A modelling methodology for the analysis of radon potential based on environmental geology and geographically weighted regression

• Radon estimate based on local regression technique (GWR) more tuned to local factors.
• Spatial non variability of local relationships captured by novel GWR based approach.
• Geological factors influence on indoor radon values studied as spatial effects.
• Multilevel tool used to pre-process radon data in order to filter building variability.

Many countries have promoted environmental studies and established national radon programmes in order to identify those geographical areas where high indoor exposure risk of people to this radioactive gas are more likely to be found (often referred to as ‘radon-prone areas’). Traditionally, the evaluation of radon potential has been pursued by means of global inference techniques. Conversely, in this paper we present a novel modelling approach, based on well established environmental software, best suited to capture the spatial variability of local relationships between indoor radon measurements and some environmental geology-related factors. The proposed strategy consists of three stages. First, a multilevel model based standardisation of indoor radon data should be carried out in order to reduce the building related variability. Then, the global and local autocorrelation indexes have to be employed to highlight the role of the local effects. The last step implies the use of the Geographically Weighted Regression(GWR) to show the differences in associations between indoor radon and the geological factors across space. The method was tested using an available geo-referenced dataset including both radon indoor measurements and geological data related to the territory of an Italian region (Abruzzo). The results are encouraging, although there are several critical issues to be addressed.