Characterization of naturally occurring radioactive material dynamics in community water systems using groundwater from Ganghwa Island, Republic of Korea

Abstract

Understanding the temporal behaviors of naturally occurring radioactive materials is important for the maintenance of community water supply systems for drinking groundwater on islands and in mountainous areas. Intensive field sampling in the Republic of Korea to characterize temporal variation in 238U concentration and 222Rn activity was performed for three community water systems on Ganghwa Island for ten years (2012–2022). 222Rn, a dissolved gas in groundwater, tended to decrease in activity as water moved from the source through community water systems, but no such trend was observed in 238U concentration. 238U and 222Rn behaviors in time series data for different CWSs reflected different geochemical processes and were differently classified accordingly. In particular, the 238U and 222Rn at points GJ and YH shows the signature of weathering: accompaniment by variation in pH and dissolved oxygen. In other time series, the 238U and 222Rn behavior shows the signature of cation exchange by seawater intrusion: accompaniment by dramatic variation in electrical conductivity. Random forest regression models developed to estimate the 238U concentration and 222Rn activity in groundwater boreholes exhibited reasonable threshold classification accuracy. The optimized threshold structure for 238U concentration and 222Rn activity variation with respect to the water quality variable of greatest statistical importance in each community water system random forest regression model corresponded to the geochemical process controlling the behavior of 238U and 222Rn in that groundwater borehole. Therefore, more intensive field sampling, covering the full data range that could emerge at each groundwater borehole, can aid in enhancing the accuracy of estimation of 238U concentration and 222Rn activity. Furthermore, the use of an in situ groundwater monitoring system in a random forest regression model facilitates the effective management of community water systems in the face of risk posed by naturally occurring radioactive materials.