T8 PM Measurement/ Estimation of Environmentally Relevant Physico-Chemical Properties
Tuesday, 15 November 2005: 1:50 PM - 5:30 PM in 337-338

(MAC-1117-798159) A least-squares adjustment procedure to improve accuracy of physico-chemical properties of organic compounds.

Schenker, Urs¹, MacLeod, Matt¹, Scheringer, Martin¹, Hungerbühler, Konrad¹, ¹ Swiss Federal Institute of Technology, Zurich, Switzerland

ABSTRACT- Partitioning properties (vapor pressure, water solubility, octanol solubility, the partition coefficients between air, water, and octanol, and their temperature dependencies) are used to describe the behavior of organic contaminants in environmental fate and exposure models. If such quantities are measured, random measurement errors will appear and systematic errors are possible, leading to datasets that are not internally consistent. Beyer et al. (Environ. Toxicol. Chem., 2002; p. 941-953) have suggested an adjustment procedure that calculates a set of internally consistent property data which, they claim, is minimally divergent from the measured quantities. We show that their adjustment procedure does not actually minimize the adjustments applied to the measured quantities, and that the adjusted values differ unnecessarily from the measured quantities. We suggest a new adjustment procedure that is based on the theory of least squares and show that the adjusted values calculated with the new method are always closer to the measured quantities than the values calculated with the Beyer method. The requirement of the least-squares theory (normal distribution of measurement errors) allows us to identify bias and systematic measurement errors if series of property data are reported. Such a bias is evident in properties of a recently reported PCB dataset (Li et al.; J. Phys. Chem. Ref. Data, 2003; p. 1545-1590). We discuss possible reasons for this bias and suggest new values for the physico-chemical properties of the PCB congeners. The new least-squares adjustment procedure also incorporates a more rigorous approach of dealing with measurement uncertainties: variances of the measured quantities can be propagated through the adjustment procedure, and variances (and confidence intervals) of the adjusted quantities can be indicated. Such information improves accuracy of model input data and helps to perform uncertainty analyses (such as Monte-Carlo simulations) of environmental fate and exposure models.

Key words: fate modeling, partition coefficients, thermodynamic constraints, measurement uncertainty