Due to fiscal and practical constraints, surface-water monitoring programs collect samples infrequently, usually at well less than every 7-days. As a result, peak contaminant concentrations are often not sampled, and the model-based estimation of peak values is challenging, requiring suitable covariates for the non-sampled days. Unfortunately, few covariate options are available, with the best proposed to date being derived from average daily streamflow (or flow), which is readily measured on all days, both sampled and non-sampled. Flow can correlate with concentration because precipitation induces runoff from fields, simultaneously increasing both flow and contaminant concentration.  The SEAWAVE-QEX model predicts the mean of the underlying contaminant process using a regression model having a multi-year linear trend, a seasonal wave, and two flow-derived covariates: the short-term flow anomaly (STFA) and the mid-term flow anomaly (MTFA). These flow covariates use the idea that the correlation of flow with concentration will depend on the deviation of recent flow from historical averages. Other transformations of flow are also possible, one example being a normalized Box-Cox flow.  This presentation compares the performance of SEAWAVE-QEX and alternative streamflow models in the prediction of maximum 1-day and 60-day rolling averages of atrazine concentration using 7-day and 14-day samples collected from the Atrazine Ecological Exposure Monitoring Program. Best models for each sampling design-target quantity combination will be identified. Coverage of prediction intervals will be evaluated.

Prediction of atrazine concentrations from monitoring data II: SEAWAVE-QEX and related streamflow models

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.