Analysis of the relationship binding in situ gamma count rates and soil sample activities: Implication on radionuclide inventory and uncertainty estimates due to spatial variability

Abstract The paper strives to identify through geostatistical simulations the parameters which build up a correlation between radionuclide activity concentrations measured on core samples and corresponding in situ total gamma count rates measured into boreholes drilled within the contaminated soil. This numerical exercise demonstrates that a linear relationship should exist between logarithmic values of in situ count rates and logarithmic values of activity concentrations when the contamination is strongly structured through space. A sensitivity analysis to some parameters (geostatistical range of the contamination structure, core sampling method, soil water content, multiple gamma-emitter contamination, etc.) is undertaken to identify which situations may impede the use of such a correlation. Then this approach is applied on Chernobyl measurements undertaken in 2015 and compared to the co-kriging method which considers the localization of the measurements and the additional measurements. It appears that co-kriging is a better estimator than linear regression, but the latter remains an acceptable way of estimating activity from gamma emitters and presents better results than lognormal regression. Therefore, total gamma logging measurements performed into boreholes of porous media contaminated by gamma-emitting radionuclides can be used for characterizing contamination and dealing with its spatial variability with the use of co-kriging.