Regression Methods for Estimating the Occurrence of Atrazine in Shallow Ground Water beneath Agricultural Areas of the United States

Ordinary least squares (OLS) methods were used to estimate the frequency with which atrazine would be detected in shallow ground water beneath agricultural settings based on results from 52 studies across the United States. Fifty-eight percent of the variability in atrazine detection frequency was explained by a 5-variable OLS equation whose terms describe (1) the amount of atrazine applied for agricultural purposes per unit area, (2) the presence of artificial drainage in the form of subsurface drains or surface trenches, (3) the amount of agricultural land in each study area, (4) the average available water-holding capacity of the soil, and (5) the vertical permeability of the soil. The final form of the OLS equation indicates that atrazine use alone is insufficient for estimating the frequency of atrazine detection in shallow ground water in unmonitored agricultural areas; soil characteristics and agricultural-management practices also appear to be important factors in determining the vulnerability of shallow ground water to atrazine contamination. Locations where artificial-drainage methods are not used — and, where atrazine use is most intensive — are those most vulnerable to atrazine contamination. Further analysis will allow for estimation of the presence or absence (logit regression) and the concentration (tobit regression) of atrazine at individual well sites in agricultural settings across the United States.