Investigating the Spatial Variation of Radon in Soil Geostatistically
25 July 1999
Oliver, M. A. Kharyat, A.L.
Key Words: geostatistics, kriging, nested survey, radon, spatial variation, variogram
Increasing concern about possible links between emissions of radon and certain types of malignant disease has led to local and regional surveys to measure radon concentrations in the soil and in dwellings. We selected an area in the Midlands of England (Derbyshire), designated as a radon affected area, to explore the nature of the spatial variation of the gas using solid state nuclear track detection. The aim of the first of three surveys was to discover the approximate scale of resolution in the variation using an unbalanced nested sampling scheme and a hierarchical analysis of variance. An area of 7 km by 7 km with two different limestone formations was sampled by a range of sampling intervals: 1 m, 4 m, 15 m, 60 m, 240 m, 950 m and 3,750 m. The components of variance from the analysis of variance were accumulated and plotted against distance to give a reconnaissance variogram. The results suggested that geology exerts a strong control on the variation, but that there are other factors involved giving rise to spatially correlated variation within the lithological units over distances of 60 m to 240 m.
To gain further insight into the variation of radon, we sampled an area of 2 km by 2 km on the Monsal Dale limestone only. The variogram was computed and modelled with a power function. It was then used with the data to estimate radon concentrations at unsampled locations by kriging (geostatistical estimation). The map showed a distinctive pattern that suggested an inverse relation between elevation and radon concentration. The large radon values in the western part were associated with low elevation, and the small values occurred where the relief is higher. This suggested that factors other than lithology might account for the variation in radon concentration, for instance soil thickness, bulk density, particle size distribution, and so on.
The third survey was a 2 km long transect along which the soil radon concentration was measured together with soil thickness, conductivity, and slope angle every 20 m. The radon concentrations along the transect varied considerably and the variogram was periodic with a wavelength of about 263 m. Ground conductivity and slope angle also were periodic with wavelengths of about 180 m. The periodic variation of radon and ground conductivity might result from variation in soil thickness associated with joint controlled solution in the underlying limestone. Resistivity measurements suggested that where the radon values were large in the West the soil is thick, and where they were small in the East the soil is shallow. Particle size distribution of the soil also showed some relation with radon: where the soil contains more sand and silt, the radon values are larger than where it contains more clay.
IV International Conference on GeoComputation, Mary Washington College, Fredericksburg, VA, USA, 25-28 July 1999.