Land Use Regression Approach to Model NO
2
–Concentrations in
a Swedish Medium-City
Mateus Habermann
1
, Monica Billger
1
, Marie Haeger-Eugensson
2,3
1
Department of Architecture, Chalmers University of Technology, Gothenburg, Sweden.
2
Department of Earth Sciences, University of Gothenburg, Gothenburg, Sweden.
3
COWI AB, Gothenburg, Sweden.
E-mail: mathab@gmail.com
Abstract. In order to visualize the geographical distribution of air pollution concentration
realistically, we applied the Land Use Regression (LUR) model in the urban area of Gothenburg,
Sweden. The concentration of NO
2
was obtained by 25 passive air samplers during 7-20 May, 2001.
Explanatory variables were estimated by GIS in buffers ranging from 50 to 500 m-radii. Linear
regression was calculated, and the most robust were attained to the multiple linear regression.
Additionally, the LUR model was compared with a dispersion model. The final model explained 81.7%
of the variance of NO
2
concentration with presence of sum of traffic within 150 m and altitude as
predictor variables. Mann-Whitney Test did not exhibit significant difference between yearly
concentrations of NO
2
measured by regulatory measurement sites and measurements from passive
samplers, thus LUR model was extrapolated for later years and mapped. The extrapolation indicated
more elevated levels of pollution for the years 2003, 2006 and 2010. The results highlight the
contribution of traffic on air quality and suggest that LUR modelling may explain the variations of
atmospheric pollution with good accuracy. In addition, the model puts focus on spatial and temporal
variability needed to describe retrospective exposure to air pollution in studies that evaluate health
effects.
Keywords: Air pollution; nitrogen dioxide; exposure modeling; geographic information system; LUR
model.
1 Introduction
Exposure assessment can be used to evaluate, at various levels of detail, the degree and linkage between
contaminant sources and concentration of hazards and receptors (e.g. humans) in the environment by
studying different exposure pathways (e.g. air, water, and soil) and routes (e.g. inhalation, ingestion,
and dermal contact) between them. As one kind of exposure assessment air pollution exposure
assessment indicates human exposure to air pollutants [1].
The most exposed receptors to air pollution include either individuals whose residences, study or work
places offices are located near to heavy traffic roads or individuals who remain long time on roads (bus
drivers, traffic guards, street vendors etc.). Therefore the environment may influence the exposure to
pollutants and thereby trigger various outcomes [2].
Although the literature has documented significant variation of outdoor air pollution at small scales
within urban areas for important pollutants such as NO
2
and black smoke [3-5], many studies assessed
exposure based only on the proximity to polluted source, e.g. proximity to busy traffic [6]. This
approach is limited because it disregards other parameters that may influence the dispersion of
pollutants such as altitude, land use, population, road type, traffic intensity, temperature and
atmospheric stagnation [7]. Therefore, recent models became more refined including some of those
parameters [1].
Ordinary dispersion modelling requires good databases which are updated frequently (at least every
five years); however, measurements are expensive if they are conducted at numerous places [1] and when
dispersion modelling (DM) does not include measurements the computational analysis can be labour
intensive. In many of urban areas levels of air pollution frequently exceed environmental standards, it is
Environmental Pollution and Protection, Vol. 3, No. 3, September 2018
https://dx.doi.org/10.22606/epp.2018.33001 71
Copyright © 2018 Isaac Scientific Publishing EPP