PREDICTING THE OCCURRENCE OF ATRAZINE IN SHALLOW GROUND WATER BENEATH AGRICULTURAL AREAS OF THE UNITED STATES
Data from 52 NAWQA studies of the quality of shallow ground water beneath agricultural areas, conducted within the study units since 1992, were used to develop and apply linear regression models that can be used to predict the occurrence of atrazine, a widely used agricultural herbicide, in areas that have not yet been monitored. The detection frequency of atrazine in each of the 52 studies was correlated with more than 100 explanatory variables that represent aspects of the physical setting and human-related factors. Fifty-eight percent of the variability in atrazine-detection frequency was explained by a five-variable model whose terms describe (1) the rate with which atrazine is 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, and (4) the average available water-holding capacity and vertical permeability of the soils. Application of the model to each 1-km2 area of the conterminous U.S. that is at least 50 percent or more agricultural land indicates that application rate alone is an insufficient variable for predicting the occurrence of atrazine in shallow ground water because soil characteristics and agricultural-management practices are important factors in determining the likelihood that atrazine will be transported to ground water.