A data set of long-term radon measurements in approximately 2200 houses in
southern Belgium has been collected in an on-going national radon survey. T
he spatial variation of indoor Rn concentrations is modelled by variograms.
A radon distribution map is produced using the log-normal kriging techniqu
e. A GIS is used to digitise, process and integrate a variety of data, incl
uding geological maps, Rn concentrations associated with house locations an
d an administrative map, etc. It also allows evaluation of the relationship
s between various spatial data sets with the goal of producing radon risk m
aps. Based on geostatistical mapping and spatial analysis, we define three
categories of risk areas: high risk, medium risk and low risk area. The cor
relation between radon concentrations and geological features is proved in
this study. High and medium Rn risk zones are dominantly situated in bedroc
k from the Cambrian to Lower Devonian, although a few medium risk zones are
within the Jurassic. It is evident that high-risk zones are related to a s
trongly folded and fractured context. (C) 2001 Elsevier Science B.V. All ri
ghts reserved.