Modelling Air Quality in the Vicinity of Major Roads in Auckland, New Zealand

Abstract

In this study, recently-recorded CO, NOX and PM10 concentrations from the ROADSIDE project have been combined with traffic flow data to assess air quality at sites close to a busy section of the Auckland Southern Motorway. The results are compared against predictions from a model designed for and tested on New Zealand roads and conditions, but which has yet to be tested on motorways. The importance of the study lies in its potential to contribute to an air-quality control programme, to identify strengths and limitations of the model, and to quantify the predictive power of the model. Motorway emission rates were found to depend largely on commuter behaviour, with emissions rising sharply during the morning and evening rush hours. The expelled pollutants may linger for well into the day, and are compounded by evening traffic, which adds pollutants to cooling air at the same time that wind speeds decline. Trucks, which may travel at any time of the day or night, and which have high emissions/vehicle, form a substantial portion of motorway traffic. This combination of factors ensures that air quality along the motorway is of low quality much of the time. To predict pollutant concentrations, a Gaussian-plume-based box-model approach known as SOSE (site-optimised semi-empirical model), developed by Dirks in 2002, was used. In SOSE, parameters are derived from linear regression of known concentrations versus reciprocal wind speeds. These parameters are then used as constants to predict unknown concentrations at the same or similar sites. In this study, model inputs were the calculated emission rates, and measured wind speeds at the ROADSIDE sites. The model could adequately predict CO and NOX levels at sites offset from the motorway, but was less successful at predicting PM10 levels, and failed at the adjacent Deas Place Reserve site for all pollutants. It was also found that SOSE tends to over-predict low concentrations and under-predict high ones, which is a limitation of the linear regression approach. We conclude that SOSE works well provided that one of its main premises is satisfied; that is, that concentration is inversely proportional to wind speed, and we explore reasons why this might not be the case for sites adjacent to the motorway.