Increased dose rate precision in combined α and β counting in the μDose system - a probabilistic approach to data analysis

The $\mu$Dose system was developed to allow the measurement of environmental levels of natural radioactive isotopes. The system records $\alpha$ and $\beta$ particles along with four decay pairs arising from subsequent decays of $^{214}$Bi/$^{214}$Po, $^{220}$Rn/$^{216}$Po, $^{212}$Bi/$^{212}$Po and $^{219}$Rn/$^{215}$Po. Under the assumption of secular equilibrium this allows to assess the specific radioactivities of $^{238}$U, $^{235}$U, $^{232}$Th decay chains and $^{40}$K. This assessment provides results with uncertainties which are correlated and, thus, require the development of an error estimation methodology which considers this issue. Here we present two different approaches for uncertainty propagation based on Monte Carlo and Bayesian methods. Both approaches produce statistically indistinguishable results and allow significantly better dose rate precision than when the correlations are not accounted for. In the given example, the dose rate precision is improved by a factor of two.