• Journal Article

FLOW-COVARIATE PREDICTION OF STREAM PESTICIDE CONCENTRATIONS

Citation

Mosquin, P. L., Aldworth, J., & Chen, W. (2017). FLOW-COVARIATE PREDICTION OF STREAM PESTICIDE CONCENTRATIONS. Environmental Toxicology and Chemistry. DOI: 10.1002/etc.3946

Abstract

Potential "peak" functions (e.g., maximum rolling averages over a given duration) of annual pesticide concentrations in the aquatic environment are important exposure parameters (or target quantities) for ecological risk assessments. These target quantities require accurate concentration estimates on non-sampled days in a monitoring program. In this paper, we examine stream flow as a covariate via universal kriging to improve predictions of maximum m-day (m = 1, 7, 14, 30, 60) rolling averages and the 95(th) percentiles of atrazine concentration in streams where data were collected every 7 or 14 days. The universal kriging predictions were evaluated against the target quantities calculated directly from the daily (or near daily) measured atrazine concentration at 32 sites (89 site-years) as part of the Atrazine Ecological Monitoring Program (AEMP) in the U.S. Corn Belt region (2008 to 2013) and 4 sites (62 site-years) in Ohio by the Nation Center for Water Quality Research (NCWQR) (1993 to 2008). Because stream flow data are strongly skewed to the right, three transformations of the flow covariate were considered: log transformation, short term flow anomaly, and normalized Box-Cox transformation. The normalized Box-Cox transformation resulted in predictions of the target quantities that were comparable to those obtained from loglinear interpolation (i.e., linear interpolation on the log-scale) for 7-day sampling. However, the predictions appeared to be negatively affected by variability in regression coefficient estimates across different sample realizations of the concentration time series. Therefore, revised models incorporating seasonal covariates and partially or fully constrained regression parameters were investigated, and they were found to provide much improved predictions in comparison to those from loglinear interpolation for all rolling average measures. This article is protected by copyright. All rights reserved.