Radon Dynamics in Granite and Calcareous Soils: Long-Term Experiments in a Semi-Arid Context

Abstract

Radon in soil poses a significant health risk when it accumulates inside dwellings. The estimation of radon potential is a difficult task due to the complex dynamics of radon within soil and its relations with the weather. This research focuses on the variability of radon activity, driven by environmental changes, assessed in two loam soils (loamy sand–granite soil and silty clay loam-calcareous soil) with different radium contents. We conducted an experiment with teow soil columns in a semi-controlled outdoor laboratory, in a warm semi-arid climate. We also examined the consequences of abundant rainfall on radon activity through artificial soil water content (SWC) experiment conditions. Statistical analyses reveal that SWC is the most significant parameter influencing radon activity in these experiments. Radon is proportional to SWC and inversely proportional to temperature, evapotranspiration, and pressure in both soils, while wind is negatively related only in the loamy sand soil. Based on our findings, we modelled radon potential considering different soils and climatic contexts. SWC influences radon potential by changing radon emanation, activity, and permeability, depending on the local soil texture and radium concentration.

1. Introduction

Radon is a noble gas that comes from the uranium decay chain. Specifically, ²²²Rn, the radiogenic isotope of ²²⁶Ra of the ³²⁸U chain, has the longest half-life (3.82 days) among radon isotopes, which makes it more useful for scientific research than the other two isotopes (²²⁰Rn and ²¹⁹Rn). Radon is widely used for understanding and predicting geological processes, such as volcanoes and earthquakes [1,2,3]. However, it also poses a health risk to humans, since the decay products of ²²²Rn are solid, radioactive isotopes that remain attached to breathable particles and continue to decay in lung tissue. As a result, radon is the second most common cause of lung cancer [4,5,6].

A major source of radon is the soil (geogenic radon), which may migrate into dwellings and become a potential health hazard when it accumulates indoors for a long period. To assess the radon risk, the geogenic radon potential is used, which is calculated from radon activity and soil gas permeability field measurements [7,8]. Natural ²²⁶Ra concentration in soils is also used to estimate radon instead of measuring it directly on the field. Geogenic radon potential is used for the creation of predictive indoor radon maps on a large scale, which provide a risk assessment of radon exposure [9,10,11,12]. Neznal et al.’s radon potential (RP) equation for building sites is the most used for a local scale [7]:

$$RP=\frac{C_{Rn}}{p{k}_g-10},$$

(1)

where C_Rn is the measured radon activity in soil air (Bq m⁻³) and pk_g = −log k_g (k_g is gas permeability in m²).

The dynamics of radon in soil are complex and conditioned by several factors, such as parent material, mineralogy, porosity, organic matter, texture, water content, and gas permeability [13,14,15,16,17]. On the one hand, the higher the uranium content of the parent material, the higher the radon activity. However, it is also necessary to consider the microscopic-scale distribution of uranium within the mineral grains resulting from the consecutive phases of pedogenesis [18]. On the other hand, soil physical properties such as pore size and soil water content (SWC) affect the emanation and diffusion coefficients of radon, and therefore soil gas permeability in unsaturated systems [19,20,21,22]. Therefore, changes in soil gas permeability have a key role in reducing or enhancing radon migration outwards, while soil type is a decisive factor when characterising radon in soil [23,24,25,26,27].

In addition, weather parameters play a fundamental role by affecting the soil parameters mentioned above, as well as conditioning the development of the soil itself. A similar soil parent material may derive into different soils due to, for instance, more humid or arid climates [28]. Also, this difference in climatic context conditions the daily and seasonal weather, affecting geogenic radon potential [29,30]. In the short term, barometric pressure, temperature, wind, and evaporation are weather parameters associated with changes in radon dynamics [31,32,33,34]. Generally, a local high wind speed, a decrease in barometric pressure, and/or an increase in temperature lead to a decrease in radon activity [12,35,36].

Furthermore, soil and environmental variables, such as water content, also modify CO₂ production and migration within the soil [37]. A great variety of studies in geosciences, including vulcanology and GHG soil emissions to the atmosphere or cave sites, among others, use CO₂ together with radon as tracer gases [38,39], even though they have different origins. Organic CO₂ is mainly formed in soils either by the proliferation and subsequent decomposition of microorganisms, or by the respiration of plants through their roots [40].

Many studies have been carried out to understand the relationship of radon with fractures and fault areas, and weather variables [41,42,43,44,45] with a closer perspective under controlled conditions—e.g., Haquin et al. [46], Nassiri-Mofakham et al. [47], and Papachristodoulou et al. [48]—but not deep enough to study the radon potential at a local scale.

Our research analyses the dynamics of radon for two types of soil, with different radium content and soil texture under a semi-arid climate (Alicante province, SE Spain). For this purpose, we developed an experimental procedure in semi-controlled, open laboratory conditions, where weather conditions and soil parameters were monitored.

First, recorded time series of all variables were examined to explore how weather and soil variables impact radon activity changes and to identify which ones are most decisive. Second, we performed an artificial soil water content experiment to further assess the effects of large rainfall on radon activity. Third, a detailed statistical analysis was performed to establish the correlations and weight of soil and weather variables related to radon. This analysis compared discrete rainfall events (with significant changes in radon activity) with the frequency response of long-term time series using wavelet analysis. Finally, based on our results, we evaluated radon activity and its response to weather parameters and soil properties in the context of radon potential estimation.

2. Materials and Methods

2.1. Weather Variables and Climate Series

According to the Köppen–Geiger classification, the study site has a Bsh hot semi-arid climate (hot steppe type; [49]). This climate classification defines climate scenarios on a global scale. Weather variables at the study site were provided by the Azotea IIG Station of the Climatology Laboratory-Interuniversity Institute of Geography of the University of Alicante, by a Davis Vantage Pro2 automatic weather station. The weather variables considered were relative humidity (%), barometric pressure (mbar), air temperature (°C), wind speed (m s⁻¹), rain (mm) and evapotranspiration (ET, mm). ET is calculated directly by the weather station with air temperature, environmental relative humidity, wind and radiation applying the Penman–Monteith Equation as implemented by CIMIS (California Irrigation Management Information System), which includes the calculation of net radiation.

For the period of monitoring, April 2021–February 2023, the average air temperature was 19.7 °C, reaching a maximum of 40.4 °C in summer and a minimum of 2.8 °C in winter. The average environmental relative humidity, barometric pressure and wind speed were 68.0%, 1017.2 mbar and 5.3 m s⁻¹, respectively. The total rainfall was 435.2 mm while the calculated ET for the same period was 950 mm.

2.2. Soil Properties

The soils studied were chosen by their clear differences in terms of mineralogy and, therefore, differences in terms of geogenic radon potential: a silt soil from a calcrete, common in the Alicante region (Spain; identified henceforth as calcareous soil) that does not belong to a priority action area or a high-RP area according to the radon potential map of Spain [10]; and a second soil of weathered granite origin (granite soil hereafter) which belongs to the central area of Spain, classified as a priority action area. Additionally, they also represent two very different soil typologies that are at the same time representative of the context of the Spanish country.

Soil samples were taken, for both soils, from horizon A and horizon B separately. Horizons represent soil layers with different characteristics. They are arranged vertically in a soil profile. After that, columns of 50 cm diameter and 110 cm depth were prepared. The soil profile differs in both soils: calcareous soil has an AC profile, while the granite soil has an ABC profile. Therefore, the calcareous soil column only had horizon A, while the granite soil was prepared with horizon A (layer 1) to complete the first 55 cm, and the remaining 55 cm with horizon B (layer 2), as at the original location.

To characterise both soils, soil samples were prepared by removing gravel and root debris, followed by sieving. Soils were oven-dried at 40 °C for 24 h to determine dry bulk density (ratio of the dry weight to the volume of a sample). The grain density (particle or real density) was determined using an AccuPyc 1330 Helium pycnometer (Micromeritics Instrument Corporation, Norcross, GA, USA). Total porosity was obtained from grain and bulk densities. Soil particle size distribution was determined by hydrometer analysis according to the Spanish standard [50], and the soil texture type was obtained by the United States Department of Agriculture (USDA) classification. The granite soil was classified as a sandy loam and loamy sand type (horizons A and B, respectively), while the calcareous soil was classified as silty clay loam soil. Their initial characteristics are summarised in

Table 1. Summary of the characteristics of the soils under study.

Initial Parameters	Calcareous Soil	Granite Soil
		Layer 1 Horizon A	Layer 2 Horizon B
Bulk density (g cm−3)	1.25	1.68	1.55
Porosity (%)	52	35	42
Soil texture:	Silty clay loam	Sandy loam	Loamy sand
(%) sand	18	72	80
(%) silt	51	22	14
(%) clay	31	6	6
238U (Bq kg−1)	21.7	53.0	48.0
226Ra (Bq kg−1)	16.9	39.7	44.4

.

The determination of ²²⁶Ra and ²³⁸U radionuclides, performed by gamma spectrometry, demonstrates the higher uranium and radium content of granite soil compared to calcareous soil. The mineralogical composition, obtained with powder X-ray diffraction with a Bruker D8-Advance diffractometer (mirror Goebel, CuKα radiation—40 kV, 40 mA—2θ: 3–60°, step size of 0.05°, scan step of 3 s; Bruker, Billerica, MA, USA), showed that the calcareous soil is dominated by calcite, with a minor amount of quartz, dolomite, feldspar, and clay minerals. The mineralogical composition of horizons A and B of the granite soil is similar, composed of quartz, plagioclase, and feldspar. Horizon A presents higher amounts of clay minerals—e.g., kaolinite—derived from the alteration of a leucocratic granite, whereas horizon B shows a higher amount of muscovite.

2.3. Monitoring System

The experiments were performed in two experimental columns that contained the soils under study (Section 2.2). The columns were placed outdoors at the campus of the University of Alicante (Alicante province, SE Spain). This study was carried out in different periods depending on the soil: April 2021 to July 2022 for the granite soil, and July 2022 to February 2023 for the calcareous soil. Two different measurement periods were established because radon could only be measured continuously in one soil at a time.

The experimental procedure was identical for both columns. The columns consisted of a 110 cm high corrugated polyethylene tube with a 50 cm diameter (Figure 1). Several sensors controlled by dataloggers were installed providing continuous data on soil parameters through the studied period. The monitoring system consisted of a set of dataloggers (HOBO H22 (Onset, Bourne, MA, USA) and CR-1000 Campbell Scientific^®—Logan, UT, USA) connected to the power supply and to a backup battery to prevent data loss. In each column, there were two different measurement points (at different depths; top and bottom, Figure 1) to continuously register the following at hourly intervals: (i) dielectric permittivity (to later calculate soil water content (SWC); accuracy: ±0.03 m³ m⁻³ [51]); (ii) soil temperature (accuracy: ±1 °C) using a 5TE automatic sensor (Decagon Devices, Pullman, WA, USA) connected to the CR-1000; and (iii) CO₂ soil concentration GMP252 Vaisala CARBOCAP sensor (reading accuracy: ±2%; Vaisala, Vantaa, Finland), with a 0–10,000 ppm measurement range, connected to the HOBO H22. Note that the operating range of the CO₂ sensor conditioned the measurement in the periods of highest soil CO₂ production, mainly spring, since the probe was saturated, and real values were not available.

In addition, at the bottom of each column (Figure 1), radon activity was continuously measured at 30 min intervals with a Barasol MC2 silicon detector (accuracy: ±50 Bq m⁻³; Algade, Bessines-sur-Gartempe, France). This device was placed in a PVC-capped tube at the bottom of the column at the target soil depth of 90–100 cm, as radon and gas permeability measurements are usually performed at a 100 cm depth [50]. This is an adequate depth to avoid dilution of radon in soil due to mixing with atmospheric radon–poor air.

2.4. Sampling System

Discrete sampling was carried out to calibrate the continuous series of radon and CO₂. A gas sampling tube was installed in each column at a depth of 90–100 cm—to achieve sampling at the target depth—to extract air from the soil (Figure 1).

Soil air samples were pumped out using a micropump and a 12 V battery and then collected in 2 L RITTER opaque gas sampling bags with lock valves (RITTER, Bochum, Germany). They were analysed with AlphaGUARD DF2000 (accuracy: ±3 Bq m⁻³; Bertin Technologies SAS, Montigny-le-Bretonneux, Paris, France) at a 0.3 L min⁻¹ pump flow and 1 min flow mode, for 15 min. The first six minutes of measurement, related to the time the gas sample needs to enter AlphaGUARD, were discarded and only the remaining 9 min were considered. This procedure only considers the measurement of ²²²Rn and avoids the ²²⁰Rn contribution (²²⁰Rn half-life is 55.6 s), and therefore, ²²²Rn values recorded with AlphaGUARD are comparable to those with the Barasol device (which measures only ²²²Rn).

The median radon value was corrected after measuring to compensate for the radon loss due to radioactive decay. Equation (2), based on the solution of Bateman equations, was used:

$$C_{Rn}=C_0{e}^{-\mathsf{\lambda }t},$$

(2)

where C_Rn (Bq m⁻³) is corrected by applying λ = 0.00761 (h⁻¹) for a particular time (t, h) [52] to obtain the actual radon activity in soil (C₀, Bq m⁻³).

Radon exhalation rate (E, Bq m⁻² s⁻¹) was evaluated since it provides information on how soil releases radon to the atmosphere, which varies according to the soil type. We used an AlphaE sensor (sigma/24 h uncertainty: ±12%; Bertin Technologies SAS) inside a dust bag placed within a polymethyl methacrylate accumulation chamber (20 cm inner diameter and 0.005 m³ of effective volume). The accumulation chamber was placed on top of a PVC collar, completing the exhalation box, which was directly connected to the soil. Radon exhalation is calculated as follows (Equation (3)):

$$E=\frac{V}{S}\frac{C_{Rn}}{t},$$

(3)

where V (m³) and S (m²) are the effective accumulation volume and the exhalation surface of the soil within this accumulation chamber, respectively, at a particular time t (s).

Finally, soil gas permeability was measured with RADONJOK (Radon V.O.S., Praha, Czech Republic), and RP was estimated by applying Equation (1) by Neznal et al. [7]. Some discrete measurements of CO₂ concentration were also performed using a CRDS analyser (Picarro G2201-i, Santa Clara, CA, USA).

2.5. Artificial Soil Water Content Experiment

After continuous monitoring of both soils, we artificially increased the soil water content, aiming to simulate the natural water content increase associated with heavy rainfalls or flood irrigation.

The experiment involved the artificial addition of water in the granite (60 L) and calcareous (43 L) soil columns. In the case of the granite soil, the 60 L was added in 3 stages of 20 L each: 20 L at the beginning of the experiment, 20 L after 6 h, and 20 L after 22 h. For the calcareous soil, 43 L of water was added in 20 L at the beginning, 13 L after two days, and 10 L after five days. The experiments for the two soil columns were not simultaneous due to the use of a single measuring device. Thus, as the water was added to the soil, the continuous monitoring system installed in the soil columns registered the soil variables explained in Section 2.3, as well as the radon exhalation rate (Figure 1). In these experiments, discrete sampling and continuous monitoring were carried out according to the methodology explained in Section 2.2 and Section 2.3. A smoothing analysis was applied to the continuous series of radon activity, using the Adjacent-Averaging method with 15 window points.

2.6. Time Series Analysis

The analysis of radon activity alongside soil and weather variables considered the detection and analysis of discrete events with significant variations in radon activity, and the frequency response of the variables by wavelet analysis.

The statistical analyses of the discrete events were carried out in two steps. First, we defined the discrete events when significant variations in radon activity occurred. The analysed events were selected within the time series because they experienced significant variations in radon activity, generally 0.15% higher than average levels. We then evaluated all variables before and after rain episodes to find out which were the most decisive on soil properties and radon activity. This study examined granite soil because it presents a sandy texture, and consequently, major variations in all variables were expected due to changes in weather conditions.

Based on the results of the previous analysis, the second step consisted of a global analysis of the time series for both soils by selecting periods with no gaps in the continuous time series. To develop these analyses, the average daily values were obtained from the hourly recorded values of all measured variables.

Principal component analysis (PCA) and multiple linear regression analysis (MLR) were applied to the obtained datasets. PCA reduces the size of the number of variables initially considered and establishes associations between those variables that are organised into components. In the MLR analysis, radon in soil acts as a dependent variable, whereas soil and weather parameters serve as independent variables. This analysis also includes the weight (magnitude of the standardised coefficients, Beta) of each independent variable in the calculation of multiple linear equations, aiming to quantify the influence of each variable on radon activity. The p-value is obtained to evaluate which variable is most significant in the fitted regressions. The lower the p-value, the greater the statistical significance of the variable analysed. Statistical analyses were performed using the SPSS v.24.0 (SPSS Inc., Chicago, IL, USA).

To complement the analyses and assess the influence of soil variables on radon activity, the cross wavelet transform (XWT) and the wavelet transform coherence (WTC) were applied between pairs of signals (one being the soil radon activity). This time–frequency analysis allows us to examine the interrelations between two time-domain signals (explained by XWT) and their coherence (WTC), leading to the identification of areas with high common power between the pair of signals in the final scalograms [53,54,55,56]. Wavelets are a tool that helps in detecting periodic patterns in radon activity, which may be related to meteorological or environmental parameters [57]. For this analysis, the Environmental Wavelet Tool (EWT) MATLAB-based code [58] was used, which incorporates the package developed by Grinsted et al. [59]. This analysis was performed for the period October 2021–July 2022 for granite soil and from July 2022 to January 2023 for calcareous soil. These periods were established according to the data requirements for performing the wavelet analysis.

3. Results

3.1. Time Evolution of the Radon Activity and the Main Weather Parameters

Figure 2 (granite soil; April 2021–July 2022) and Figure 3 (calcareous soil; July 2022–February 2023) show the temporal evolution of all variables involved.

3.1.1. Granite Soil

The mean value of radon was 4215 ± 960 Bq m⁻³ (Figure 2). Radon activity was generally below this average in the dry season (mainly summer) and usually higher than this value in the rest of the seasons. Radon was close to its maximum with multiple peaks due to radon oscillations often related to rain (Figure 2 (e)). The average concentration of CO₂ was 5414 ± 2284 ppm and 4502 ± 2462 ppm (upper and lower layer, respectively; Figure 2), and its highest concentration also corresponds to the rainy periods (mainly spring) with a maximum concentration of 28,994 ± 423 ppm (discrete sampling). The mean soil temperature was 19.3 °C (Figure 2), and it showed diurnal and seasonal patterns. Soil temperature undergoes a large-scale tendency to increase from the end of winter onwards, reaching the highest temperatures in summer, and then gradually decreasing until winter (38.5 °C and 2.8 °C, respectively). The most significant diurnal temperature contrast occurs during the winter–spring months (Figure 2). The mean soil water content was 0.074 m³ m⁻³ and 0.058 m³ m⁻³ (upper and lower layer, respectively; Figure 2). The gas permeability of the granite soil has a typical value of 10^−10.63 m² (pk_g = 10.63; where pk_g = −log k_g (m²)) measured at SWC = 0.058 m³ m⁻³. It can be classified as very permeable according to Neznal et al. [7].

3.1.2. Calcareous Soil

Compared to the granite soil, the variation in radon in calcareous soil over time was slightly less sensitive to changes. The mean radon activity was 1363 ± 814 Bq m⁻³ (Figure 3), and its activity was also strongly dependent on the soil water content. The CO₂ soil concentration has a mean value of 1471 ± 485 ppm and 2328 ± 415 ppm (upper and lower, respectively; Figure 3), and presents lower continuous values than granite soil. The average soil temperature was 20.6 °C, with the highest and lowest values in summer and winter (39.5 °C and 3.4 °C, respectively). The mean soil water content (SWC) was 0.126 m³ m⁻³ and 0.120 m³ m⁻³ (upper and lower measurements, respectively; Figure 3). The percentage of sand, silt, and clay (% SSC) in the calcareous soil behaves differently than in the granite soil, and consequently, the calcareous topsoil experienced considerably less variation after rain episodes (Figure 3 (a); October 2022). For an SWC value of 0.101 m³ m⁻³, the calcareous soil was less gas-permeable (pk_g = 10.79) than granite due to its fine texture and larger SWC.

3.2. Influence of Weather and Soil Parameters on Radon Dynamics

The impact of weather factors on radon was obtained from the analysis of continuous time series (Figure 2 and Figure 3). In granite soil, the series began during the early spring rains of 2021, which resulted in elevated radon activity (above the mean 4215 Bq m⁻³) concurrent with the maximum CO₂ (Figure 2 (a)). Fluctuations in surface soil water content (SWC), mostly below the average (0.074 m³ m⁻³), showed discrete peaks at the end of May 2021 (Figure 2 (b)), aligning with a significant rainfall episode. Thereafter, radon activity decreased in late summer, coinciding with a reduction in precipitation and an increase in temperature (Figure 2). In autumn, radon peaks increased during rainy periods. Upper-layer SWC values rose with autumn rainfall, along with the highest radon peaks (Figure 2 (c)). Wind gusts from late autumn 2021 until early winter 2022, without rain episodes, are related to radon fluctuations (Figure 2 (d)). Radon stabilised above the mean after January 2022. March 2022 rains led to a substantial increase in upper-layer SWC, while values in the lower layer remained unchanged (Figure 2 (e)). Radon and SWC increased, reaching their highest levels in April and May 2022. The radon peak in May 2022 coincided with the second-highest rainfall event, leading to a notable increase in CO₂ (Figure 2 (e)). At the beginning of June 2022, radon decreased again, in conjunction with a decrease in CO₂ and soil moisture, while temperatures rose.

As observed in the granite soil, from July 2022 to the beginning of January 2023, temperature and SWC in the calcareous soil gradually decreased in the seasonal trend while the radon increased (Figure 3). SWC was affected by the most important rainfall event of the series (Figure 3 (a)). Rain events during mid- and late-autumn 2022 caused a notable impact on radon activity, particularly following two significant rainfall episodes in October and November 2022 (Figure 3 (a,b)). Then, the wind gusts coincided with a decline in radon. Additionally, the late autumn period (Figure 3 (c), concluding the December 2022 rain events) showed a low-pressure trend, aligning with a slight increase in radon activity. In winter 2022, strong wind gusts were recorded along with a decrease in radon activity (Figure 3 (c)).

The two continuous soil series clearly show how the weather variables influence radon dynamics, especially in granite soil. Both SWC and temperature were identified as conditioning variables, and SWC was the parameter likely to exert the most significant influence on radon activity [43,60]. The water content is also analysed in Section 3.3. High temperatures in summer (i.e., also higher ET) are seasonally associated with low SWC values and low radon activity [61,62].

In fact, the response of the radon variations to the SWC depends on the initial SWC in each seasonal period. Therefore, for a semi-arid climate such as Alicante, radon activity generally increases more with longer periods of rainfall events (Figure 2 (e)). Thus, maximum values of radon peaks occurred during periods of high soil moisture linked to rainfall, as Pinault et al. [63], Tareen et al. [64], and Naskar et al. [65] described. In the calcareous soil, the occurrence of maximum values of radon peaks following rainfall events was delayed when compared to the granite soil, likely due to its loamy and clayey texture (Table 1).

Both soils exhibit different responses to rainfall events related to SWC variations closer to the surface and deeper layers. Close to the soil surface, SWC achieved higher and more variable values, compared to the more stable SWC in the lower layer, less influenced by weather conditions [30,31]. In the case of calcareous soil, the SWC of the deeper layer was less affected and remained nearly constant after rainfall events, which lasted from a few hours to days until the water reached a peak in the lower SWC layer. This was particularly evident during continuous rainfall periods (Figure 2 (e)).

The increase in water in the upper soil layer reduces soil gas permeability. Consequently, radon activity rises in the lower soil layer due to the combined effects of recoil and the gas being trapped in the soil under a water-saturated layer [12], which reduces gas transport [14,26,66,67,68].

CO₂ fluctuations also respond to SWC variations, similar to radon, and although radon and CO₂ have different sources, they are commonly used as tracer gases in various geoscience disciplines [38,39]. In our context, the concentration of CO₂ is linked to microbiological activity and is dependent on water content and temperature [69,70]. Both CO₂ and radon increased with SWC, mainly in spring and autumn (Figure 2 (c,d) and Figure 3 (a)). They also share a similar downward trend at the end of spring and during the summer period (Figure 2 (e)), regarding an increase in soil temperature and evapotranspiration, and a drop in microbiological activity (in case of CO₂).

Barometric pressure and high wind velocities can influence radon activity, although their impact is generally considered less significant. They have a more pronounced impact on radon activity in highly permeable soils [12]. However, in the current time series, SWC variations related to rain events do not allow a proper visual evaluation of their influence on radon activity changes. Large-scale barometric pressure changes and high wind velocities correlated inversely with radon activity (Figure 2(d) and Figure 3 (c)). The decrease in barometric pressure draws soil air to the surface, affecting the radon activity in deeper soil layers. Similarly, high wind velocities locally depressurise, and radon activity decreases [35].

3.3. Artificial Soil Water Content Experiment

The results obtained in the previous analyses highlighted the relevance of water content on soil radon activity in the studied soils. For this reason, we forced soil water conditions with an experiment that was developed to better understand the relationship between soil water content and soil radon variations.

In granite soil, soil temperature showed natural daily temperature variations, while the SWC changes were entirely consequences of the artificial addition of water (Figure 4). Water content was first detected at the upper sensor 2 min after the first water addition. It flowed downwards by water infiltration, and it was detected in the lower measurement point 4.03 h after the first water addition. The different SWC peaks (0.230, 0.230, and 0.246 m³ m⁻³ in the upper layer; 0.144 and 0.145 m³ m⁻³ in the lower layer; Figure 4) are a consequence of the different water additions. As the soil becomes saturated, the water infiltration is constant from the upper to the lower point of the soil column.

Radon activity at the beginning of the experiment (3301 Bq m⁻³) was below the average measured in continuous monitoring (4215 Bq m⁻³; Figure 2). With the addition of water, radon activity sharply increased, reaching up to 9170 Bq m⁻³ (Figure 4). This increase occurred in a period ranging from 14 to 22 h from the beginning of the experiment. Afterwards, radon activity remained stable for approximately 4 days, from which it decreased and returned closer to its initial values (2266 Bq m⁻³). After the fifth day of the experiment, the continuous measurement of radon was replaced by discrete daily measurements (Figure 4) that confirmed the reduction in its activity over time. Concerning CO₂, a similar pattern is described, with an increasing concentration in response to the addition of water to the soil. Both gases reached maximum values as a fast response to water addition and showed a drop nearly four days after the beginning. Maximum concentrations of CO₂ (greater than 14,000 ppm) were not measured by the continuous probes but by taking discrete samples.

The radon exhalation rate shows a decrease with respect to the value measured before the addition of water (first measurement of the series; Figure 4). As the soil porous network was filled with water, radon could not migrate through the soil pores towards the atmosphere. Consequently, exhalation values recorded on the soil surface are the lowest with maximum water contents.

Before conducting the artificial soil water content experiment, pk_g was 11.00 (k_g = 10^−11.00 m² is considered high gas permeability according to Neznal et al. [7]). After the second water addition, soil gas permeability decreased to its minimum (pk_g = 11.46). This coincided with the growth of radon activity and its highest recorded level. Gas permeability has a clear variation throughout the experiment (from 10^−11.00 to 10^−11.46 m²). In addition, there was a notable difference in soil temperature between the beginning and the end of the experiment, considering both minimum and maximum (maximums began around 15 °C, while at the end, they were close to 25 °C; Figure 4). This was accompanied by a decrease in radon activity, which was in concordance with the soil drying process and the removal of water from the pore space, allowing the migration of radon outwards.

Figure 5 shows the artificial soil water condition experiment performed in the calcareous soil, which displays different soil variable dynamics compared to the granite soil, especially regarding radon. Soil temperature was also influenced by the daily air temperature oscillations and the gradual warming trend. Water added to this soil was less than that in the granite due to significantly reduced vertical infiltration through the soil column, related to its field capacity. Consequently, water remained stagnant on the soil surface, causing ponding. This caused all calcareous soil variables to evolve more slowly.

Water infiltration was detected in the upper sensor 4 min after the first water addition, while the detection in the lower sensor occurred 2 days and 2.2 h later. Thus, the maximum SWC in the bottom layer (nearly 0.20 m³ m⁻³) was lower compared to the upper soil layer (nearly 0.30 m³ m⁻³; Figure 5). However, the variation in SWC in the lower layer did not cause a change in gas permeability, as with granite (pk_g remained nearly constant at 10.73).

Radon activity shows two pulses: one during the period of water addition and another just after this period. In the first one, radon activity increased progressively until it reached a peak on day 5 (3930 Bq m⁻³; Figure 5), then decreased and returned to values similar to the beginning. The second pulse recorded a similar trend but with slightly higher activities, reaching an overall maximum on day 13 (5292 Bq m⁻³; Figure 5). Between days 13 and 14, the maximum activities occurred, and then radon decreased and began more pronounced fluctuations.

The latter coincides with a barely perceptible decrease in the Upper SWC, while the Lower SWC remains constant at a value that shows a double contrast to that at the beginning. This in turn coincides with the increase in temperature.

Radon exhalation rates were lower compared to the granite soil, even slightly lower after starting the experiment because of stagnant surface water, as well as the radium content and the water retention related to its texture. Finally, regarding CO₂ concentration, it gradually increased until the highest radon activity was reached (day 13). From then on, CO₂ showed similar dynamics to radon but with a larger amplitude between peaks and drops. It reached its maximum recorded level 21 days after the beginning of the experiment.

The experiments showed that there was more water than air in the pore system, mainly in the upper layer. This increases radon emanation [71]. When the water content is very high, gas permeability decreases and radon is diluted in water, resulting in no soil–atmosphere exchange (radon exhalation rate decreases). In lower layers, radon activity increases until the upper layer is no longer saturated as the soil dries out, favoured by rising temperatures. It is at this point that the gas permeability increases (increased exhalation rate [72]), and radon activity decreases due to exchange with the atmosphere. In the case of calcareous soil, the process is more complex, conditioned by a lower content of sand-sized particles and a higher percentage of silt and clay.

4. Discussion

This section aims to discuss the data recorded from both soil time series in order to understand the processes and variables affecting radon dynamics. First, the statistical determination of the relationship between radon variability and the other variables is addressed. According to the results obtained, the granite soil was initially selected to study discrete rain events with significant variations in radon because its texture allowed us to obtain clearer evidence of changes in radon and the other variables. Based on this analysis, we conducted a global analysis of the time series for both soils at daily intervals. Additionally, the long-term radon activity pattern was characterised using wavelet analyses to evaluate the interrelations between pairs of variables. Finally, based on our results, we evaluated and estimated radon potential estimation related to radon response to weather parameters and soil properties.

4.1. Time Series Analyses

4.1.1. Statistical Analyses of the Granite Soil

This analysis considered the detection of discrete events with significant variations in radon activity related to rainfall episodes. The main variables associated with these events are initial radon activity (Rn₀), maximum radon reached after the rain episode (Rn_max), soil water content in the upper soil layer at Rn_max (Upper SWC_{Rn max}), pressure at Rnmax (P_{Rn max}), soil temperature at Rn_max (Soil T_{Rn max}), wind speed at Rn_max (Wind_{Rn max}), and accumulated rainfall (Rain).

The principal component analysis (PCA) of these variables defined two principal components (PCs), which accounted for 62.3% of the total variance (

Table 2. Principal component analysis (PCA) and multiple linear regression (MLR) of discrete events with significant variations in radon related to each rainfall episode in the granite soil. PC corresponds to principal components in the PCA. B and Beta are the coefficients and the standardised coefficients of the MLR, respectively, with their associated p-value. Significant codes: p-value: 0.05 ‘*’; >0.1 ‘-’. Rn_max acts as the dependent variable.

Variables	Principal Component Analysis		Multiple Linear Regression
	PC1	PC2	B	Beta	p-Value
Rnmax	0.871	−0.007	65,674.43		*
Rain	0.654	0.225	77.94	0.31	*
Rn0	0.702	−0.274	0.34	0.33	*
Upper SWCRn max		0.161	−515.95	−0.03	-
PRn max	−0.399	−0.652	−58.84	−0.35	*
WindRn max	0.024	0.780	60.63	0.14	-
Soil TRn max	−0.456	0.754	−88.78	−0.43	*

). PC1 links Rn_max with initial Rn₀, and Rain with Upper SWC_{Rn max}. Thus, the Rn_max reached after the rain episode depends on the initial radon activity before the rain event. As described previously in Section 3.2, rain mainly defines SWC in the upper soil layer and is the weather factor that affects the most radon variations. PC2 associates weather parameters Wind_{Rn max}, P_{Rn max} and Soil T_{Rn max} (similar to air temperature pattern), and they act differently when radon reaches its maximum after the rain episode.

Multiple linear regression (MLR) analysis aims to evaluate the significant weight of each variable (acting as independent variables) on the estimation of the maximum radon reached after the rain episode (acting as the dependent variable). In other words, it is a qualitative approach to evaluate which variable affects Rn_max the most, rather than providing a predictive equation. The combination of all the variables considered (Table 2) to estimate Rn_max leads to an excellent goodness-of-fit, with Pearson’s coefficient, R, being 0.816. In the MLR, Rn₀, P_{Rn max} and Soil T_{Rn max} can be considered significant variables in the estimation of Rn_max (Table 2). This analysis focuses on the variation in radon related to each rainfall episode in granite soil, where Wind_{Rn max} does not significantly influence Rn_max dynamics.

Based on these results related to rainfall episodes, we analysed the entire time series at a daily frequency. The dataset contains the following variables: radon activity (Rn), soil water content in the upper soil layer (Upper SWC), soil temperature (Soil T), wind speed (Wind), and pressure (P). Moreover, here, we considered evapotranspiration (ET). Rain was excluded from these analyses because it was discrete throughout the entire time series. Instead, SWC, which represents rainfall effect on soil, was evaluated.

In this case, PCA results in three components that explain 85.6% of the variance and show that Rn was found in two principal components (

Table 3. PCA and MLR of entire time series at a daily frequency in the granite soil. PC corresponds to principal components in the PCA. B and Beta are the coefficients and the standardised coefficients of the MLR, respectively, with their associated p-value. Significant code: p-value: 0 ‘****’. Rn acts as the dependent variable.

Variables	Principal Component Analysis			Multiple Linear Regression
	PC1	PC2	PC3	B	Beta	p-Value
Rn	−0.639	0.613	0.379	22,690.39		****
Upper SWC	−0.329	0.890	−0.050	9834.87	0.60	****
Soil T	0.886	0.063	0.401	−53.09	−0.57	****
Wind	0.249	0.245	−0.911	−173.26	−0.57	****
P	−0.580	−0.532	0.100	−17.35	−0.13	****
ET	0.795	0.325	0.196	5210.02	0.25	****

), PC1 and PC2. In PC1, Rn was inversely related to Soil T and ET, whereas it was positively related to barometric pressure. As ET increases, SWC decreases, causing radon to migrate from the soil into the atmosphere. Conversely, an increase in P retains soil air in the lower soil layer. In PC2, radon (Rn) was positively related to Upper SWC and inversely related to P, highlighting the importance of SWC in the accumulation of radon at the lower layer. PC3 only includes Wind and it was not directly related to Rn.

The MLR analysis found that all variables are significant, considering the daily dataset (Table 3). Pearson’s coefficient is 0.859, and it considers Upper SWC, Soil T, and Wind to have the most statistical weight in the model (higher standardised coefficients—Beta values). Despite PCA concluding that Rn and Wind act differently, the MLR analysis highlights that Wind was a weather parameter that must be considered on a long-term scale in these soils and the climate context.

Different studies have demonstrated that the correlation of radon activity with meteorological variables is complex and does not follow a specific pattern, being dependent on the study site [72,73,74]. Regarding similarities with this granite soil, Papachristodoulou et al. [45] and Tareen et al. [64] also described a positive correlation with pressure, and Turk et al. [75] observed a positive correlation with water content in soil.

4.1.2. Statistical Analyses of the Calcareous Soil

The PCA of the calcareous soil explains 67.1% of the variance, described by two principal components (

Table 4. PCA and MLR of the entire time series at a daily frequency in the calcareous soil. PC corresponds to principal components in the PCA. B and Beta are the coefficients and the standardised coefficients of the MLR, respectively, with their associated p-value. Significant codes: p-value: 0 ‘****’; 0.01 ‘**’; >0.1 ‘-’. Rn acts as the dependent variable.

Variables	Principal Component Analysis		Multiple Linear Regression
	PC1	PC2	B	Beta	p-Value
Rn	−0.499	0.320	−12,020.20		****
Upper SWC	0.790	0.397	3982.40	0.19	-
Soil T	0.929	0.229	−4.19	−0.13	-
Wind	0.096	−0.830	9.94	0.10	-
P	−0.537	0.489	12.86	0.29	**
ET	0.867	−0.028	−2128.16	−0.28	-

). Rn was included in PC1, although its loading coefficient was low (−0.499). In other words, it is not well described in PC1, which hinders the correlation with other variables. Therefore, Rn is positively correlated with P (although it has a low loading coefficient, like in the granite soil) and inversely correlated to Upper SWC, Soil T, and ET. PC2 only contains the Wind variable.

MLR analysis reveals a low linear correlation between the variables studied, as Pearson’s coefficient for the MLR is 0.442. Most of the variables have low statistical weight in the estimation of radon. However, the sign of the coefficients in the MLR equation (B) clarifies the relationship between Rn and soil and weather parameters. Thus, Rn was positively correlated with Upper SWC, Wind, and P, and negatively correlated with Soil T and ET, which is in agreement with the general trend of the calcareous soil in the time series.

As calcareous soil has a silty clay loam texture, its hydraulic properties and water retention are different from granite (loamy sand soil, Figure 2, Figure 3, Figure 4 and Figure 5), as demonstrated in both the long-term time series and the experimental rain study. Furthermore, the time series in the calcareous soil is shorter than in the granite soil, and it does not include rainfall events that could affect the SWC comparably to the granite soil. Consequently, these results are not as conclusive as those observed in the granite soil, which emphasises the need for individual analysis of each soil for a better understanding of the radon potential.

4.1.3. Wavelets

For the granite soil, the results of wavelet analyses confirmed the influence of soil water content, soil temperature, evapotranspiration rate, pressure, and wind speed on radon (Figure 6 and Figure 7), also highlighted in the previous analysis.

For the period selected to perform the analysis, important changes in SWC caused a response in radon activity (Figure 7a,b). In the granite soil, rainfalls of November 2021, March–April 2022, and May 2022, which caused an increase in SWC, resulted in an evident increase in radon activity, in all cases occurring simultaneously. Therefore, the individual analysis of both signals established a predominant periodicity from one to several days with the rainfall episodes when they occurred (Figure 7a,b). Within these episodes, SWC and radon activity coincided in the time–frequency domain, showing interrelations explained by cross wavelet transform (XWT) and its wavelet transform coherence (WTC). Furthermore, the effect of pressure and wind variations on radon was also highlighted in the analysis (Figure 6e,f and Figure 7g–j) with marked areas with high common power in the scalograms. These areas are higher in the radon–pressure analysis, indicating a larger influence of these variables [76]. Concerning the influence of soil temperature and ET (Figure 6c,d and Figure 7c–f), the wavelet analysis does not reveal conclusive results, as common areas with strong power (in the case of radon–Soil T or radon–ET) were not detected in the time–frequency domain.

For the calcareous soil, wavelet analysis (Figure 6 and Figure 8) identified areas with high common power between pairs of signals, which might be significant in the case of pressure or wind (Figure 6k,l and Figure 8g–j). For these variables, XWT and WTC identify areas of high common power that coincide with the results of the individual analysis of the different signals (Figure 8e,f) in which the different predominant periodicities are established. The scalogram identifies common behaviours between coincident periodicities that exist for these variables and radon. However, SWC’s interrelation with radon (XWT; Figure 8a) is weaker compared to granite soil. It appears continuously for the time range in the 64-day period band, but neither in the WTC (Figure 8b) nor in the individual SWC analysis (Figure 6a). No significant common power areas are identified for soil temperature and ET (Figure 6i,j and Figure 8c–f).

4.2. Radon Potential

The detailed analysis of long-term time series has demonstrated that radon dynamics are largely determined by weather parameters and soil properties, and their interactions. Soil water content (SWC) was the parameter likely to exert the most significant influence on radon in soil air. However, its response to rainfall events and variability within the soil depth depends on the type of soil. Barometric pressure, temperature, and wind also affect radon soil activity, although their impact was less significant.

Thus, processes that alter radon activity and its mobility in the soil have an impact on the radon hazard assessment. As previously introduced, the geogenic radon potential can be considered an indicator of radon activity hazard, since radon gas exhalating from the soil is the main contributor to indoor environmental accumulation. It conceptualises ‘what Earth delivers in terms of radon’ from geogenic sources—e.g., radionuclides in soils, rocks, and fractures—to the atmosphere [9,77]. The widely used Neznal equation for the radon potential of a building site (Equation (1) [7]) includes soil gas permeability and radon activity, and it leads to a radon index class. Therefore, soils with high radium and uranium content and high permeability present high RP values, implying a higher probability of indoor radon infiltration due to geological factors. RP can be classified as low (RP < 10), medium (10 ≤ RP < 35), and high (RP ≥ 35), according to the Neznal et al. [7] classification.

The soils studied are classified as low risk, due to the relatively low concentrations of radium and uranium (Table 1). Initially, we chose granite soil, because it belongs to a radon-prone area, to compare it with the local calcareous soil. However, the radon activity recorded in the long-term series, the gas permeability, and the warm semi-arid climate define this soil as low risk. Consequently, the RP assessment could not be performed properly. Nevertheless, our findings on radon dynamics and their response to changes in the environment and soil properties can be used to parametrise and model RP variations. Soil water content and soil texture are quantitatively related to radon activity and gas permeability, and therefore to radon potential.

Radon activity can be estimated from radium activity, since ²²²Rn is the ²²⁶Ra decay product. Therefore, the theoretical radon activity can be obtained by the following consolidated expression (Equation (4))—e.g., Åkerblom [78], Giustini et al. [79], and Voltaggio et al. [80]:

$$C_{Rn}=C_{Ra}·\epsilon ·{\rho }_b,$$

(4)

where C_Ra (Bq kg⁻¹) is ²²⁶Ra concentration, ε is the emanation coefficient (-), and ρ_b (g cm⁻³) is the bulk density. The radon emanation coefficient in soil is a function of soil texture, water content, and temperature, and it is obtained from Equation (5) [81,82]:

$$\epsilon ={\epsilon }_0·\left[1+a\left(1-\mathrm{e}\mathrm{x}\mathrm{p}\left(-b·S_W\right)\right)\right]·\left[(1+c)·T\right],$$

(5)

where ε₀ is the radon emanation coefficient at 25 °C for dry soil; S_w is the water saturation (-); T (°C) is the soil temperature; and a, b, and c are three parameters fitted that depend on the soil properties (

Table 5. Fitted parameters and ²²²Rn emanation coefficient for clay, silt, and sand [82].

Coefficient	Clay	Silt	Sand
ε0	0.10	0.14	0.18
a	1.85	1.73	1.53
b	18.8	20.5	21.8
c	0.012	0.01	0.011

). If, as a first approximation, the residual (volumetric) water content is considered zero and the saturated (volumetric) water content is equal to the porosity, ϕ, then S_w = SWC/ϕ. Radon emanation coefficients can be estimated for any type of soil texture considering a weighted sum of % SSC.

Table 5 shows the % SSC for typical clay, silt, loam, and sandy soils. For example, according to Equations (4) and (5), sandy soil has a radon activity of 15.60 kBq m⁻³, whereas clay soil has 9.54 kBq m⁻³ (for soil with ρ_b = 1.7 g cm⁻³ and C_Ra = 50 Bq kg⁻¹ at 20 °C). This calculation reveals the strong influence of soil texture on radon activity and how heterogeneities in the soil profile may determine the local radon activity, without considering other crucial aspects such as radon transport and soil hydraulic properties. Furthermore, the influence of temperature on radon activity is negligible compared to soil texture, considering that daily soil temperature variations at a 100 cm depth are less than 3 °C (Figure 2, Figure 3, Figure 4 and Figure 5). For example, a change from 20 to 30 °C reduces radon activity from 15.49 to 16.87 kBq m⁻³ (for C_Ra = 50 Bq kg⁻¹ and a loam soil defined in

Table 6. Percentage of sand, silt, and clay (% SSC), and its related n and pk_g values according to the USDA soil texture classification.

Parameter	Clay	Silt	Loam	Sandy
Sand (%)	5	5	42	90
Silt (%)	5	90	41	5
Clay (%)	90	5	17	5
n	1.20	1.64	1.44	2.30
pkg	12.5	12.3	12.8	11.4

).

However, gas permeability in unsaturated soils is also included in RP and varies significantly with %SSC and SWC. Increasing the water content in soils leads to a decrease in gas permeability, which influences RP, because the presence of water in porous materials decreases both the size of the pores and the effective porosity, and at the same time increases tortuosity [20]. The estimation of the unsaturated gas permeability, k_g (S_w), can be formulated as follows (Equation (6)):

$$k_g\left(S_W\right)=k_g\left(1-S_W^{\tau }\right)·\left[1-{\left(1-{\left({1-S}_W^{m^{-1}}\right)}^m\right)}^2\right],$$

(6)

where k_g is the variable-saturated gas permeability, m results from the parameter n of the van Genuchten water retention curve as m = 1 − n⁻¹, and τ is the tortuosity value (commonly τ = 0.5) [83]. Applying the approach proposed by [14] for RP studies, k_g and n are obtained from the Rosetta PTF using % SSC, together with the bulk density of the soil as input variables (Table 5).

Figure 9 analyses the impact of water content and soil texture on both radon activity and gas permeability (Figure 9a–c), as well as the radon potential (Figure 9d–f). This comparative analysis contemplates four different soils (clay, silt, loam, and sandy; Table 6) and three different ²²⁶Ra activities, representing soils with low to high radon activity [12]. The calculations considered a bulk density of 1.7 g cm⁻³—a value corresponding to a loam soil—and a porosity of 0.5—an average value of the porosity of a soil—at 20 °C. Due to the emanation coefficient, radon activity strongly increases from S_w = 0 to 0.1 (SWC = 0.05 for ϕ = 0.5) and remains nearly constant as water saturation increases. Relatively small fluctuations in water content at low values will lead to a high variability in radon concentration, as the long-term radon series also showed (Figure 2 and Figure 3). As previously explained in Section 3.3, this variability is amplified with particle size (i.e., sandy soils). However, the changes in gas permeability are remarkable in clay soils, becoming more impermeable as water saturation increases (resulting in higher pk_g values; Figure 9a–c).

Water saturation influences RP evolution by changing radon activity and permeability (Figure 9d–f). As with radon activity, RP reaches its maximum at S_w = 0.1 due to radon emanation. Therefore, small changes in SW (before reaching S_w = 0.1) can result in RP rising from low to medium–high—e.g., hot desert climate—whereas RP decreases as water saturation increases due to the reduction in gas permeability. On the contrary, a water-saturated soil will present a low RP risk index—e.g., temperate oceanic climate.

This analysis emphasises the influence of weather parameters and soil characteristics on radon potential. Here, we have primarily evaluated soil water content, which is related to precipitation. However, barometric pressure, temperature, and wind can also affect radon potential and may make an accurate estimation difficult. This analysis particularly highlights how local soil texture heterogeneity can lead to significant variations in radon potential and risk index classification. Therefore, a comprehensive analysis of pedological properties such as soil structure, particle size distribution, water content distribution in the soil profile, clay patches, organic matter, pH, etc., should be performed to better understand the dynamics of radon gas in soils and, consequently, radon potential classification.

5. Conclusions

Long-term monitoring of loamy sand and silty clay loam soils highlights the importance of soil texture and weather parameters on radon dynamics. Rainfall plays a key role due to its direct impact on soil water content in the context of a warm semi-arid climate.

Statistical analysis of the time series reveals that radon activity undergoes strong seasonal variations related to rain and soil water content. For both soils, at the study depth of 90–100 cm, radon activity is above the average level during rainy periods, mainly in spring, whereas it is below these levels during dry periods, particularly in summer (4215 ± 960 Bq m⁻³ and 1363 ± 814 Bq m⁻³, in granite and calcareous soils, respectively). An increase in soil water content (SWC) enhances radon emanation. However, if this increase in SWC is high enough, the topsoil layer becomes saturated and impedes radon movement, causing a significant increase in radon activity.

For SWC to be significantly affected at a depth of 90–100 cm, there must be heavy and/or prolonged rainfall. Given these conditions, a measurable increase in SWC will occur much faster in the loamy sand soil compared to the silty clay loam soil. Radon increments due to SWC variation can be measurable within hours or days, depending on soil texture. Our artificial water experiments, which forced SWC to be higher than those usual in the studied region, demonstrate that humid climates with greater rainfall would result in soils with higher radon activities (for a given radium concentration).

Multivariate analysis reports how weather parameters determine the radon dynamics. SWC is likely to exert the most significant influence on radon in soil. However, its response to rainfall events and variability within the soil depth depends on the soil type. Barometric pressure, temperature, and wind also affect radon activity in soil, although their impact is less significant in the studied soils. Loamy sand soil shows statistically more robust results than silty clay loam soil. Thus, radon is positively related to SWC and negatively related to temperature, evapotranspiration, and pressure for both soils, while wind speed is only negatively associated with the loamy sand soil. Wavelet analysis confirms the interrelation in the time–frequency domain of SWC and radon in loamy sand soil. In this soil, wavelet analysis also highlights the effect of variations in pressure and wind speed on radon activity. However, wavelet results for the silty clay loam are less decisive, probably due to the absence of rainfalls in the monitoring period, and also due to the inherent properties of the soil.

Finally, the analysis of the radon potential (RP) for the studied soils under a warm semi-arid climate is classified as low risk due to low ²²⁶Ra concentration. The response of radon activity to changes in the environment has been parametrised to model RP variations under different soils and climate conditions. Thus, theoretical RP analysis confirms the significant radon fluctuations recorded in the long-term radon series. Small fluctuations in water content at low soil VWC have a large impact on radon concentration, and afterwards, the activity remains stable. Water saturation influences RP evolution by changing radon activity and permeability applied to local soil texture heterogeneity and ²²⁶Ra concentration. Continuous monitoring is proven to be crucial for understanding seasonal radon changes, especially regarding the estimation of RP. Considering climate and weather variations and soil textures is essential not only for assessing RP and developing strategies for new buildings and mitigating possible razon risk hazards in existing constructions but also for creating more reliable large-scale radon risk maps.