Model assumptions and parameters used in an earlier study of the vulnerability of ground water in Kent County, Michigan, to atrazine contamination were reviewed and compared with other studies. The review indicated that model assumptions are consistent with those used in other models and that the parameters assigned in the Kent County model are within the broad range commonly found in the literature. Model assumptions most likely to limit the accuracy of the previous study include those of uniform transport and steady-state flow. Simulation results are more sensitive to parameter estimates for atrazine half life, organic-carbon content, and organic-carbon partition coefficient than to other model parameters.

Potential atrazine detection probabilities and concentrations of atrazine reported in the previous Kent County Study were compared with data from other studies. Detection probabilities of 14 measurements of atrazine concentrations in ground water from shallow wells in the southern Lower Peninsula of Michigan were compared with detection probabilities based on potential atrazine concentrations computed for Kent County. Results indicate that the distribution of detection probabilities based on measured concentrations is similar to that based on adjusted potential concentrations. Potential concentrations were adjusted for effects of differences between sampling and modeling depths and for differences between the uniform application rate used for potential concentrations and the percentage of Kent County that is likely to be treated with atrazine. Potential concentrations of atrazine in the Kent County Study were within the wide range of concentrations measured in other states.

A stratified random sampling strategy was developed to verify expected atrazine concentrations in ground water within Kent County. The strategy helps identify strata, determine the optimum allocation of ground-water samples within defined strata, and project the sampling error. Implementation of the strategy was illustrated by use of potential atrazine concentrations computed in the previous Kent County Study. Once detailed information on historical application rates of atrazine is developed, expected atrazine concentrations can be computed by use of the vulnerability model and used to implement the sampling strategy. Sampling results may be used to verify the effectiveness of the vulnerability model and local estimates of historical atrazine application rates by use of analysis of variance.