Estimating exposure to traffic-related pollution within a GIS environment

This thesis applies, evaluates and compares methods for estimating exposure to traffic-related pollution within a GIS environment. The methods were used in two contrasting case studies; Greater London and Sheffield, where they were selected on basis of data availability and resolution. The methods used in this research were CALINE3, DMRB, ADMS-Urban and ISC3 (air pollution dispersion models), kriging and co-kriging (spatial interpolation), SAVIAH (regression method) and traditional exposure indicators. Calculated estimates were validated by comparing them to monitored NO2 data. In the Sheffield case study the best methods were then used to analyse relationships between traffic-related pollution and respiratory health. Evaluation of the performance of the various methods found that none of the methods used in Greater London worked very well, although ISC3 and kriging tended to give more reliable results. In Sheffield DMR.B and SAVIAH gave the best estimates of monitored pollution levels. Traditional exposure indicators were only used in Sheffield of which ‘density of main roads within 150 metres’, ‘traffic flow within 150 metres’ and ‘HGV flow within 150 metres’ provided the most reliable estimates. In general, the quality of all exposure measures was highly dependent on the quality of input data. This is largely due to the fact that most variation of traffic-related pollution occurs close to main roads. In Greater London the quality of data was clearly inadequate. In Sheffield, where data was of a higher quality, results were better. No substantial or significant associations were found between the exposure measures and health outcome in the Sheffield case study. In Sheffield, this research also showed that passive sampling of NO2 provided a reliable measure of relative levels of air pollution across an urban area. It also showed that none of the models were able to detect raised NO2 concentrations due to accumulation of pollution from the city, as a result of wind direction. The results of this research show that, although the methods used here can help in the investigation of relationships between traffic-related pollution and health, there is a major need to improve methods for modelling exposure to air pollution. An important development could be to link different models together within a GIS environment, in order to improve the ability to use available information and exploit the different capabilities of the models. In order to detect the effects of traffic-related pollutants on chronic health, estimates are needed across large populations. Linkage of the methods applied here, would be particularly useful to model spatial and temporal variations in these types of studies