VARIATIONS IN ENVIRONMENTAL TRITIUM DOSE ESTIMATES DUE TO METEOROLOGICAL DATA AVERAGING AND UNCERTAINTIES IN PATHWAY MODEL PARAMETERS

The large amounts of tritium produced at the Savannah River Site (SRS) coupled with the current dose reconstruction study at the facility em phasize the importance of ensuring accurate and efficient prediction o f tritium doses to the public. Presently, dose estimates to the genera l population in the site vicinity are calculated annually using a five year meteorological database. Determining whether detailed monthly do se estimates are necessary or whether annual averaged data is sufficie nt offers the potential for more efficient dose prediction. In this st udy, off site collective committed doses and maximum individual doses due to atmospheric tritium releases were calculated according to the m ethods outlined in the U.S. Nuclear Regulatory Commission's Regulatory Guide 1.109 and compared using monthly versus five-year meteorologica l data and source terms. Site-specific variables not currently utilize d at SRS for annual dose estimates also have been included. In additio n, the range of predicted doses, based on the distribution in model pa rameters given in the literature, were estimated. Finally, a sensitivi ty analysis was performed in order to determine the influence of model inputs on dose estimates. Results corroborate previous studies by ind icating that the primary contributor to infant tritium dose is the ing estion of milk, while for all other age groups, the most important pat hway is the ingestion of vegetation. These relative pathway contributi ons remain constant throughout the year for infants; for children, tee nagers, and adults, however, inhalation and absorption of tritium thro ugh the skin increases in relative importance in the months of June to September. It was found that the model utilized was most sensitive to dose factors, the ratio of the specific activity of tritium oxide in vegetation to the specific activity of atmospheric tritium oxide, and breathing rates. Most importantly, it was found that over a five-year period, the use of averaged meteorological data results in total indiv idual doses that are only 2 to 6% higher than doses determined monthly , depending on the pathway of interest.