REGRESSION MODEL FOR PREDICTING THE CONCENTRATION OF ATRAZINE RESIDUES IN SHALLOW AGRICULTURAL GROUNDWATER ACROSS THE CONTERMINOUS UNITED STATES

Tobit regression methods were used to develop a model for predicting the summed concentration of atrazine and deethylatrazine (referred to as atrazine residue) in shallow groundwater. Atrazine residue concentration data were collected from more than 1,000 monitoring wells located in various agricultural settings across the conterminous United States between 1993 and 2001 as part of the U.S. Geological Survey’s National Water Quality Assessment Program. Explanatory variables in the model describe the source and transport of atrazine residues in the subsurface and include the (1) intensity of atrazine application for agricultural purposes (kg/km²), (2) permeability of the least permeable soil layer and the average available water-holding capacity, (3) amount of artificial drainage (trenches and subsurface drains), (4) rate of recharge, and (5) redox condition of each sample. The model, which explains about 45 percent of the variability in atrazine residue concentrations measured in shallow agricultural groundwater, was used to predict atrazine residue concentrations in unmonitored areas as well as the probability of exceeding a specified concentration threshold.