Development of algorithms and approximations for rapid operational air quality modelling

In regulatory and public health contexts the long-term average pollutant concentration in the vicinity of a source is frequently of interest. Well-developed modelling tools such as AERMOD and ADMS are able to generate time-series air quality estimates of considerable accuracy, applying an up-to-date understanding of atmospheric boundary layer behaviour. However, such models incur a significant computational cost with runtimes of hours to days. These approaches are often acceptable when considering a single industrial complex, but for widespread policy analyses the computational cost rapidly becomes intractable. In this paper we present some mathematical techniques and algorithmic approaches that can make air quality estimates several orders of magnitude faster. We show that, for long-term average concentrations, lateral dispersion need not be accounted for explicitly. This is applied to a simple reference case of a ground-level point source in a neutral boundary layer. A scaling law is also developed for the area in exceedance of a regulatory limit value.