Abstract
Community groundwater systems serve as the primary source of household water for over 81 million people in the United States. In many residences, groundwater may contain significant concentrations of radon-222, serving as a potential source of human exposure. An extensive database of measured radon concentrations in community groundwater systems is available from the National Inorganic and Radionuclides Survey (NIRS). However, estimation of key descriptive statistics and distribution functions for these data is complicated by the fact that the data are censored, that is, a portion of the NIRS samples have concentrations below the minimum reporting level of 100 pCi/L. A variety of parameter estimation techniques were investigated for fitting the NIRS data, including maximum likelihood and regression methods. In general, the NIRS data were found to be well-described by lognormal distributions. The resulting lognormal distributions, along with quantitative estimates of uncertainty, are presented for five groundwater system sizes and for eight geographic regions.