A geostatistical approach for mapping the geogenic soil gas radon potential in a south Italy area

Radon is a natural radioactive gas emanated constantly in small amounts from the Earth to theatmosphere. Many scientists use to measure outdoor soil gas radon concentrations to assess thegeogenic radon potential because it is the main source for indoor radon concentrations independenton the construction features of building. Spatial distribution of soil gas radon concentrations hasbecome an important issue in terms of radiological protection because constitute a serious humanhealth hazard. Geostatistical methods provide us a valuable tool to study spatial structure of radonconcentration and mapping its spatial distribution. Stochastic simulation is a development ofgeostatistics and estimates the conditional cumulative distribution functions at each location.Statistical information deriving from stochastic simulation allows to estimate the probability thateach pixel exceeds a threshold value and to produce the probability map of high radonconcentrations in the soil gas. The main aim of this paper was to explore the spatial structure of soilgas radon concentration in a south Italy area and mapping the geogenic soil gas radon potential.Another aim was mapping the risk of occurrence of high soil radon gas concentration. Theexperimental area was located in the Catanzaro-Lamezia plain (south Italy) with a surface of about1105 km2. It is a graben bordered by E-W trending normal faults and constitutes the central part ofthe Calabro-Sicula rift-zone. Measurement of radon concentration were made at 4420 pointscollecting soil gas radon into Lucas cells and then measuring their alpha activity in the laboratory.Measurements were made in July 2004 and the samples were collected as uniformly as possible withan average sampling density of 4 samples per km2. To reduce the influence of few high values of soilgas radon a Multi-Gaussian approach was used. An isotropic model was fitted to the experimentalvariogram including three basic structures: 1) a nugget effect; 2) a spherical model with a shortrange=3.78 km and 3) a spherical model with a long range=23.90 km. Stochastic simulation was usedto map the risk of occurrence of high soil radon gas concentration: 500 alternative equi-probableimages of the unknown radon concentration were generated using the conditional sequentialGaussian simulation algorithm. Counting the number of times that each pixel exceeded the thresholdvalue and converting the sum to a proportion we produced the probability map of exceedence. Weused as threshold value the upper quartile of the data distribution function because in Italy no radonlevel of risk in outdoor air exists. Map of soil gas radon concentration and probability map reflectedthe impact of fractures and faults on the spatial distribution of radon concentration because they actas a preferential way for gases migration. The results showed that the highest radon values occurpreferentially along elongated zones similar to the most representative trends obtained bygeomorphological and mesostructural analyses, i.e. E-W trends and, secondarily, NW-SE orientations.