Elsevier

Atmospheric Environment

Volume 40, Issue 3, January 2006, Pages 542-553
Atmospheric Environment

Predicting long-term average concentrations of traffic-related air pollutants using GIS-based information

https://doi.org/10.1016/j.atmosenv.2005.09.067Get rights and content

Abstract

Global regression models were developed to estimate individual levels of long-term exposure to traffic-related air pollutants. The models are based on data of a one-year measurement programme including geographic data on traffic and population densities. This investigation is part of a cohort study on the impact of traffic-related air pollution on respiratory health, conducted at the westerly end of the Ruhr-area in North-Rhine Westphalia, Germany.

Concentrations of NO2, fine particle mass (PM2.5) and filter absorbance of PM2.5 as a marker for soot were measured at 40 sites spread throughout the study region. Fourteen-day samples were taken between March 2002 and March 2003 for each season and site. Annual average concentrations for the sites were determined after adjustment for temporal variation.

Information on traffic counts in major roads, building densities and community population figures were collected in a geographical information system (GIS). This information was used to calculate different potential traffic-based predictors: (a) daily traffic flow and maximum traffic intensity of buffers with radii from 50 to 10 000 m and (b) distances to main roads and highways.

NO2 concentration and PM2.5 absorbance were strongly correlated with the traffic-based variables. Linear regression prediction models, which involved predictors with radii of 50 to 1000 m, were developed for the Wesel region where most of the cohort members lived. They reached a model fit (R2) of 0.81 and 0.65 for NO2 and PM2.5 absorbance, respectively. Regression models for the whole area required larger spatial scales and reached R2=0.90 and 0.82. Comparison of predicted values with NO2 measurements at independent public monitoring stations showed a satisfactory association (r=0.66). PM2.5 concentration, however, was only slightly correlated and thus poorly predictable by traffic-based variables (r<0.3).

We concluded that NO2 and soot can be considered truly traffic-related pollutants, and that GIS-based regression models offer a promising approach to assess individual levels of exposure to these pollutants.

Introduction

Traffic related emissions are today's leading cause of air pollution due to the international success in reducing emissions of pollutants from, e.g. domestic heating and industry in the past decades (Brunekreef and Holgate, 2002; World Health Organization, 2000; Technical Working Group on Particles, 1997). Therefore, the focus of recent research has shifted to health impacts of pollutants caused by emissions of motorized vehicles. Evidence on the health impact of PM, especially of its fine respirable fraction, has been cumulating since the early 1990s (Brunekreef and Holgate, 2002). There is an ongoing debate whether NO2 has any direct health impact or is simply an indicator for other traffic-related pollutants and its reaction products (WHO Working Group, 2003). Long-term exposure to air pollution is now of heightening concern, since chronic effects on respiratory health have been documented in several studies (Brauer et al., 2002; Hoek et al., 2002a; Abbey et al., 1999; Pope et al., 1995; Dockery et al., 1993).

Former studies in this field were mainly hampered by severe deficiencies in assessing the individual exposure of the study subjects. Since direct measurement of the total exposure for each study subject is not feasible, indirect methods of exposure assessment are necessary (Hertz-Picciotto, 1998). Between-city comparisons, which relied on city-wide averages, did not try to determine individual exposure and were prone to be confounded by intangible differences between the cities (Hoek et al., 2002a). Other studies used substitute measures such as distances to streets or self-reported exposure status, the validity of which might be questionable (Hoek et al., 2002b; Heinrich et al., 2005).

The development of geographical information systems (GIS), which have gained wide appreciation during the last 20 years, offers the opportunity to predict pollutant concentrations on a fine spatial scale (Burrough and McDonnell, 1998). Since more and more geographic data of high resolutions are being collected by public agencies, and the software is capable of overlaying such different data sets, the use of rather sophisticated methods for exposure assessment has become feasible. A convenient approach to include exogenous geographic information as well as actual measurements into exposure estimation models is regression mapping. The SAVIAH study (Briggs et al., 2000, Briggs et al., 1997) which applied this practicable technique to the estimation of NO2 annual average concentrations has a pioneer role in its propagation.

The present results have been obtained in the framework of a study conducted in the region of Wesel and the city of Duisburg in North-Rhine Westphalia (NRW), Germany, according to the protocol of the international study TRAPCA on the impact of traffic-related air pollution on the development of inhalant allergy, asthma and other respiratory conditions in children. In this paper, we explored whether and to what extent PM2.5, PM2.5 absorbance or NO2 concentrations could be reasonably predicted by traffic-based variables in an area, which comprises urban industrial cities and large adjacent rural areas. We computed a variety of traffic-related predictors in order to test their suitability.

A major interest was to find out which spatial scales were meaningful for predicting pollutant concentrations. On the basis of the available geographic information, we developed regression models and assessed their validity. Finally, we applied these models to the coordinates of the cohort member addresses to get individual estimates for the levels of exposure to traffic-related air pollution.

Section snippets

Study design

The prospective study design is similar to the German part of the TRAPCA study, which was carried out in the city of Munich in Bavaria (Brauer et al., 2003; Gehring et al., 2002; Hoek et al., 2001a). While this study was restricted to a highly urbanized region in southern Germany, the present study represents a wider range of rural, suburban and urban areas in the western part of Germany. The study centre is located in Wesel, a small town of about 60 000 inhabitants situated at the river Rhine

Study population

In the combined GINIplus and LISAplus cohort, 2922 birth addresses and 2457 current addresses located in NRW have been retained. For 1946 participants both addresses were available; 1426 (73.3%) of them had never moved during the follow-up period. From the data of the NRW land surveying office, 2674 (91.5%) birth addresses and 2122 (86.4%) current addresses could be geocoded. For 140 addresses, coordinates had to be taken from adjacent buildings in the same street. The prediction models were

Discussion

For NO2 concentration and PM2.5 absorbance we managed to establish well-fitted regression models. Relying on only three or four predictors, the models could explain a very substantial percentage of variation over the total area (R2=0.90 and 0.82, Table 4(b)) and an acceptable percentage of variation within the Wesel region (R2=0.81 and 0.65, Table 4(a)).

The concentration of PM2.5 mass has long been realized to be a large-scale phenomenon (Hoek et al., 2002b). This observation is strongly

Conclusion

Strong associations of traffic density with annual averages of NO2 concentration and PM2.5 absorbance as a proxy for soot, but not with PM2.5 mass concentration were observed. Total traffic figures in neighbourhoods of the coordinates of interest, which can be calculated using geographical information systems, turned out to be especially suitable predictors. Variables of large and small spatial scales had to be combined to establish highly predictive regression models. The application of such

Acknowledgement

The measurement campaign and the investigation of the six year old children were financed by a Grant of the BMU for the IUF (BMU A6 FKZ: 20462296). We thank the environment agency NRW for providing us with the data of traffic counts and the Geodata centre NRW for the data of building density. E. Jermann (IUF) did the NO2 determinations, thanks for that perfect job. We thank also the Research Institute for Prevention of Allergies and Airway Diseases, Department of Paediatrics, Marien-Hospital

References (30)

  • D.J. Briggs et al.

    Mapping urban air pollution using GIS: a regression-based approach

    International Journal of Geographical Information Science

    (1997)
  • I. Brockow et al.

    Beeinflussung des Stillverhaltens durch familiäre atopische Vorbelastung und unterschiedliche Information über atopiepräventive Ernährung

    Monatsschrift Kinderheilkunde

    (2003)
  • P.A. Burrough et al.

    Principles of Geographical Information Systems

    (1998)
  • J. Cyrys et al.

    Comparison between different traffic-related particle indicators: elemental carbon (EC), PM2.5 mass, and absorbance

    Journal of Exposure Analysis and Environmental Epidemiology

    (2003)
  • D.W. Dockery et al.

    An association between air pollution and mortality in six U.S. cities

    New England Journal of Medicine

    (1993)
  • Cited by (119)

    • Maternal exposure to ambient air pollution during pregnancy and lipid profile in umbilical cord blood samples; a cross-sectional study

      2020, Environmental Pollution
      Citation Excerpt :

      Our findings for the total street length across buffers of 300 m and 500 m were not conclusive. Previous studies have shown that higher street length as an indicator of traffic density was significantly correlated with higher levels of air pollution, especially in the smaller buffer sizes (e.g., 100 m) (Hochadel et al., 2006; Momtaz et al., 2019; Weichenthal et al., 2016). Similarly, we observed that the correlations between the total street length in a 100 m buffer and air pollutants were stronger compared to those for 300 m and 500 m buffers (Fig. 2).

    • Transit-oriented development and air quality in Chinese cities: A city-level examination

      2019, Transportation Research Part D: Transport and Environment
    View all citing articles on Scopus
    View full text