Comparison of AIRDOS-EPA predictions of ground-level airborne radionuclide concentrations to measured values. Final report

The precision of the AIRDOS-EPA code in predicting ground-level radionuclide concentrations at varying distances from the source is high, after adjustment for incorrectly specified source characteristics or possible measurement bias at each site. The precision of the model is best measured by the constrained regression of adjusted predicted values versus measured values for all sites. The coefficient of determination (R2) of this regression is 0.77 with a slope of 0.92, indicating that the predicted concentrations are statistically equivalent to the measured values after adjustment of site-specific bias. At the 50% confidence level, the predicted concentrations at all sites are within 50% of the measured values after adjustment