EVALUATING SOURCES OF ARSENIC IN GROUNDWATER IN VIRGINIA USING A LOGISTIC REGRESSION MODEL

We used a logistic regression model to investigate if geologic and/or other factors are linked to the probability of having elevated arsenic (As) concentrations above 5 parts per billion (ppb) in groundwater in the state of Virginia (USA). Measured As concentrations in groundwater (n = 5,632) were used as the dependent variable. Geologic units, lithology, soil series and texture, land use, and physiographic province were used as initial regressors (independent variables) in the model. However, due to multicollinearity issues, during model refinement we focused attention solely on geologic units. The presence of three geologic units at a spatial location, including Triassic-aged sedimentary rocks and igneous intrusives of the Culpeper Basin, and Devonian-aged shales/sandstones in the northwestern region of Virginia, resulted in a higher probability for As occurrence above 5 ppb in groundwater. Due to the few (<5%) observations having As concentrations > 5 ppb in our data set, caution is needed when applying the model to predict As concentrations in other parts of the state. However, our results are useful for identifying areas of Virginia, defined by underlying geology, that are more likely to have elevated As concentrations in groundwater. Due to the ease of obtaining publically available data and the accessibility of Geographic Information Systems, the study approach can be applied to other areas with existing datasets of As concentrations in groundwater and accessible data on geology, soils, and other environmental factors.