Predictive Uncertainty and Parameter Sensitivity of a Sediment-Flux Model: Nitrogen Flux and Sediment Oxygen Demand

Estimating model predictive uncertainty is imperative to informed environmental decision making and management of water resources. This paper applies the Generalized Sensitivity Analysis (GSA) to examine parameter sensitivity and the Generalized Likelihood Uncertainty Estimation (GLUE) to estimate the predictive uncertainty of a sediment-flux water quality model with previously reported nitrogen and SOD data obtained from Chesapeake Bay sediment samples. The sediment-flux model comprises analytical solutions which describe the processes of ammonia production by anaerobic decomposition of sediment organic matter, nitrate production by nitrification, diffusive transport, denitrification, and sediment oxygen demand (SOD). Results show that total ammonia flux sensitivity to the parameters is more complex than that of SOD and nitrate flux. The diffusive boundary-layer thickness, which control mass transfer across the sediment-water interface, is identified to be the most sensitive parameter. Preliminary results show that the posterior model predictive uncertainties are overly underestimated as significant number of observations fall outside the GLUE-estimated 90% confidence limits. These results are prelude to further and refined analysis to improve the estimation of model predictive uncertainty and posterior probability distributions of the parameters.