The seasonal wave with adjustment for streamflow model (seawave-Q) is combined with a stochastic model for the random deviations of observed daily concentrations from modeled concentrations.   The deviations are modeled using a first-order autoregressive process with seasonal variance and observational error.  Model parameters are estimated using maximum likelihood estimation.  A modified Kalman filtering algorithm is used to compute the likelihood function and predict daily concentrations given sparse and censored concentration data.  Conditional simulations can be used to estimate the probability of exceeding specified benchmarks for n-day moving-average concentrations.  The methodology is demonstrated using simulated and actual data from selected locations.  A procedure for evaluating benchmark exceedances for unsampled locations is proposed, in which (1) streamflow data are replaced by precipitation and temperature data, and (2) exceedance probabilities from the stochastic model are conditioned on concentration statistics from the Watershed Regressions for Pesticides (WARP) model.

Stochastic model for daily pesticide concentrations, and application to estimating benchmark exceedance probabilities

248th National Meeting (2014)

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