Hence doubling the 226Ra in the ground should double concentrations in homes, giving a lognormal
distribution which, if plotted as a histogram on a logarithmic x axis, is just shifted to the right without
changing shape. If the measurements of the source term (226Ra) have normally distributed
uncertainties, the appropriate variables to use for modelling would be arithmetic mean source value
and GM radon in homes (Figures 1 and 5 above) . In contrast, the relationship between soil gas and
indoor radon (Figures 9 and 10) is best defined using GMs on both axes because both variables are
lognormally distributed. It therefore seems to be appropriate to use these models for predicting
indoor radon levels which are likely to occur at the 226Ra and soil gas radon concentrations found or
predicted at near-surface radioactive waste disposal sites. These models should be used in
preference to those based on the log (y)-linear (x) relationship between LnGM indoor radon and
18
estimated 226Ra (Figures 2 and 6), which was used in this study for the comparison between WTLS,
LS and Theil regression models.
Hence doubling the 226Ra in the ground should double concentrations in homes, giving a lognormaldistribution which, if plotted as a histogram on a logarithmic x axis, is just shifted to the right withoutchanging shape. If the measurements of the source term (226Ra) have normally distributeduncertainties, the appropriate variables to use for modelling would be arithmetic mean source valueand GM radon in homes (Figures 1 and 5 above) . In contrast, the relationship between soil gas andindoor radon (Figures 9 and 10) is best defined using GMs on both axes because both variables arelognormally distributed. It therefore seems to be appropriate to use these models for predictingindoor radon levels which are likely to occur at the 226Ra and soil gas radon concentrations found orpredicted at near-surface radioactive waste disposal sites. These models should be used inpreference to those based on the log (y)-linear (x) relationship between LnGM indoor radon and 18estimated 226Ra (Figures 2 and 6), which was used in this study for the comparison between WTLS,LS and Theil regression models.
การแปล กรุณารอสักครู่..
