PARENT SESSION

6A - LCIA - Toxicity/RA
Hall 2
8:30 AM - 10:15 AM, Wednesday, 30 April 2003
Chair: Jolliet, O.¹, ¹
Co-chair: McKone, T.², ²

(WE2/4) Comparison of results from probabilistic and statistical exposure models for life-cycle impact and sustainability assessments.

McKone, Thomas^{1, 2}, Bodnar, Agnes^{1, 2}, MacLeod, Matthew², ¹ University of California School of Public Health, Berkeley, California, USA² Lawrence Berkeley National Laboratory, Berkeley, California, USA

ABSTRACT- Life-cycle impact assessment (LCA) and sustainability assessment both use of source-to-dose calculations to quantify the potential impacts of industrial emissions. It is recognized that source-to-dose calculations involve model, parameter (data), and scenario uncertainties. The common approach for characterizing uncertainties in source-to-dose estimates is probabilistic modeling--an approach in which uncertainties are propagated through a model using repeated simulations, i.e. Monte Carlo modeling. An alternative approach is statistical uncertainty modeling. In contrast to the probabilistic approach, we use statistical analyses to obtain useful information from seemingly random measurements. These statistical models are developed directly from data to gain insights into the nature of randomness and to establish the relationship between factors such as emissions and intake. In this presentation we will explore the advantages and limitations of these two methods for characterizing uncertainties in source-to-dose relationships for persistent bio-accumulative pollutants. We express source-to-dose relationships using the intake fraction (iF), which is defined as the mass of chemical taken up by the receptor population divided by the mass of chemical released. Using both probabilistic multimedia exposure models and statistical models developed from emissions data, food surveys, and dose biomarkers, we develop probability distributions to characterize intake fraction for food-chain exposures to air emissions of dioxin compounds and polycyclic aromatic hydrocarbons (PAHs). We then compare the distributions obtained from the two approaches. In contrast to the PAH compounds, we find a better correlation between the multimedia probabilistic model results and the statistical model results for dioxin-like compounds. This appears to be due in part to the persistence, widespread distribution in the environment, and food-chain dominant exposures of the dioxin compounds.

Key words: source-to-dose, life-cycle impact, intake fraction, uncertainty