Targeting lead in the multimedia environment in the continental United States.

The U.S. Environmental Protection Agency's (EPA) lead attainment strategy for air is being expanded to address geographic areas with the potential for multimedia, multipathway exposures to lead. Geographic Information Systems (GIS) technology is used to coordinate information from various databases to identify areas of potential concern. The data retrieval and decision processes used in identifying priority sources from each medium and in evaluating identified areas of concern are described in this paper. Only EPA databases with reliable locational information were used to facilitate accurate mapping and allow correlation with other data sources. The sources of lead loadings to air, water, and soils were mapped using either latitude and longitude or zip code, or county centroids for data lacking longitudinal and latitudinal coordinates (such as the drinking water data). A multimedia cluster of lead sources was identified at the county level, since all the facility data in the five databases could be mapped to this level. An impact factor and weighting system was devised to combine the information on the number of facilities and their relative size in developing a ranking of the multimedia lead clusters of concern in each region. The counties with the highest number of points were considered clusters of highest concern for multimedia lead sources. Two separate lists of the clusters were developed according to a point system. One identified 10 multimedia lead clusters in each of the 10 EPA regions, and the other identified the 100 clusters of highest concern in the country as a whole. The project is designed to be a first step in targeting future efforts to identify potential environmental problems associated with lead. The analyses presented in this paper provide a first look at the areas in the country where there is a potential for multimedia exposure to lead. A more refined analysis at the zip code level was subsequently developed to provide a good understanding of the issues pertaining to potential exposure at the neighborhood level. The results of this analysis will ultimately help the EPA and the states to target implementation and enforcement in areas of high potential lead exposures.

Performance comparison of a neural network with human observers on a visual target detection task.

An experiment is described which compares the performance of a neural network to human performance on a visual task which consists of detecting a target in a background image of correlated noise. A three-layer, feed-forward, multi-layer perceptron is trained to indicate the presence or absence of a target in images also presented to human observers. The basis for the comparison between the network and the human observers is the receiver operating characteristic (ROC) curve. Network performance is comparable to human performance for this particular task.