Problems Associated with Planning BOD Experiments

Possible inadequacies of commonly used long-term BOD experimental strategies are studied using actual BOD data and the first-order reaction model as a vehicle of analysis. The relation between the times and numbers of observations and the statistical quality of the estimated parameters k and L a is shown, where statistical quality is measured by the size of the joint confidence region and the degree of parameter correlation. Merely making a large number of observations does not necessarily ensure that precise parameter estimates will be obtained. On the other hand, if the observations are efficiently located a small number of observations can give precise parameter estimates.