Recurrence of Extreme Concentrations

We have known for many years that atmospheric dispersion modeling should include not only predictions of mean concentrations of pollutants, but also information about possible extreme events, i.e. probabilistic statements about deviations from the predicted mean concentrations. One way of complying with this requirement is to predict, as a function of time and space, the variance together with the mean of the concentration. With an assumption about the form of the probability density it will then be possible to calculate the probability that, at any given time, the concentration will exceed a particular value. Relevant discussions can be found in Chatwin and Sullivan (1993) and references therein.