Showing papers on "Uncertainty quantification published in 2000"

Watershed-level uncertainties : Implications for phosphorus management and eutrophication

Abstract: Eutrophication is often the critical problem associated with surface waters having impaired water quality. Managing eutrophication in lakes and reservoirs typically involves the control of phosphorus inputs since it is often the limiting nutrient and because nitrogen and carbon are difficult to control. Management activities to address water quality impairment require a watershed-level approach. However, uncertainty and stochasticity are ubiquitous in watershed-level assessment and management. We describe and demonstrate a method for incorporating uncertainty analyses in watershed-level modeling using Wister Lake in Oklahoma as an example. Natural stochasticity, parameter error, and error due to aggregation of data are quantified separately to allow for a comparison of the various sources of uncertainty in modeling activities. Parameter error was found to be the greatest contributor to overall output uncertainty. Several example management scenarios are presented to illustrate the utility and flexibility of including uncertainty quantification in watershed-level management and decision making.

Methodology for characterizing modeling and discretization uncertainties in computational simulation

Abstract: This research effort focuses on methodology for quantifying the effects of model uncertainty and discretization error on computational modeling and simulation. The work is directed towards developing methodologies which treat model form assumptions within an overall framework for uncertainty quantification, for the purpose of developing estimates of total prediction uncertainty. The present effort consists of work in three areas: framework development for sources of uncertainty and error in the modeling and simulation process which impact model structure; model uncertainty assessment and propagation through Bayesian inference methods; and discretization error estimation within the context of non-deterministic analysis.

Mathematical and Numerical Approaches in Performance Assessment for Radioactive Waste Disposal: Dealing with Uncertainty

Abstract: Publisher Summary This chapter discusses how uncertainty should be treated in a performance assessment (PA) for a radioactive waste disposal facility or some other complex system. In such assessments, uncertainty is often divided into two components: (i) stochastic or aleatory uncertainty, which arises as the system under study can potentially behave in many different ways, and (ii) subjective or epistemic uncertainty, which arises from a lack of knowledge about quantities that are assumed to have fixed values within the computational implementation of the PA. A PA for a radioactive waste disposal site or some other facility, is a complex mathematical calculation. In such a calculation, it is important to have a clear view of both the formal mathematical components on which the PA is based and the numerical procedures used in the approximation of these components. In addition, appropriate numerical procedures and software that implements these procedures are essential parts of a PA. However, it is also important to maintain the conceptual distinction between a mathematical model and the numerical procedures used to implement that model.

Uncertainty Quantification for Delineation of Wellhead Protection Areas Using the Gauss‐Hermite Quadrature Approach

Abstract: Missing from most wellhead protection area (WHPA) delineation studies is a measure of the uncertainty associated with model predictions. A quantitative representation of that uncertainty can be used by regulators to implement different degrees of protection for areas with different degrees of certainty. Uncertainty analysis was performed for one- and five-year WHPA delineation for two municipal wells within a buried-valley glacial-outwash aquifer in southwestern Ohio. An approximation of the three-point Gauss-Hermite quadrature formula was used. This method is an alternative to simple Monte Carlo random sampling, typically requiring fewer model runs. It results in model-prediction expected values and variances and quantifies parameter main and two-way interactive effects. This study explored the collective impact of the uncertainties associated with six model parameters. Parameter probability density functions (PDFs) were based on field data and modified using a ground water flow model to ensure that all combinations of parameter values, within their PDFs, yielded an acceptable model calibration. The one-year WHPA for the upgradient well had a high degree of associated uncertainty represented by the difference in size between the WHPA low and high 95% confidence interval limits. The large uncertainty was due to the uncertainties associated with model parameter values, especially effective porosity and horizontal hydraulic conductivity. River conductance also had a substantial impact on the WHPA predictions. The WHPAs for the downgradient well and the five-year WHPA for the upgradient well were limited by upgradient hydrogeologic boundaries; their prediction was, therefore, less sensitive to the uncertainties inherent in the model parameters.