Methodologies for risk mapping of radon gas at various scales in Tondela and Oliveira do Hospital region (center of Portugal)

Inhalation of radon gas exposes the lungs to ionizing radiation which significantly contributes to the equivalent dose received by a human body. European Union advises Member States to identify radon prone areas (RPA), characterized by a significant percentage of dwellings above the national reference level (RL).

The presented work aims to evaluate the use of Receiver Operating Characteristic (ROC) curve analysis to map RPAs at a small-scale (from 1:25 000 to 1:1 000, henceforward called “regional” and “local” scale respectably), using interpolated surfaces of total gamma radiation (TGR) as proxy and point data of radon concentration in dwellings as the observed variable.

The case-study areas are in the center of Portugal (Tondela and Oliveira do Hospital) where outcrops different coarse-grained biotite granites (Beira’s Granite) and metasediments of Beiras’ Group, frequently as small enclaves hosted in the granites. An intense network of faults is also characteristic of these regions.

At Tondela area the geospatial analysis and ordinary kriging interpolation of TGR, on a regional scale, evidenced: a) a geological control on this variable; b) a structural control on anomalies by N35ºW orientated faults and by the intersection of these structures with others, namely N75ºE and N55ºE; c) and an anisotropic covariance of equally spaced points with N35^oE oriented major axis. At Oliveira do Hospital, where at a regional scale just data of anomalies was available, the log-normal distribution of background values was simulated based on high-definition data obtained at a local scale. The results are consistent with the structural control pattern identified at Tondela. The best classifiers identified by the ROC analysis were 175 cps and 450 cps, respectably for Tondela and Oliveira do Hospital regions.

Establishing a 10% probability of dwellings with concentrations of radon above RL ( ) to define an RPA, all the areas were classified as RPAs. At Tondela region, the lowest risk area represents 25% probability of exceeding the RL and the highest risk area 52%. At Oliveira do Hospital almost the entire region represents 56% exceedance probability. The highest risk area is spatially related to intense anomalies and represent 78% exceedance probability.

For the geological context studied, the use of TGR proved to be suitable for radon gas risk mapping. The ROC curve analysis enabled to significantly classify higher and lower risk areas within high-risk regions, considering the small-scale variability. The ROC analysis did not produce a classifier properly calibrated to the RL but one that improves the cost-benefit of the classification relatively to the natural prevalence of the studied areas.