In-situ measurements of radon concentrations in soil gas at a site on the Cape Flats
Manavhela, Ramudzuli Fijiant
MetadataShow full item record
Measurements of radon (222Rn) concentration in soil gas are routinely used to locate geological fault zones. This study was undertaken to investigate the systematic effects that influence radon soil-gas measurements, in preparation for the first such fault zone measurements in South Africa. The measurements were made at two zones (called A and B for ease of reference) on the iThemba LABS site, which is located in Faure, on the Cape Flats, Western Cape (South Africa). There are no known faults located in the vicinity (~ 10 km) of the site. The radon measurements were made using a RAD7 (Durridge) radon monitor, which makes use of alpha spectrometry. A steel soil probe, which is inserted into the soil, was used to transfer radon from a certain depth in the soil to the RAD7 monitor via a flexible tube. Measurements were made at five sampling areas (three in zone A (A1, A2 and A3) and two in zone B (B1 and B2)), on the site during the months of May, August and October 2006. The variation of soilgas radon concentration as a function of depth, time of day and meteorological data were studied. The depths at which measurements were made were generally 25, 50, 75 and 100 cm. (In August and October, some of the depths were not considered due to a high water table.) Atmospheric temperature, air humidity, wind speed and atmospheric pressure were measured using a portable weather tracker (Kestrel 4000). The highest radon concentrations recorded were 11000 ± 776 Bq.m-3 and 26900 ±1370 Bq.m-3 at a depth of 100 cm in zone A and B, respectively (May). Generally, the radon concentrations were significantly quenched during the August and October measurements relative to the May values. The May soil-gas radon concentration profiles as a function of depth were fitted using a chi-square minimization procedure in order to extract diffusion lengths. A functional form derived from a onedimensional diffusion model was used for the fitting. One the parameter of the fit function is the 226Ra activity concentration. This was determined by means of gammaray spectrometry (in-situ and laboratory-based). A MEDUSA-type detector system comprising a CsI(Na) scintillator crystal and a GPS signal receiver were used for the in-situ measurements. A high-purity germanium detector (2.6 keV FWHM resolution at 1.33 MeV, 45 % relative efficiency) was used for the laboratory measurements. Significant seasonal variations in soil-gas radon concentration were observed (concentrations were generally much lower during or after the rainy season due to the presence of a high water table, and possibly the presence of more lush vegetation causing more radon to escape to the atmosphere). Significant (up to a factor of 2) local variation in radon concentration was observed at a particular depth, in a particular zone during autumn (just after the dry season). The soil-gas radon concentrations (for depth > 25 cm) in zone B were ~ factor of 2 higher than in zone A. The difference is largely due to the fact that the 226Ra soil activity concentrations in zone B were higher than in zone A (by a factor of ~ 1.6). Significant correlations (with coefficients > 0.6) were found between soil-gas radon concentrations and soil 226Ra activity concentrations (using a combination of zone A and B data). No significant, robust correlations (with correlation coefficients > 0.6) were found between soil-gas radon concentrations and meteorological parameters. Often contradictory results (in terms of the sign of the coefficient) were found. From measurements made at two depths (25 and 50 cm) in zone A, over a 24 hour period, there is evidence of a statistically significant variation between night and day radon concentrations, with the former being higher. The one-dimensional diffusion model used to fit the measured radon concentration depth profiles failed to fit the data when the results for a depth of 25 cm were included. When the 25 cm data were excluded the fits yielded diffusion lengths that are more realistic, although significant variations in diffusion length were found in a particular zone.