Urban-scale modelling of reactive air toxics. Does chemical transformation matter?
CSIRO, December 2008
- Urban-scale modelling of reactive air toxics. Does chemical transformation matter? (PDF - 3,517 KB) | (RTF - 18,369 KB)
This report documents work undertaken for the Clean Air Research Programme project "Modelling Reactive Air Toxics and Exposure", the aim of which is to investigate the level of chemical detail that is required to accurately and defensibly model the urban exposure of selected reactive air toxics (formaldehyde, toluene and xylene).
The project involved the use of a three-dimensional numerical weather prediction and chemical transport system to model air toxic concentrations in Melbourne and Sydney. Formaldehyde, toluene and xylene were modelled both as chemically reactive species and as non-reactive tracer species. Differences in peak 24-hour concentrations and, (where appropriate) annual average concentrations were used to assess the veracity of using the tracer assumption to model these species. It was found that formaldehyde concentrations were underpredicted when chemical transformation was not considered, and this occurred because formaldehyde is generated as an intermediate product of the oxidation of other volatile organic compounds present in the urban atmosphere. Ignoring chemical transformation caused peak 24-hour concentrations to be underpredicted by up to 80%. When concentration was weighted by population density (as a surrogate for health impact), impacts were underpredicted by 30-60%. As a result it is recommended that future urban airshed modelling of formaldehyde should consider the chemical production and loss pathways of this species. In the case of toluene and xylene, use of the tracer assumption caused peak 24-hour and annual average concentrations to be slightly over predicted (less than 1%) while the exposure outcome was overpredicted by up to 30%. However the modelled peak concentrations of toluene and xylene were small compared to NEPM investigation levels (typically less than 15%) and thus the use of the tracer assumption and the generation of a slightly conservative outcome appears to be justified.