Release of 222 Rn from some soils

Measurements have been made of 222Rn release from diverse soils in the region surrounding Malaga, Spain. These flux measurements were carried out by two methods. A direct method using a static chamber technique and another indirect method obtained from concentration profile measurements of 222Rn in the soil air. The effects of meteorological variables and other parameters on 222Rn flux were studied. The factors that most affected the instantaneous value of 222Rn release were humidity and soil thermal gradient. The directly measured 222Rn fluxes at investigated sites are higher than 222Rn fluxes derived by the indirect method.


Introduction
Rn is an inert radioactive element with a half-life of 3.8 days. It belongs to the radioactive uranium series and occurs in soil gas in varying concentrations. In recent years, Rn has been used as a tracer for the origin and trajectory of air masses (Prospero and Carlson, 1970;Karol, 1974;Wilkniss et al., 1974;Larson and Bressan, 1980). Rn has also been used in more quantitative studies particularly to determine vertical matter diffusion coefficients (Druilhet and Fontan, 1972;Hsu et al., 1980). Another interesting use of Rn is as an indicator of the vertical stability of the lower atmosphere at a given site through a study of the equivalent mixing height, a parameter which is a characteristic of vertical stability (Guedalia et al., 1980). In problems relating to conservation of mass of atmospheric Rn an accurate knowledge of Rn release is essential.
The present research was undertaken with three principal goals: (1) to characterize the flux of Rn from the soils of Ma´laga; (2) to study which factors are responsible Correspondence to: C. Duen as for the variation in flux from one site to the next and (3) to compare the flux of Rn by two different methods.
However, the Rn release value is in itself an important parameter, as it is directly related to the presence or absence of seismic activity (King, 1978;Duen as and Fer-na´ndez, 1988) and also to the effects of underground nuclear explosions. However, the numerical value of Rn release for a given terrain is not the only important factor. It is equally necessary to know the variations of the release that may be caused by changes in meterological conditions. The characteristics of the soil itself are important, including the amount of parent Ra present, the porosity and permeability of soil and the degree of water saturation. There is a great confusion and a lot of argument about the results obtained, especially regarding the influence of certain meteorological parameters.
In the present work, the release of Rn was measured by two methods, including closed accumulation and an indirect method based on an understanding of the vertical profile of the Rn present in the soil air in the first few meters below the ground surface. The Rn release has been measured in four types of soil in the area around Ma´laga city (Spain) over various times. Quantitative relations between the Rn release and several meteorological variables and characteristics of the soil were found. Rn transport in the unsaturated soil zone is by molecular diffusion. We have determined the diffusion coefficient and the emanation fraction for all the soils studied. Finally, the Rn release measured by accumulation method is compared for each soil with the exhalation obtained indirectly from concentration profile measurements in the soil air.

Theoretical considerations
Gas transport in the unsaturated soil zone occurs through molecular diffusion; then, the Rn flux can be calculated from Fick's first law: where Jo 0L is the flux density, D 0L is the bulk diffusion coefficient for Rn through the volume of the porous medium and C 0L is the concentration of Rn in the interstitial gas.
Fick's second law follows from Eq. (1) by a conservation principle with the added terms due to Rn decay and emanation from the solids of the medium, assuming a homogeneous Rn source strength and a depth independent diffusion coefficient D 0L : where is the porosity of the medium defined as the ratio of void volume to total volume, 0L is the Rn decay constant and 0L is the emanation power of the medium into the interstitial volume. 0L (Bq m\ s\) is estimated from soil parameters and the decay Rn constant (Nazaroff et al., 1988) according to: where f is the emanation fraction, defined as the quotient between the Rn atoms that enters the pore soil and the Rn atoms generated in soil, B is the density of soil grain and C 0?\ (Bq kg\) is the radium activity concentration in the soil.
Equation (2) may be solved in the steady state using the boundary conditions C 0L P0 at zP0 and C 0L "C 0L at zPR: where: is a denominated relaxation depth and C 0L is the radioactive equilibrium concentration value at great depth.
From Eqs. (4), (5), (2) and (3), the equation of emanation fraction of Rn is obtained: This equation permits us to obtain the emanation fraction of Rn from soil as a function of physical characteristics of soil, such as porosity, density and radium content and C 0L . The flux at the soil surface, J 0L , as a function of relaxation depth is therefore: All the equations are derived under the assumption that the gas transport in the soil water phase can be neglected. This assumption is true for Rn because the molecular diffusion coefficient of Rn is 10 times smaller in water than in air. In Eq. (1) we have neglected the where K is the permeability, is the dynamic viscosity and p is the absolute pressure. In the study soils, the K values have been small. The soil properties will be exposed after this paragraph.

The chamber
The closed accumulation methods were of the type described by Wilkening et al. (1972). The chamber used consists of a cylindrical container 0.55 m in diameter and 0.32 m in height. The chamber was made of stainless steel with a sampling tube. A small electric fan inside the chamber was used to maintain uniform mixing of emitted gases. The fan was operated a few minutes before the accumulation period was finished. Equal pressure between the environment and the interior of the chamber was maintained through an orifice with 3 mm diameter. The chamber was placed at a depth of 1-2 cm in the soil.

Radium content
The radium content of the soil is typically given as an activity concentration per unit mass. Soil cores were obtained in the first centimeter of the soil. Samples were dried in the laboratory, sieved to remove rock fragments and sealed in Marinelli type container of 450 cm for several weeks. Samples were counted directly on an intrinsic germanium detector for 24 h to obtain sufficient precision of measurement. The intrinsic germanium detector has an efficiency of 25%, a resolution of 2 KeV and is surrounded by shielding material to reduce the background count. The 610 KeV gamma of Bi was used as a measurement of Ra activity after several weeks had elapsed between sealing and gamma counting to permit Bi to grow into equilibrium with Ra. Careful calibration was carried out using soil samples of similar density with Eu uniformly distributed and sealed in the Marinelli recipient analogues used for the soil cores.

Radon in air
For the Rn analysis, air was sampled into a 1 l or 0.5 l lucite walled cylindrical cell, previously evacuated, as a modified design of the original Lucas cell (Lucas, 1957) developed by Quindo´s et al. (1991). The walls of the cell were coated internally with SZn(Ag) coated mylar, but the main difference from the traditional one is the ability to open the cell after use by removing the bottom. The background count of this kind of cell ranged from 0.7 to 1.3 c.p.m., with counting times of 30 min, resulting in a lower limit of detection around 10 Bq/m. Rn was measured by counting the alpha particles emitted by Rn and its daughter products, Po and Bi, when they reach radioactive equilibrium. The precision of the measurements was about 5% taking only the statistical error into account (Carretero, 1994). Typically, counting intervals of 30 min to 2 h were employed. The cell developed by Quindo´s et al. (1991), was calibrated in the Lovelace Inhalation Toxicology Research Institute, Albuquerque, New Mexico, USA. Afterwards the cell was recalibrated in our laboratory obtaining an average efficiency of 0.510$0.015.

Interstitial Rn air
Measurements of Rn in soil air concentrations were made by inserting some stainless steel sampling tubes into the soil at several depths, according to the soil type. The 10 mm diameter tubes ended in a 3.5 cm diameter, 6 cm long filtration chamber which was designed to permit the aspiration of interstitial air from the soil. The chamber was filled with glass fibers to prevent the aspiration of soil or clay particles. The upper ends of the tubes were sealed except for the time that samples were being taken. A small purging took place before and after sampling was completed, by direct (filtered for dust) intake into evacuated scintillation cells for subsequent Rn analysis.

Location and properties of soils
All the samples were taken in four sites in an area located outside the city of Ma´laga. The climate of Ma´laga is warm, temperate with hot summers and little rain, particularly during measurement periods in the last 2 years. Figure 1 shows the location of study soils and Table 1 exhibits the characteristics of the soils mentioned.
We studied soil properties such as particle-size analyses, density, porosity and permeability. Table 2a shows the grain size distribution and Table 2b the different properties.
Particle size analyses were carried out by the sieve method as in Jime´nez and de Justo (1975). Porosity was determined in the laboratory from real and apparent density measurements of a soil sample obtained by perforation of the soil with a core sampler which is a hollow cylinder whose volume is known. Permeability was determined in situ assuming Darcy's law and measuring the flow for a fixed pressure drop.
The meteorological data (wind speed, air temperature, air humidity and atmospheric pressure) were supplied by the Ma´laga Airport Observatory. Soil temperatures at 15 cm and 80 cm depth were measured with a Crisson digital thermometer with a platinum sounding probe. Soil humidity was measured during all the samplings on the first centimetre of soil.

The Rn release by the direct method
Soil gas flux was measured directly by the accumulation method (Wilkening et al., 1972) using the chamber described in Sect. 3.1. The concentration change was where C 0L is the concentration increase during t and h is the height of the chamber. The average accuracy of Rn flux by direct method has been 7%. Table  3 shows the average flux of Rn accompanied by an indication of the standard deviation found, the activity of Ra and the total number of measurements for each soils.
Altogether, we obtained 235 individual Rn release measurements for soils situated close to Ma´laga city. The range of variations of these values is between 3 · 10\ and 25 · 10\ Bq m\ s\, which is coherent with the global distribution of Rn release except for soil M3. Taking into account the different methods existing in order to determine the Rn release in the interphase soil-atmosphere, Rn release oscillates between 0.5 and 2.5 atoms cm\ s\ (10 · 10\ and 52 · 10\ Bq m\ s\), according to existing abundant literature on the measurement of Rn release (Israe¨l, 1951;Servant, 1964;Pearson and Jones, 1965;Wilkening et al., 1972;Sisigina, 1967;Turekian et al., 1977;Mochizuki and Sekikawa, 1978;Martı´n Salas et al., 1979;Quindo´s, 1980;Lambert et al., 1982;Nero and Nazaroff, 1983;Polian et al., 1989 and Graustein and Turekian, 1990. The sampling time was always at the same time of day (1500 UT). The average Rn release from M3 soil is very low for two reasons: (a) their typical location on the coast (30 m from the sea shore) where the water table is about 1.20 m below the surface in dry periods; for this, the humidity at depth is high, the migration of Rn towards the interphase soil-atmosphere making difficult and (b) the number of samplings when rains fell close to the sampling, being 50% of all measurements.
The Rn release is related to the activity of Ra for each soil. A dependence of Rn release on the activity of Ra was observed which may be expressed by the following equation: JM 0L "1.2 · 10\Ra!2.5 · 10\ which gives a correlation coefficient of 0.93 and a degree of confidence higher than 90%.
According to Sisigina (1967) and Do¨rr and Mu¨nnich (1990) the Rn release from soil may be influenced by the radium content and the grain size diameter of soil particles. This is confirmed by the observation of lower Rn fluxes (3 · 10\!10 · 10\ Bq m\ s\) on sandy soils (M3 and M4) and higher Rn fluxes (12 · 10\-25 · 10\ Bq m\ s\) on loamy and clayey soils (M1 and M2).
In relation with the temporal variation of the Rn release for each soil, the Rn release from the clayey soil shows higher variations than the sandy soils (see Table 4). Consequently, the Rn release from clayey soils should present seasonal variations more significantly than sandy soils. This behavior is corroborated in this because the Rn release from clayey soils exhibits better correlation with the characteristic paraeters of soil and meteorological parameters than the Rn release from sandy soils. This behavior is shown in the following paragraph.

Factors affecting values of Rn release
In this study, we have excluded the Rn release from soil M3 because the number of measurements were scarce.  Figure 2 shows the correlations between the Rn release and soil humidity on soils M1, M2 and M4. Table 4 exhibits the coefficients of equation of regression (m and b), the correlation coefficient (r) and the degree of confidence (p) for the soils mentioned.
Soil humidity appears as the decisive factor in release values. Generally, the presence of water in soil produces a reduction of soil porosity and consequently a diminution in gas diffusion rates, in particular, in the top decimeter of soil. This retards Rn release and increases the concentration of Rn in the top layer near the interphase soil-atmosphere. In soil with a high degree of permeability, the presence of water in soil is less important than for soils with low permeability values. The soils M1 and M2 have, respectively, slight and small permeability, while M4 is moderately permeable. For this reason, the correlation between Rn release and soil humidity is better for M1 and M2 than M4. This behavior is analogous to those reported by other investigators, Cox et al. (1970), Megumi andMamuro (1972), Martı´n Salas et al. (1979) and Quindo´s (1980).

Ra release and soil thermic gradient
We have found a correlation between Rn release and thermal gradient for soil temperature between 15 and 80 cm deep for soils M1 and M2. The results of this correlation are shown in Fig. 3 and Table 5.
The direct correlation between Rn release and the soil thermal gradient has originated because an increasing surface soil temperature produces an enhanced surface release. Moreover, the relation between Rn releases and soil thermal gradient is consequence of soil humidity. When the temperature of the top soil is high, the soil humidity is low and consequently Rn release would be greater than when the temperature of the top soil is low, a situation associated with negative thermal gradients linked generally to high soil humidity. This behavior is analogous to those reported by Martı´n Salas et al. (1979) and Ferna´ndez (1981).

Rn release and atmospheric pressure
The existence of a significant atmospheric pressure effect is not surprising because many experiments have shown this relationship. Moreover, mathematical models predict such an effect (Clements and Wilkening, 1974;Edwards and Bates, 1980). Figure 4 shows Rn release measured at 1500 (UT) in each case as a function of the barometric pressure averaged over 12 h before the sampling was completed for M1 and M2 soils. Table 6 shows the characteristic parameters of adjustments.

Rn release and wind speed
Model calculations for wind to be reported in detail elsewhere indicates that its effect is to increase release. Enhancement of release due to wind has been previously reported (Kovach, 1946;Martı´n Salas et al., 1979;Duen as and Ferna´ndez, 1983). In this study we have found a direct correlation between Rn release and the wind speed averaged over 2 h before sampling only for soil M2. Moreover, this correlation has been obtained by fixing the soil humidity at values greater than or equal to 10%. We suppose that the correlation obtained is coherent becausse the increase of wind speed is linked to turbulent mixing that increases the Rn gradient between soil and air and this enhances Rn release. Figure 5 shows the correlation between Rn release and wind speed (vN ) for soil M2. Table 7 presents the coefficients of equation of regression (m and b), the correlation coefficient (r) and the degree of confidence ( p) for soil M2. We conclude that soil parameters and variable meteorological effects influence Rn release from soils M1  and M2 more than from M4. Rn release from soils M1 and M2 presents oscillations higher than from M4, consequently, the Rn release from soils M1 and M2 would be more affected by the parameters considered than M4 soil.  In particular, for sandy soil Ferna´ndez et al. (1983), Schery et al. (1984) and Do¨rr and Mu¨nnich (1990) have not found a correlation between Rn release and meteorological parameters; Ferna´ndez et al. (1983) observed the influence of soil humidity on Rn release from a sandy soil at a sampling point with a very wet climate; Schery et al. (1984), for a sandy soil with semi-arid climate, found correlation between Rn release and barometric pressure; finally, Do¨rr and Mu¨nnich (1990) did not find a correlation between Rn release and parameters, such as meteorological variations and soil characteristics in a wet climate.
The mean concentration profiles of Rn concentration in the study soils are given in Fig. 6. Rn concentration profiles increases with depth to a constant value, C 0L , which varies according to the soil types. The solid curves indicate the Rn concentration profiles fitted to an exponential function by least squares method, Eq. (4). Table 8 shows the characteristic parameters of the adjustments carried out, C and zN for Rn. The average relaxation depth is in good agreement with the theoretical value of 100 cm for Rn (Nazaroff et al., 1988). Also, the number of profiles achieved (N), the chi squared ( ) and the degree of confidence (p) results are given in Table 8. The results of adjustments are very acceptable because the majority exhibit a degree of confidence greater than or equal to 99%; this behavior confirms the supposed hypothesis that the transport in the unsaturated zone is really due to molecular diffusion.

Diffusion coefficient
We evaluated the average diffusion coefficient in each soil from Eq. (5), using measured porosity ( ) and the calculated value of relaxation depth (zN ). Table 9 shows the porosity and the diffusion coefficient for each soil studied.
Using Table 9, we conclude that the average diffusion coefficient for soils M2 and M4 was 15.6 · 10\ cm/s and 3.25 · 10\ cm/s for soils M1 and M3. Considering that the molecular diffusion coefficient of Rn in the air may be taken as 0.1 cm/s and in the water as 10\ cm/s (Tanner, 1964), we thought that the obtained values for diffusion coefficient should be related to the average humidity of soil during the measurement period. Evaluating the average humidity for each soil, we have not found any important differences between the humidities of the different soils to explain the wide range of values that we found on soils studied for diffusion coefficients.
The soils gas sampling at the greatest depth was different for each soil, when we analyzed the diffusion coefficient for each soil together with the greater depth, their location and soil properties. The M2 site provided a clayey soil, covered by old trees (Ficus and Eucalytus). The soil gas sampling at depths of 10, 15, 30, 45 and 60 cm from installations permanently situated on tree roots, which joined with possible internal retractions produced by their clayey nature, may be the factors that help increase the diffusion of Rn between 60 cm depth and the surface. The M2 site presents the highest value of diffusion coefficient. The M1 site presents a very heterogeneous structure in relation to their composition and grain size with clay on the surface and a mixture of sand and gravel in depth. The soil gas sampling at depths of 18, 47, 75, 150 and 122 cm are from permanent installations. The M1 site exhibits the smallest values of diffusion coefficients. The M3 sandy soil also presents a small diffusion value of coefficient; this value may be attributed to their typical location and the rains that occured closely the sampling time. The soil M4 presents a diffusion coefficient that is with in the range of sandy soils with small humidity soil (Duen as and Ferna´ndez, 1987).

Emanation fraction of Rn from sampling soils
We have calculated from Eq. (6) the emanation fraction or emanation coefficient of Rn using measured radium content, the bulk density, the porosity and the calculated value of C 0L for the four study soils. These results, summarized in Table 10, indicate a range of 2.3 to 10.7%.
The average value obtained for the emanation fraction for the soils sampled was 7%. According to Rama and Moore (1984), the fraction emanation is between 1-10% depending on the effective soil particle surface and thus, on grain size distribution. The M3 soil presents the smallest emanation fraction and precisely this soil has also the smallest Rn release. Several factors have been demonstrated in several studies to have a large impact on the emanation coefficient, such as the moisture content and the grain size distribution (Nazaroff et al., 1988). When estimating the different parameters used to evaluate the emanation fraction of Rn, Eq. (6), the main differences are of the values of C and the soils that exhibit the greatest (M4) and smallest value (M3) of the radioactive equilibrium concentration are precisely those which have the highest and lowest emanation fractions respectively.

Comparison between Rn obtained fluxes for direct and indirect method
The Rn fluxes have been evaluated by two methods. A direct method, accumulation, before exposed, and an indirect method from Eq. (7). Table 11 shows the average Rn fluxes by direct and indirect methods for all investigated sites. In all soils sampled, the flux evaluated by direct method is higher than the flux obtained by indirect method. These results are analogous to those reported by other papers (Kraner et al., 1964;Schroeder et al., 1965;Schery et al., 1984).
The greatest difference between direct and indirect flux is exhibited by M3 soil; this difference may be attributed to its locational peculierities already mentioned. The flux evaluated by indirect method has been obtained by supposing that the transport of Rn in the soil air is controlled by molecular diffusion. Thus, the results obtained may indicate that there may be other mechanisms of transport operating in addition to diffusion such as the convective processes or some fissures that augment the flux obtained by direct method. In soils M1 and M2 the quotient between direct and indirect flux is higher than the same quotient for soil M4. In M1 and M2 we have found a good linear correlation between the direct flux of Rn and the thermal gradient of soil between 15 and 80 cm (already exposed), while for M4 we have not found any correlation for mentioned parameters.

Conclusions
The data obtained characterize the flux from the surface of the most widespread generic soil of Ma´laga. The factors that most affect the value of Rn release were humidity and soil thermal gradient. Other factors that also affect the value of Rn release are barometric pressure and wind speed. The directly measured Rn fluxes at investigated sites are higher than Rn fluxes derived by the indirect method.