The equal-intensity events special case models of odour annoyance provide robust logical frameworks within which detailed analyses may be undertaken of the efficacy of threshold and percentile based odour criteria to predict and limit annoyance.
Models in which both the frequency and intensity of impact events are assumed to be associated with annoyance highlight what appear to be significant shortcomings in the efficacy of single percentile odour criteria to predict nuisance odour impact extents. Consideration of these models suggests that a multi-percentile approach may significantly mitigate these shortcomings, however more sophisticated approaches may ultimately be required to maximise skill of dispersion modelling to predict nuisance odour impact extent.
Compared to single percentile criteria, the suggested multi-percentile approach allows for a wider range of meteorological conditions that may be important in terms of impacts to influence calculated separation distances. This attribute may potentially aid resolution of a noted disparity between complaints data and some single percentile model calculations (Schauberger et al., 2008). These models also strongly support a particular interpretation of the community annoyance curve concept proposed by Watts and Sweeten (1995) that resolves difficulties with an alternative interpretation of this concept suggested by Wallis and Cadee (2008).
The analysis undertaken in this paper suggests that the identified short-comings of single percentile criteria and differences in impact environments at study locations may have been contributing factors to some seemingly disparate dose response study findings. The reviewed empirical data appear to be broadly consistent with this hypothesis, but are not conclusive.
The models of annoyance developed here and the regulatory definition of nuisance odour proposed by Watts and Sweeten (1995) address the question of which percentile is “best” for describing odour exposure, however they raise new questions which require resolution. These relate to the limits of the range of event frequencies or percentiles to be considered in models of annoyance, the nature of adverse community response incurred at these percentile levels and the significance of additive effects when thresholds are approached or exceeded at multiple percentile levels in multi-percentile approaches.
While shortcomings of single percentile odour criteria have been previously identified by a number of researchers, to date these shortcomings have been largely overlooked by regulators. It is suggested that this earlier work, in combination with the models of annoyance developed here provide an improved concept of odour criteria that may assist in the development more skilful and better harmonised jurisdictional criterion frameworks.