Sensitivity and uncertainty analysis for river quality modelling

Abstract

Uncertainty analysis for model simulation is of growing importance in the field of water quality management. The importance of this concern is provided by recent public awareness over health risks from improper disposal of toxic wastes as well as by the continuing emphasis on risk assessment. The first step in the chain of risk assessment is the quantification of the error in predicting water quality. In each mathematical modelling application, different uncertainties are involved. The uncertainty sources can be classified into different categories (in this study, as model-input uncertainty, model-structure uncertainty, model-parameter uncertainty and measurement errors). These different types of uncertainty sources determine collectively the total uncertainty in the model results. In this paper, the relative contributions of uncertainties associated with each source are studied for the physico-chemical water quality modelling of a river in Belgium. This provides information as to where available modelling resources should be focused.