Application of a spatial model to estimate lead concentrations in soil at skeet and trap ranges

The accumulation of lead at skeet and trap ranges poses a potential risk to people using the sites for recreational purposes. EPA risk assessment guidance recommends using the 95-percent upper confidence limit (95th UCL) of the arithmetic mean of site contaminant data as the exposure concentration in the risk assessment equation. The parametric, non-spatial methods included in EPA guidance produce an estimated exposure concentration that is often greater than the highest concentration observed on the site and implicitly assumes that each data point represents an equal area of the site. The common practice of locating samples in known or suspected "hotspots", or areas of high contamination ('sample clustering'), often produces a biased estimate of the exposure concentration when the arithmetic average of the sample data is used to estimate the exposure concentration.; Three methods were used to investigate the effect of sample clustering on the 95th UCL, using soil Pb data from five skeet and trap ranges. Two of the methods (normal-based and lognormal-based) are included in EPA guidance on estimating the exposure concentration; the third method was the bootstrap. The effect of sample clustering on the exposure concentration and the uncertainty in the estimated exposure concentration was determined by weighting the soil Pb data using Voronoi polygon areas.; The above analysis produced six different estimates of the exposure concentration and the uncertainty in the exposure concentration for each site. Comparison of the six estimates indicated the spatial arrangement of the data should be considered in the estimate of the exposure concentration. The spatially-weighted bootstrap method is preferred over the spatially-weighted normal and lognormal-based methods for the data analyzed.