The impact of the congestion charging scheme on air quality in London. Part 1. Emissions modeling and analysis of air pollution measurements.

TitleThe impact of the congestion charging scheme on air quality in London. Part 1. Emissions modeling and analysis of air pollution measurements.
Publication TypeJournal Article
Year of Publication2011
AuthorsKelly F, Anderson RH, Armstrong B, Atkinson R, Barratt B, Beevers S, Derwent D, Green D, Mudway I, Wilkinson P
Corporate AuthorsHEI Health Review Committee
JournalRes Rep Health Eff Inst
Issue155
Pagination5-71
Date Published2011 Apr
ISSN1041-5505
KeywordsAir Pollutants, Air Pollution, Carbon Monoxide, Environmental Monitoring, Humans, London, Models, Theoretical, Nitric Oxide, Nitrogen Dioxide, Particulate Matter, Research Design, Time Factors, Vehicle Emissions
Abstract

On February 17, 2003, a congestion charging scheme (CCS*) was introduced in central London along with a program of traffic management measures. The scheme operated Monday through Friday, 7 AM to 6 PM. This program resulted in an 18% reduction in traffic volume and a 30% reduction in traffic congestion in the first year (2003). We developed methods to evaluate the possible effects of the scheme on air quality: We used a temporal-spatial design in which modeled and measured air quality data from roadside and background monitoring stations were used to compare time periods before (2001-2002) and after (2003-2004) the CCS was introduced and to compare the spatial area of the congestion charging zone (CCZ) with the rest of London. In the first part of this project, we modeled changes in concentrations of oxides of nitrogen (NOx), nitrogen dioxide (NO2), and PM10 (particles with a mass median aerodynamic diameter < or = 10 microm) across the CCZ and in Greater London under different traffic and emission scenarios for the periods before and after CCS introduction. Comparing model results within and outside the zone suggested that introducing the CCS would be associated with a net 0.8-microg/m3 decrease in the mean concentration of PM10 and a net 1.7-ppb decrease in the mean concentration of NOx within the CCZ. In contrast, a net 0.3-ppb increase in the mean concentration of NO2 was predicted within the zone; this was partly explained by an expected increase in primary NO2 emissions due to the introduction of particle traps on diesel buses (one part of the improvements in public transport associated with the CCS). In the second part of the project, we established a CCS Study Database from measurements obtained from the London Air Quality Network (LAQN) for air pollution monitors sited to measure roadside and urban background concentrations. Fully ratified (validated) 15-minute mean carbon monoxide (CO), nitric oxide (NO), NO2, NOx, PM10, and PM2.5 data from each chosen monitoring site for the period from February 17, 2001, to February 16, 2005, were transferred from the LAQN database. In the third part of our project, these data were used to compare geometric means for the 2 years before and the 2 years after the CCS was introduced. Temporal changes within the CCZ were compared with changes, over the same period, at similarly sited (roadside or background) monitors in a control area 8 km distant from the center of the CCZ. The analysis was confined to measurements obtained during the hours and days on which the scheme was in operation and focused on pollutants derived from vehicles (NO, NO2, NOx, PM10, and CO). This set of analyses was based on the limited data available from within the CCZ. When compared with data from outside the zone, we did not find evidence of temporal changes in roadside measurements of NOx, NO, and NO2, nor in urban background concentrations of NOx. (The latter result, however, concealed divergent trends in NO, which fell, and NO2, which rose.) Although based upon fewer stations, there was evidence that background concentrations of PM10 and CO fell within the CCZ compared with outside the zone. We also analyzed the trends in background concentrations for all London monitoring stations; as distance from the center of the CCZ increased, we found some evidence of an increasing gradation in NO and PM10 concentrations before versus after the intervention. This suggests a possible intermediate effect on air quality in the area immediately surrounding the CCZ. Although London is relatively well served with air quality monitoring stations, our study was restricted by the availability of only a few monitoring sites within the CCZ, and only one of those was at a roadside location. The results derived from this single roadside site are not likely to be an adequate basis for evaluating this complex urban traffic management scheme. Our primary approach to assessing the impact of the CCS was to analyze the changes in geometric mean pollutant concentrations in the 2 years before and 2 years after the CCS was introduced and to compare changes at monitoring stations within the CCZ with those in a distant control area (8 km from the CCZ center) unlikely to be influenced by the CCS. We saw this as the most robust analytical approach with which to examine the CCS Study Database, but in the fourth part of the project we did consider three other approaches: ethane as an indicator of pollution dispersion; the cumulative sum (CUSUM) statistical technique; and bivariate polar plots for local emissions. All three were subsequently judged as requiring further development outside of the scope of this study. However, despite their investigative nature, each technique provided useful information supporting the main analyses. The first method used ethane as a dispersion indicator to remove the inherent variability in air pollutant concentrations caused by changes in meteorology and atmospheric dispersion. The technique had the potential to ascertain more accurately the likely impacts of the CCS on London's air quality. Although this novel method appeared promising over short time periods, a number of concerns arose about whether the spatial and temporal variability of ethane over longer time periods would be representative of meteorologic conditions alone. The major strength of CUSUM, the second method, is that it can be used to identify the approximate timing of changes that may have been caused by the CCS. This ability is weakened, however, by the effects of serial correlation (the correlation of data among measurements in successive time intervals) within air pollution data that is caused by seasonality and long-term meteorologic trends. The secure interpretation of CUSUM requires that the technique be adapted to take proper account of the underlying correlation between measurements without the use of smoothing functions that would obscure a stepped change in concentrations. Although CUSUM was not able to provide a quantitative estimation of changes in pollution levels arising from the introduction of the CCS, the strong signals that were identified were considered in the context of other results from the study. The third method, bivariate polar plots, proved useful. The plots revealed important characteristics of the data from the only roadside monitoring site within the CCZ and highlighted the importance of considering prevailing weather conditions when positioning a roadside monitor. The technique would benefit from further development, however, in transforming the qualitative assessment of change into a quantitative assessment and including an estimate of uncertainty. Research is ongoing to develop this method in air-quality time-series studies. Overall, using a range of measurement and modeling approaches, we found evidence of small changes in air quality after introduction of the CCS. These include small decreases in PM10, NO, and CO. The possibility that some of these effects might reflect more general changes in London's air quality is suggested by the findings of somewhat similar changes in geometric means for weekends, when the CCS was not operating. However, since some evidence suggests that the CCS also had an impact on traffic volume on weekends, the CCS remains as one possible explanation for the observed pattern of changes in pollutant concentrations. In addition, the CCS was just one of a number of traffic and emission reduction schemes introduced in London over the 4-year study period; if the other measures had an impact in central London, they might partly explain our findings. Although not the aim of this study, it is important to consider how the trends we observed might be translated into health effects. For example, given that London already has NO2 concentrations in excess of the permitted limit value, we do not know what the effects of an increase in NO2 created by diesel-exhaust after-treatment for particles might mean for health. Further, although it is not likely that NO affects health, the decrease in NO concentrations is likely associated with an increase in ozone concentrations (a pollutant associated with health effects), as has been seen in recent years in London. These and other similar issues require further investigation. Although the CCS is a relatively simple traffic management scheme in the middle of a major urban environment, analyzing its possible impact on air quality was found to be far from straightforward. Using a range of modeling and monitoring approaches to address the impact of the scheme revealed that each technique has its own advantages and limitations. The placement of monitoring sites and the availably of traffic count data were also identified as key issues. The most compelling lesson we take away from this study is that such work is impossible to undertake without a coherent multi-disciplinary team of skilled researchers. In conclusion, our study suggests that the introduction of the CCS in 2003 was associated with small temporal changes in air pollutant concentrations in central London compared with outer areas. However, attributing the cause of these changes to the CCS alone is not appropriate because the scheme was introduced at a time when other traffic and emissions interventions, which might have had a more concentrated effect in central London, were also being implemented.

Alternate JournalRes Rep Health Eff Inst
PubMed ID21830496