Mapping radon-prone areas by lognormal modeling of house radon data.

Results of radon measurements in houses are generally distributed approximately lognormally, whether large or small areas are considered. The properties of the lognormal distribution allow the proportion of houses above a threshold level to be estimated even when there are insufficient data to calculate the proportion directly. In this paper the properties of the data set of radon measurements made in UK houses with etched-track radon detectors are examined, with the purpose of using them to map radon-prone areas. Individual results are normalized depending on the average outdoor temperature during the measurement to estimate the annual average radon level. For certain types of maps, results may also be normalized to estimate the radon level in a standard house or mixture of house types on the same site. Methods to use the lognormal model to estimate the geometric mean radon concentration and geometric standard deviation for data grouped by area have been developed. These data are then used to map radon-prone areas of England and Wales.