Uncertainty analysis of correlated stability constants

Every attempt of using a computer to model reality has two main uncertainties: the conceptual uncertainty and the data uncertainty. The conceptual uncertainty deals with the choice of model selected for the simulation and the data uncertainty is about the precision and accuracy of the input data. They are often determined experimentally and may thus be encumbered by a number of uncertainties. Normally when treating uncertainties in input data these data are treated as independent variables. However, since many of these parameters are determined together they are actually correlated. This paper focuses on chemical stability constants, a most important parameter for chemical calculations based on speciation. Commonly in the literature they are at best given with an uncertainty interval. We propose to also give the covariance matrix thus giving the opportunity to really assess correlations. In addition we discuss the effect of these correlations on speciations.