Mapping ambient urban air pollution at the small area scale : a GIS approach

Air pollution is an emotive and complex issue, affecting materials, vegetation
growth and human health. Given that over half the world's population live
within urban areas and that those areas are often highly polluted, the ability to
understand the patterns and magnitude of pollution at the small area (urban
environment) level is increasingly important. Recent research has highlighted,
in particular, the apparent relationship between traffic-related pollution and
respiratory health, while the increasing prevalence of asthma, especially
amongst children, has been widely attributed to exposure to traffic-related air
pollution. The UK government has reacted to this growing concern by
publishing the UK National Air Quality Strategy (DOE 1996) which forces all
Local Authorities in England and Wales to review air quality in their area and
designate any areas not expected to meet the 2005 air quality standards as Air
Quality Management Areas (AQMAs), though what constitutes AQMAs and
how to define them remains vague.

Against this background, there is a growing need to understand the patterns and
magnitude of urban air pollution and for improvements in pollution mapping
methods. This thesis aims to contribute to this knowledge. The background to
air pollution and related research has been examined within the first section of
this report. A review of sampling methods was conducted, a sampling strategy
devised and a number of surveys conducted to investigate both the spatial
nature of air pollution and, more specifically, the dispersion of pollution with
varying characteristics (distance to road, vehicle volume, height above ground
level etc). The resultant data was analysed and a number of patterns identified.
The ability of linear dispersion models to accurately predict air pollution was
also considered. A variety of models were examined, ranging from the
simplistic (e.g. DMRB) to the more complex (e.g. CALINE4) model. The
model best able to predict pollution at specific sites was then used to predict concentrations over the entire urban area which were then compared to actual
monitored data. The resultant analysis, indicated that the dispersion model is
not a good method for predicting pollution concentrations at the small area
level, and therefore an alternative method of mapping was investigated. Using
the ARC/INFO geographical information system (GIS) a regression analysis
approach was applied to the study area. A number of variables including
altitude, landuse type, traffic volume and composition etc, were examined and
their ability to predict air pollution tested using data on nitrogen dioxide from
intensive field surveys. The study area was then transformed into a grid of
10m2, regression analysis was performed on each individual square and the
results mapped. The monitored data was then intersected with the resultant
map and monitored and modeled concentrations compared. Results of the
analysis indicated that the regression analysis could explain up to 61 per cent of
the variation in nitrogen dioxide concentrations and thus performed
significantly better than the dispersion model method. The ease of application
and transferability of the regression method means it has a wide range of
applied and academic uses that are discussed in the final section.