its short-lived progeny are responsible for health hazards caused by inhalation; it is a human carcinogen and the second most prevalent cause of lung cancer after tobacco smoking [4]. A national reference level recommended by the World Health Organization (WHO) is 100 Bq/m³ [4]. The European Commission’s Directive EURATOM 59/2013 has set a reference level of 300 Bq/m³ as the national action level threshold [5]. In terms of radiological protection and health preservation, indoor radon exposure should be the primary focus because people spend most of their time indoors (either at home or in the workplace). Some researchers have provided evidence of an increase in the risk of exposure at radon levels below 200 Bq/m³ (the concentration that requires action in many countries) [6], although the WHO proposes that the chosen level should not exceed 300 Bq/m³ [4]. The incidence of lung cancer depends on the person’s age at exposure; it is two-fold higher for adolescents than for adults at the same dose [7,8]. Radon-related lung cancer could be prevented in 35–40% of cases if radon levels in buildings were reduced below 100 Bq/m³ [9].

On the other hand, radon is a good indicator of crustal activity and seismic movements. Radon travels toward the ground surface mainly through cracks and faults in the Earth’s crust through diffusion and advection. Diffusion is driven by a concentration gradient (described by Fick’s law) and occurs over short distances, while advection is driven by pressure gradients (described by Darcy’s law) and is important for long-distance transport. However, migration by diffusion is negligible in comparison to advection [10]. Both radon transport mechanisms depend on soil porosity and permeability and are influenced by the stress field at the same time [11,12].

Since the amount of radon itself is usually too small to form pressure gradients, it must be carried via a gas mixture flow (He, H₂, CO₂, SO₂, CH₄, and H₂S) through a fissure space [13]. When earthquake stress accumulates in the Earth’s crust, a change in the strain field occurs. The formation of new pathways under tectonic stress leads to changes in gas transport and an increase in volatile compounds traveling from the deep layers to the soil surface. Fluids play a crucial role in controlling the strength of crustal fault zones; episodic fracturing leads to enhanced permeability and changes in fluid flow, which are highly related to anomalous changes in radon concentration [14].

In earthquake-prone regions, radon is usually the most preferred geo-gas to be monitored as an earthquake precursor. The abrupt emission of high radon concentrations through cracks, along with the release of other gases suitable for radon transport to the ground surface, is attributed to strained rocks before a sudden slip. Due to pre-seismic stress or rock fracture, spike-like peaks in radon concentrations (conspicuous short-term changes) occur before a major earthquake. Additionally, spatial and temporal variations in radon measurements in soil and groundwater can provide information about geodynamic events occurring in an active fault zone [15].

Meteorological conditions (rainfall, pressure, soil humidity, temperature, and wind) control radon concentrations in soil gas. These parameters change the characteristics of soil/rock and influence the rate of radon transport, with shallow soil layers more affected than deeper layers. Pressure has a major influence on radon concentrations in soils: a decrease in pressure causes an increase in radon exhalation from the ground, whereas an increase in pressure forces radon into the ground, causing its dilution [16]. During winter months, the air temperature is lower than the soil temperature, resulting in an upward radon movement, whereas during summer, the soil temperature is lower than the air temperature, and therefore the upward flux is reduced. Radon concentrations in the soil are also controlled by soil moisture and rainfall through fissures opening in the surface [17].

Anomalous radon concentrations have been reported before strong earthquakes. The complex behavior of radon gas before an anomaly was reported as a gradual increase in radon levels three months before an earthquake, followed by a marked increase two weeks and a sudden decrease one week before a shock; after an earthquake, radon levels return to values before the pre-seismic events [18]. Other studies have reported fluctuations in radon concentrations around a mean value, with a peak of about two or three times the mean value preceding some seismic events. Anomalies have been associated with changes in crustal strain and are likely related to local seismicity [19]. In addition, some authors have reported rather significant radon anomalies across fault lines [20]. During 15 years of continuous monitoring of geophysical events on Mt.Etna, which is characterized by tectonic and volcanic phenomena, as well as numerous earthquakes, it has been shown that with an increase in radon levels, the daily rate of earthquake and strain release increases, which corresponds to the beginning of an eruption [21]. Terray et al. also confirmed releases of high radon concentrations during eruptive activity in the volcanic environment of Mt. Etna [22]. Many studies have been conducted that identify radon as a seismic precursor and provide evidence of the association between its complex behavior and earthquake phenomena [23,24,25,26,27,28].

In this study, the results of indoor radon monitoring were analyzed with two main objectives: (1) to assess the inhalation dose of students and employees exposed to radon levels during their stay in the laboratory (lab exercises) and (2) to examine if earthquakes in the tectonically active Balkan region induce radon anomalies which could be used as precursors.

2. Materials and Methods

2.1. Characteristics of the Building

The subject of this research is the physics laboratory located in the Faculty of Sciences and Mathematics in Kosovska Mitrovica (N42.897°, E20.867°), Serbia. The building itself has three stories with a semi-excavated basement. It was built on sandy soil in the 1970s using concrete, red bricks, and lime as the primary materials. Five years prior to radon monitoring, the building passed energy renovation procedures: the exterior walls of the building have been thermally insulated with Styrofoam, and all the windows have been replaced with leak-proof PVC (polyvinyl chloride). External insulation leads to less air permeability and an increase in the radon infiltration rate, particularly in the basement (lower part) of the building [29]. This, in principle, occurs in the cold season due to greater indoor/outdoor temperature differences.

The laboratory is located in the basement (approximately 1.3 m underground on the south side) in the center of the building. It has parquet flooring, and the interior walls were later partitioned with plasterboards. Air renewal is provided through natural ventilation. The windows are not open frequently, only briefly in the morning. During the heating season (October–April), in addition to central heating, an electric heater is used.

2.2. Underlying Geology

Kosovska Mitrovica is well-known for the mining activities at the Trepča complex during the previous 80 years. The study area contains significant deposits of lead–zinc ore and other metals (Ag, Au, and Cu). In terms of geological diversity, it ranges from the Ordovician–Silurian to the Quaternary. Larger masses of intrusive rocks have been formed due to strong volcanic activity in the past [30]. The area is characterized by a deep fault as a result of magmatic activity. Different vertical movements due to tectonic shifts have generated a network of seismic faults. One of the strike directions of young (neotectonic) faults extends in the NW–SE direction following the river valley [31]. The region is classified as a moderate-to-high hazard area according to the European Seismic Hazard Map (ground shaking is expressed in terms of the unit of gravitational acceleration, g) [32]. In the Balkan region, anomalies appear at distances that are sometimes much larger than typical source dimensions. Earthquakes occur in a strain field higher than 10⁻⁹/s, with some occurring in strain fields higher than 10⁻⁸/s, while the slip rate of active faults is 0.1–0.5 mm/year [32]. These phenomena affect increased radioactivity since radon can easily travel through faults toward the ground surface. A recent study reported that this area is prone to radon, due to high radon levels measured in a large number of dwellings [33].

2.3. Measurement Methods

Radon levels were measured using alpha spectrometry by placing an Airthings Corentium Home digital detector in a passive diffusion chamber. This detector is unaffected by other radiation and measures radon levels in the range of 0–9999 Bq/m³. Device uncertainty is less than 10% for one-month measurements. Measurement accuracy from 7 days to 2 months is 5–10% at the typical level of 200 Bq/m³. The first results on the detector’s display are available after 6–24 h, indicated as long-term average (LTA; updated once a day) and short-term average (STA; last-day values updated hourly) radon concentrations [34]. The long-term values are constantly averaged during the measurement period by the detector itself.

The radon measurements were performed from 28 August 2023 to 7 June 2024 during working days (from Monday to Friday). The radon detector was placed on the laboratory shelf away from doors and windows (1.5 m from the floor, at least 2 m from windows/doors). Radon data were continuously recorded at the same time each day (between 8 and 9 AM) prior to opening the windows. Almost all measurements were taken, except for a few days.

2.4. Dose Assessment

The effective radon dose for students and employees, for one semester, in the laboratory was estimated as follows:

$$E=\sum _{i=1}^{15}{C}_i·DC·T$$

(1)

where E is the effective dose (mSv) for an adult; C_i is the long-term average radon activity concentration during the working week i(Bq/m³); DC is the dose coefficient for radon inhalation, determined as 3.6 × 10⁻⁶ mSv h⁻¹ per Bq m⁻³(recommended by the UNSCEAR [1], which corresponds to an equilibrium factor of 0.4 between radon and its progeny for most indoor conditions); and T is 40 h, considered the exposure time of one week of full-time work (8 h per day). In this scenario, the dose was not estimated during exam periods due to lower exposure times.

2.5. Earthquake Prediction Method

Besides meteorological parameters that influence radon fluctuations, large daily differences and peaks in radon concentrations can also be associated with ground shaking due to geotectonic complexity in a certain area. Crustal anisotropy, discontinuities, or poor contact along some faults can prevent further stress transfer. Several models have been developed to determine the relationship between physicochemical parameters of radon anomalies and seismic events [12]. The dilatancy model reveals crustal activity in the preparation zone of an earthquake [35,36]. An increase in tectonic stress causes the extension and opening of cracks in a porous, cracked rock. Fluid flows into the opened cracks; decreases the pore pressure; and due to cracking, increases radon diffusion from the rock matrix to the fluid pores. This modifies pore strength and pressure, leading to an increased number of cracks and main ruptures.

Radon concentration increases upon crack formation in rocks. Afterward, radon emission becomes stable or decreases before an earthquake. The width of the zone covered by the stress load is proportional to the magnitude and depth of an upcoming earthquake. The “precursor phenomena” are caused by pressure variations that change the characteristics of rocks. An empirical relationship between earthquakes and indoor radon levels has been proposed by Hauksson and Goddard [37]. The minimum magnitude M (the Richter scale) required for a radon anomaly at a distance D (km) from the epicenter of an earthquake and the site of the observed radon anomaly are determined as follows:

$$M=2.4{ log}_{10}D-0.43$$

(2)

To estimate the strain impact at the monitoring site, in cases where only the magnitude is known, the empirical equation given by Dobrovolsky et al. [38] should be applied:

$$R_D={10}^{0.43 · M}$$

(3)

where R_D (km) is the strain radius of the effective precursory manifestation zone. This should be performed for all earthquakes because a small-magnitude earthquake next to the monitoring site might have the same strain impact as a large-magnitude event at a greater distance [39].

The earthquake database made available by the Seismological Survey of Serbia [40] has been used for the analysis.

3. Results and Discussion

3.1. Inhalation Dose from Radon Exposure of Students and Employees

Radon concentration in the laboratory averaged by the detector itself during the measuring period (in working days) was 303 Bq/m³, indicating an increase in levels in the autumn–winter period and a decreased level in spring (Figure 1). This is in good agreement with other studies [41,42,43,44,45] since indoor radon concentrations typically exhibit seasonal variation due to changes in meteorological conditions. Radon levels may also increase in the cold season due to heating in the building/workroom.

The maximum long-term average (LTA) radon concentrations were recorded at the end of January, and minimum levels were detected at the end of August. The occupational rate of the Physics Laboratory varied throughout the monitoring period depending on when exams or classes were held. Hence, the survey distinguishes four periods in order to account for the more realistic scenario of students’ and employees’ exposure to radon (Figure 1). The short-term average (STA) radon measurements exhibited very high fluctuations due to daily changes in meteorological parameters. These variations were extremely high in the middle of January, probably due to light snow and frozen soil, and in the middle of May, likely due to long-lasting light rain and drizzle [46]. The LTA concentrations indicate that the radon level is at least 1.5 times higher during the autumn–winter season than during the rest of the year (Figure 1).

For dose estimation, low LTA radon concentrations obtained two days after battery replacement were discarded from the whole dataset due to detector reset. Hereafter, LTA and STA radon concentrations were averaged for each week of both semesters. Based on LTA values, Equation (1) allows calculation of the averaged inhalation doses from radon exposure of students and employees during the first and second semesters, which were 0.81 mSv and 0.73 mSv, respectively. In comparison with the annual average dose from radon inhalation of 1.15 mSv, which assumes an exposure time of 7000 h, recommended by the UNSCEAR [1], the total dose (1.54 mSv) is very high considering only 1200 h of exposure in both semesters. Radon inhalation doses according to LTA and STA radon concentrations averaged by working week in both semesters are presented in Figure 2. An almost smooth dose line is observed according to the LTA values (Figure 2), except for the part of the line that is related to the second battery exchange. In fact, it can be seen (Figure 1) that radon concentrations took longer to re-establish a normal level during the fourth week of the second semester (Figure 2). In addition, lower STA values were recorded immediately after restarting the measurement, which subsequently led to lower LTA values due to averaging by the detector itself. When STA radon concentrations were used for dose estimation, the calculated value was 9% higher (1.68 mSv). In real conditions, these results indicate the need for remedial actions against radon exposure.

3.2. Exploration of Potential Earthquakes

3.2.1. Analysis of Characteristic Radon Peaks

The second aim of this study was to determine whether seismic events occurring in the Balkan region could generate a radon anomaly that could be clearly identified. In other words, the purpose was to determine if radon could be used as a reliable precursor of seismic activity. Ground shaking is accompanied by the anomalous emanation of geo-gases (including spike-like radon peaks) before seismic events in an area. A significant radon anomaly is considered when radon activities differ from the mean value ± 2SD [47]. Based on the statistical analysis of daily radon variations for each season (

Table 1. Descriptive statistics of STA values grouped by season.

Period of Measurement	28 August 2023– 29 September 2023	1 October 2023– 30 March 2024	1 April 2024– 7 June 2024
Mean ± SD (Bq/m3)	290.3 ± 154.1	406.7 ± 231.8	303.2 ± 210.1
Range [Min–Max] (Bq/m3)	621 [132–753]	1125 [74–1199]	1133 [72–1205]
Threshold for radon anomaly [Mean ± 2SD] (Bq/m3)	598.5	870.3	723.4

), only three spike-like radon peaks (marked by a red arrow in Figure 3) meet the threshold condition and can be considered as significant radon anomalies. For each of them, the earthquake database for seismic events in the Balkan area [40] was analyzed according to both Equations (2) and (3) in order to assess if those radon anomalies could have been triggered by tectonic activity occurring in the forthcoming days or weeks. In such a tectonically active area where earthquakes occur weekly, it is quite difficult to relate radon anomalies to a given tectonic event, as close and small-magnitude events are potential candidates and so are further but bigger earthquakes. A selection of potential earthquakes is presented in

Table 2. Some earthquakes in the Balkan region [40], analyzed according to Figure 3 and Equations (2) and (3).

Location/Country	Coordinates N, E (°)	Date of Peak /Earthquake	D (km)	Error on D N, E (km)	H (km)	RD (km)	ML	MR	ΔRn (Bq/m3)
K.Mitrovica	42.823, 20.680	27 September 2023 6 October 2023	17.4	±2.144, ±1.551	2	4.5	1.5	2.55	583 *
Vučitrn	42.788, 21.048	2 November 2023 26 November 2023	19.1	±2.039, ±1.405	5	7.2	2.0	2.64	678
K.Mitrovica	42.891, 20.786	5 December 2023 13 December 2023	6.6	±2.178, ±1.715	2	4.9	1.6	1.54	521
Bosnia and Hercegovina	44.329, 17.903	11 December 2023 30 December 2023	283	±1.981, ±1.583	12	127.9	4.9	5.45	652
Albania	41.006, 19.829	9 January 2024 13 January 2024	230	±1.751, ±2.892	16	64	4.2	5.24	1008 *
Raška	43.299, 20.616	19 January 2024 27 January 2024	49.1	±1.372, ±1.302	3	11.9	2.5	3.63	891
Novi Pazar	43.244, 20.499	19 January 2024 10 February 2024	48.7	±1.567 ±1.259	4	10.8	2.4	3.62	891
Nikšić, Montenegro	43.007, 18.623	6 March 2024 14 March 2024	183	±1.658, ±1.929	12	231.7	5.5	5.0	616
Montenegro	43.029, 18.629	6 March 2024 3 April 2024	182.7	±1.486, ±1.564	15	86.1	4.5	5.0	616
Podujevo	42.877, 21.284	9 April 2024 17 April 2024	38.3	±2.064, ±1.760	5	5.4	1.7	3.37	599
Novi Pazar	43.112, 20.628	9 April 2024 10 May 2024	30.6	±1.023, ±1.080	4	11.9	2.5	3.14	599
Blaževo	43.240, 20.901	9 April 2024 10 May 2024	38	±1.420, ±1.263	6	13.1	2.6	3.36	599
Orahovac	42.535, 20.837	10 May 2024 13 May 2024	40.4	±1.607, ±1.349	10	16	2.8	3.43	913 *

D—Distance as the crow flies; H—Hypocenter; R_D—Strain radius according to Equation (3); M_L—Magnitude; M_R—Required magnitude for radon anomaly according to Equation (2); ΔRn—Difference between radon peaks; *—corresponds to the red arrow in Figure 3.

and further discussed hereafter.

A characteristic radon peak occurred at the end of September with a 583 Bq/m³ difference in radon concentration with respect to the seasonal baseline (Table 1) and a raw activity well above the decision threshold for more than 24 h. Interestingly, an earthquake occurred in Kosovska Mitrovica, which is in very close proximity, a few days later, on 6 October 2023. The radon peak on 27 September 2023 could tentatively be envisaged as a precursor of this forthcoming tectonic event. Nonetheless, both the strain radius R_D and the magnitude M_L of this event do not satisfy the empirical laws given in Equations (2) and (3) and an unambiguous relationship between the radon anomaly and the earthquake cannot be postulated (Table 2).

The analysis of earthquake occurrences in Albania (Table 2) also eliminates the potential event as a candidate due to the low strain radius of effective precursory manifestations (64 km). The third event (Orahovac) satisfied the radon anomaly criteria, but the distance from the measurement site and the event’s magnitude did not match. The potential reasons for radon peak occurrences are discussed below.

3.2.2. Analysis with Adjustment of Empirical Laws

At this point, none of the selected peaks appear to be earthquake-related. In addition, the applied empirical laws are nearly identical as they define almost the same field for the precursor manifestation zone and agree within 5% (Figure 4, right).Therefore, all satisfying earthquakes must be located above the curves. If the empirical law (lower lying curve on the graph) defined by Equation (2) is chosen to represent earthquake magnitude vs. distance as the crow flies, two of the recorded earthquakes(with M_L marked in bold, Table 2) appear above the curve (Figure 4). The K. Mitrovica event (13 December 2023), which is close to the monitoring site (6.6 km), is a real candidate that can be related to a radon anomaly, given the distance error of about 3.9 km, but without a clear radon anomaly within at least a month before it occurred (Figure 3). The second satisfying event, a 5.5-magnitude earthquake in Nikšić, Montenegro (14 March 2024, Table 2) was followed by serious aftershocks in the Balkan region. A clear association between this seismic event and significant radon peaks at the monitoring site was not identified until two months earlier, so this could be attributed to tectonic activities in the Earth’s crust before a stronger earthquake in the area.

If 10% and 20% uncertainties are applied to the parameters of Equation (2), then an enlarged precursor manifestation field includes three earthquakes characterized by an important magnitude (in Bosnia and Herzegovina, Albania, and Montenegro) and events close to the measurement site (Figure 4). In this regard, these distant events could have caused ground shaking. Although it is not entirely clear, a connection with radon anomalies that occurred months before should not be ruled out, as radon degassing precedes tectonic activity [22]. It now appears that earthquakes relatively close to the measurement site (Vučitrn and the last three events in Table 2) have been able to produce a radon anomaly. Positive outcomes are quite obvious for the Novi Pazar and Orahovac earthquakes. However, a clear well-defined radon anomaly (meeting the decision threshold presented in Table 1) may only be related to the Orahovac event, but as will be discussed below, may also reflect external meteorological forcing.

It is interesting to mention a series of four smaller earthquakes (magnitudes ranging from 1.5 to 2.5) that occurred within 9 h in Raška (27 January 2024) along the fault zone. Approximately eight days prior to these earthquakes, the difference in radon concentration was 891 Bq/m³. However, these events did not meet the criteria (Table 2, Figure 4) to be considered as potential candidates.

According to the above, it is very difficult to assess the time scale in which a produced radon anomaly could be a precursor to an earthquake and whether any radon fluctuation not related to anomalies could be an indicator of tectonic activity. Radon anomalies are not clear precursors of seismic activity, but seismic activity may contribute, to some extent, to radon anomalies [39].

Nevertheless, more cases with no clear trend between earthquakes and radon have been found. This may be attributed to meteorological conditions in the area, which significantly affect radon fluctuations. Radon spikes were significantly fewer and less marked at both the beginning of September to mid-October and at the end of the study period (April to June 2024). A striking exception to this pattern is the peak in May 2024, which might be due to rainy conditions at that time. This suggests a strong influence of meteorological conditions on radon emanation. Its mobility partly depends on the ideal gas law, meaning that both temperature and atmospheric pressure will play a major role in radon degassing in the atmosphere. Due to its solubility in water, precipitations will also control its release, either preventing or promoting its escape to free air depending on how water circulates in soils.

4. Conclusions

Long-term radon monitoring revealed a high average value of 303 Bq/m³ in the laboratory. A seasonal pattern was clearly observed; radon activities peaked during the autumn–winter season. The calculated dose of radon inhalation for students and employers in two academic semesters was higher than the annual effective dose recommended by the UNSCEAR. In this case, strategies aimed at reducing radon levels are necessary and desirable. In addition, radon monitoring is always important because even lower radon concentrations may cause harmful consequences on health.

Accordingly, radon measurements are a useful tool in investigations of geodynamical events related to ground shaking in the broader region. However, some unclear patterns were identified between radon levels and earthquakes. Radon anomalies observed during this study most likely derive from meteorological conditions and external forcing but it cannot be ruled out that the strong tectonic activity in the Balkans area contributes to the radon budget in addition to external forcing. In the future analysis of earthquakes, continuous monitoring data should be assessed, especially in the vicinity of active faults, in order to find a clear, causal connection between prompt radon releases and the occurrence of earthquakes.