Statistical evidence of the geological control over radon soil gas concentrations and its implications for mapping radon potential

Determining how radon varies spatially over a given area in the natural environment is important for defining high risk areas and has implications for building practice. To achieve the former, radon concentrations in three areas of the English Midlands were surveyed. The first area comprised a single rock formation near Hereford (described elsewhere). The others, reported here, were areas where elevated concentrations of radon were expected. One was an area comprising two major rock types near Burton in Derbyshire where a multistage sampling design was used to determine the approximate spatial scale of variation in soil radon concentration. The third was an area of more complex geology near Nottingham, where sampling along a transect enabled the structure and scale of variation to be determined. In all of the areas radon concentrations varied considerably, both over large and small distances. The data were analysed using methods embodied in geostatistics. The results showed that structure in the spatial variation of radon for the Burton and Nottingham surveys at the longer scale could be attributed to the effect of lithology. The latter appears to account for approximately 50% of the total variation in both surveys. These results have important implications for mapping radon and also for building programmes, insurance, etc. They also suggest that to estimate radon reliably at the local level by interpolation would generally require very intensive sampling, i.e. at a scale of metres rather than kilometres. However, stratification of an area based on geology, with sampling within the strata designed to estimate average radon concentrations optimally, would provide reasonable estimates in certain situations for somewhat less sampling effort. This could provide a quick and efficient means of assessing the general radon potential of a given area. The best estimate of radon concentration at an unsampled site would then be the mean for that particular rock formation. However, if exact values are needed then radon should be measured in situ.