A steady-state model of pesticide leaching through the unsaturated zone was used with readily available hydrologic, lithologic, and pesticide characteristics to estimate the vulnerability of the near-surface aquifer to atrazine contamination from non-point sources in Kent County, Michigan. The modelcomputed fraction of atrazine remaining at the water table, RM, was used as the vulnerability criterion; time of travel to the water table also was computed. Model results indicate that the average fraction of atrazine remaining at the water table was 0.039 percent; the fraction ranged from 0 to 3.6 percent. Time of travel of atrazine from the soil surface to the water table averaged 17.7 years and ranged from 2.2 to 118 years.

Three maps were generated to present three views of the same atrazine vulnerability characteristics using different metrics (nonlinear transformations of the computed fractions remaining). The metrics were chosen because of the highly (right) skewed distribution of computed fractions. The first metric, rm = RM^λ (where λ was 0.0625), depicts a relatively uniform distribution of vulnerability across the county with localized areas of high and low vulnerability visible. The second metric, rm^λ-0.5, depicts about one-half the county at low vulnerability with discontinuous patterns of high vulnerability evident. In the third metric, rm^λ-1.0 (RM), more than 95 percent of the county appears to have low vulnerability; small, distinct areas of high vulnerability are present.

Aquifer vulnerability estimates in the RM metric were used with a steady-state, uniform atrazine application rate to compute a potential concentration of atrazine in leachate reaching the water table. The average estimated potential atrazine concentration in leachate at the water table was 0.16 μg/L (micrograms per liter) in the model area; estimated potential concentrations ranged from 0 to 26 μg/L. About 2 percent of the model area had estimated potential atrazine concentrations in leachate at the water table that exceeded the USEPA (U.S. Environmental Protection Agency) maximum contaminant level of 3 μg/L.

Uncertainty analyses were used to assess effects of parameter uncertainty and spatial interpolation error on the variability of the estimated fractions of atrazine remaining at the water table. Results of Monte Carlo simulations indicate that parameter uncertainty is associated with a standard error of 0.0875 in the computed fractions (in the rm metric). Results of kriging analysis indicate that errors in spatial interpolation are associated with a standard error of 0.146 (in the rm metric). Thus, uncertainty in fractions remaining is primarily associated with spatial interpolation error, which can be reduced by increasing the density of points where the leaching model is applied.

A sensitivity analysis indicated which of 13 hydrologic, lithologic, and pesticide characteristics were influential in determining fractions of atrazine remaining at the water table. Results indicate that fractions remaining are most sensitive to the unit changes in pesticide half life and in organic-carbon content in soils and unweathered rocks, and least sensitive to infiltration rates.

The leaching model applied in this report provides an estimate of the vulnerability of the near-surface aquifer in Kent County to contamination by atrazine. The vulnerability estimate is related to water-quality criteria developed by the USEPA to help assess potential risks from atrazine to the near-surface aquifer. However, atrazine accounts for only 28 percent of the herbicide use in the county; additional potential for contamination exists from other pesticides and pesticide metabolites. Therefore, additional work is needed to develop a comprehensive understanding of the relative risks associated with specific pesticides. The modeling approach described in this report provides a technique for estimating relative vulnerabilities to specific pesticides and for helping to assess potential risks.