Examination of the uncertainty in air concentration predictions using Hanford field data

The accuracy of an environmental transport model is best determined by comparing model predictions with environmental measurements made under conditions similar to those assumed by the model, a process commonly referred to as model validation. Over the past several years, we have done a variety of validation studies with the popular Gaussian plume atmospheric dispersion model using data from tests conducted on the Hanford site. Data for short-term releases of small particles for a range of release heights from surface level to 111 m have been used. Downwind distances examined have ranged from a few hundred meters to 12.8 km, depending on the particular data used. Measured and predicted ground-level centerline, crosswind-integrated, and 22.5 degrees sector-averaged air concentrations have been compared. Up to six different sets of atmospheric dispersion parameters and three different atmospheric stability class specification schemes have been examined. Overall, dispersion parameters based on measurements made near Jülich, Federal Republic of Germany, give the best comparisons between observed and predicted air concentrations. The commonly used vertical temperature gradient method for determining atmospheric stability class consistently gives poor results. The accuracy of air concentration predictions improves when dry deposition processes are included in the model. Further validation studies using various Hanford data sets are planned.