As we said above such models have a "design accuracy" of + or - 50% and are generally considered OK if accurate to within a factor of 3.

We must also realise that the model outlined above dos not allow for reflection from an inversion.

The main advantage- of this approach is its wide acceptance, to the extent that in several jurisdictions it is the basis of legally required planning legislation when chimney heights are considered.

If we plot distance along axis y or z as the controlled variable against concentration of pollutants we will see a distribution with a mean corresponding to the centreline with value H metres as the value of z if we measure z from the base of the stack.

The standard deviations of these distributions, which both theory and observation expect us to believe will be normal distributions, are the dispersion coefficients which we require for our calculations.

We will recall that the model is based on a good theoretical basis but the values of dispersion coefficients are, if the model is to be of any use at all based, at least to some degree on empirical measurements.

Let us look first of all at the most commonly used system for dispersion coefficient determination and then at some other alternatives.