Accounting for Spatial Relationships in Probabilistic Risk Analysis.* Jeanne C. Simpson, Golder Associates, 4104 148th Ave. NE, Redmond, WA 98052
There are two important spatial relationships that should be considered in public health risk assessments: 1) the spatial distribution of contaminants of concern over the area of interest; and 2) the spatial distribution of individuals who may be exposed to the contaminants of concern. Geostatistical techniques (kriging) are used to predict average concentrations, standard deviation of the predictions and spatial correlations of the predictions within a grid over the area of interest. The spatial distribution of individuals within this grid, under recreational and residential scenarios are developed. This spatial distribution of possible use patterns by individuals is incorporated with the spatial distribution of the contaminants to produce an exposure distribution for the individuals. The exposure distribution includes estimates of both the stochastic variability inherent in the population of individuals and the uncertainties about that population and the magnitude and location of the contaminants of concern. An example of the development of an exposure distribution using public domain geostatistical software and simple spreadsheet based Monte Carlo programs is given.
*Work funded by Monsanto Corporation and Golder Associates, Inc.