Paper No. 2
Presentation Time: 9:15 AM
SPATIAL VARIATION OF SOIL LEAD IN AN URBAN COMMUNITY GARDEN
BUGDALSKI, Lauren, Department of Geology, Wayne State University, Detroit, MI 48202, LEMKE, Lawrence D., Department of Geology, Wayne State University, 0224 Old Main, 4841 Cass, Detroit, MI 48202 and MCELMURRY, Shawn P., Civil and Environmental Engineering, Wayne State University, Detroit, MI 48202, bc0254@wayne.edu
Soil lead pollution resulting from historical industrial and residential land use practices is an issue of serious concern in urban communities throughout the developed world. In Detroit, Michigan, USA, air deposition from industrial point sources, emissions from mobile sources during the era of leaded gasoline, and residual contamination from lead based paint contribute to prevalent background levels of soil lead, even in residential areas. The recent emergence of an urban gardening movement in Detroit has increased awareness of the importance of thorough, accurate soil lead testing. The heterogeneous nature of soil can lead to variability at very small scales, making characterization of a city lot to be used for gardening difficult, however. USEPA guidelines for sampling standard city residential lots allow for as few as two surface composite samples comprised of 3 to 5 subsamples to characterize the soil lead concentration of an individual lot. In some circumstances, such minimal sampling schemes may lead to poor site characterization with areas of elevated concentration potentially overlooked.
The short-scale spatial variability of soil lead in a Detroit urban community garden was investigated using geostatistical methods to determine the number of samples needed to accurately characterize the lot. A total of 150 composite soil samples spaced ten, five, two, or one meter apart in a nested grid were taken across a 20 by 30 meter urban community garden and analyzed for total Pb using flame atomic absorption spectroscopy (AAS). Bioaccessibile lead, which serves as a proxy for bioavailable lead, was also measured for each sample. Omnidirectional variograms were constructed for Pb soil concentrations and bioaccessiblitiy, and ordinary kriging was used to interpolate values between measured locations. The results indicate notable spatial variability, even at one meter sample spacing. A cumulative distribution function of mapped soil lead concentrations was constructed and random sampling of the mapped concentrations was simulated to evaluate the probability of calculating a false negative (i.e., estimating average lead concentration below recommended action levels) and determine the number of samples required to accurately characterize the average soil lead concentration in an urban residential lot.
