Application of airborne geophysical survey data in a logistic regression model to improve the predictive power of geogenic radon maps. A case study in Castleisland, County Kerry, Ireland.

In this study, a novel methodology was investigated to improve the spatial resolution and predictive power of geogenic radon maps. The data inputs comprise indoor radon measurements and seven geogenic factors including geological data (i.e. bedrock and Quaternary geology, aquifer type and soil permeability) and airborne geophysical parameters (i.e. magnetic field strength, gamma-ray radiation and electromagnetic resistivity). The methodology was tested in Castleisland southwest Ireland, a radon-prone area identified based on the results of previous indoor radon surveys. The developed model was capable of justifying almost 75 % of the variation in geogenic radon potential. It was found that the attributes with the greatest statistical significance were equivalent uranium content (EqU) and soil permeability. A new radon potential map was produced at a higher spatial resolution compared with the original map, which did not include geophysical parameter data. In the final step, the activity of radon in soil gas was measured at 87 sites, and the correlation between the observed soil gas radon and geophysical properties was evaluated. The results indicate that any model using only geophysical data cannot accurately predict soil radon activity and that geological information should be integrated to achieve a successful prediction model. Furthermore, we found that EqU is a better indicator for predicting indoor radon potential than the measured soil radon concentrations.

Machine learning in environmental radon science.

Temporal dynamic as well as spatial variability of environmental radon are controlled by factors such as meteorology, lithology, soil properties, hydrogeology, tectonics, and seismicity. In addition, indoor radon concentration is subject to anthropogenic factors, such as physical characteristics of a building and usage pattern. New tools for spatial and time series analysis and prediction belong to what is commonly called machine learning (ML). The ML algorithms presented here build models based on sample and predictor data to extract information and to make predictions. We give a short overview on ML methods and discuss their respective merits, their application, and ways of validating results. We show examples of 1) geogenic radon mapping in Germany involving a number of predictors, and of 2) time series analysis of a long-term experiment being carried out in Chiba, Japan, involving indoor radon concentrations and meteorological predictors. Finally, we identified the main weakness of the techniques, and we suggest actions to overcome their limitations.

Applicability of radon emanometry in lithologically discontinuous sites contaminated by organic chemicals.

The applicability of radon (222Rn) measurements to delineate non-aqueous phase liquids (NAPL) contamination in subsoil is discussed at a site with lithological discontinuities through a blind test. Three alpha spectroscopy monitors were used to measure radon in soil air in a 25,000-m2 area, following a regular sampling design with a 20-m2 grid. Repeatability and reproducibility of the results were assessed by means of duplicate measurements in six sampling positions. Furthermore, three points not affected by oil spills were sampled to estimate radon background concentration in soil air. Data histograms, Q-Q plots, variograms, and cluster analysis allowed to recognize two data populations, associated with the possible path of a fault and a lithological discontinuity. Even though the concentration of radon in soil air was dominated by this discontinuity, the characterization of the background emanation in each lithological unit allowed to distinguish areas potentially affected by NAPL, thus justifying the application of radon emanometry as a screening technique for the delineation of NAPL plumes in sites with lithological discontinuities.