Elsevier

Atmospheric Environment

Volume 37, Issue 35, November 2003, Pages 4927-4933
Atmospheric Environment

A pragmatic mass closure model for airborne particulate matter at urban background and roadside sites

https://doi.org/10.1016/j.atmosenv.2003.08.025Get rights and content

Abstract

Twenty-four hour samples of airborne PM10 particulate matter have been collected as coarse and fine fractions using automated dichotomous samplers at four paired roadside and urban background locations in London and Birmingham, UK. The samples have been analysed for sulphate, nitrate, chloride, organic carbon, elemental carbon, iron and calcium and the data have been used to construct a simple model of aerosol chemistry. It is assumed initially that the major components are ammonium sulphate, ammonium nitrate, sodium chloride, elemental carbon, organic carbon and mineral dusts (for which iron and calcium are tracers).

This leaves a small proportion of mass unaccounted for, which we attribute to strongly bound water. Increasing the ammonium sulphate and ammonium nitrate content by 29% allows 100% of mass to be accounted for with a high percentage of variance in 24 h mass concentrations explained.

Introduction

Airborne particulate matter continues to give rise to concern as a result of its, now well established, adverse effects on human health. Whilst some doubt still exists as to which particle metric (e.g. number, surface area or mass) or size fraction is most closely related to the adverse health outcomes (Harrison and Yin, 2000), current air quality standards in both Europe and North America are expressed in terms of mass concentration and therefore control policies are directed at reducing the mass concentration of particles in the atmosphere.

The formulation of cost-effective abatement strategies for airborne particulate matter is crucially dependent on knowledge of the contributions of individual source categories to airborne concentrations. Such information may be gained from deterministic models, but such models suffer from many weaknesses including poor information on the magnitude of certain particle source categories (e.g. resuspension). The alternative is to use receptor models and there has been some success in the apportionment of both particle mass and specific components of airborne particles to specific source categories using models based on multivariate statistics (Thurston and Spengler, 1985; Chan et al., 1999; Harrison et al., 1996). However, much can be learnt from simpler models based purely on analysis of major chemical components; e.g. in the UK, the Airborne Particles Expert Group (APEG, 1999) used a three component model of atmospheric particles to construct projections of the influence of abatement policies on future airborne concentrations of PM10. The components in the model were primary combustion particles (mainly from road traffic), secondary sulphates and nitrates and coarse particles, including sodium chloride and resuspended soils and road dusts. While such a model is clearly a major over-simplification, it does bring considerable clarification to what is otherwise a very complex problem. Taking the three component model as a starting point, Turnbull and Harrison (2000) devised a slightly more sophisticated multiple regression model using analytical data for sulphate, nitrate, chloride and black smoke (as a marker for combustion aerosol), which accounted for a large proportion of the variance in the mass data.

The aim of the currently described work is to extend and improve those simple models. To date, aerosol chemistry and mass closure models have typically analysed a large proportion of the elements of the periodic table in an attempt to gain as complete a knowledge as possible of the chemical composition (Young et al., 1994; Chan et al., 1997; Kim et al., 2000; Andrews et al., 2000; Chow et al., 2002); however such an approach is too complex to be applicable on a routine basis and despite very extensive chemical analyses has generally failed to account for the full gravimetrically determined mass of particles, an outcome which has been attributed to the presence of unanalysed strongly bound water. In this work we have, therefore, deliberately restricted ourselves to the analysis of a small number of components which can be used as tracers of major aerosol constituents, in order to provide a simple but effective model. In all, some seven chemical components are analysed and the mass reconstructed in a semi-deterministic manner, rather than purely by regression analysis. The method is therefore a hybrid between the comprehensive chemical analysis method on the one hand, and the simpler statistical procedures on the other.

Section snippets

Air sampling

Sampling was conducted at four pairs of sites, three in London and one in Birmingham. Each site pair is known by the area in which it is located, i.e. High Holborn, Elephant and Castle, and Park Lane in London, and Selly Oak in Birmingham. Each site pair comprises a roadside location close to a heavily trafficked highway and a background location within 1 km of the roadside location but at least 50 m from any busy highway. The difference between the roadside and background locations may be taken

Results and discussion

As explained in the introduction, the aim of this work was to provide a much simplified protocol for chemical analysis of airborne particles which nonetheless took account of the major aerosol components. Table 1 shows the species which were analysed, and the range of concentrations encountered. The reasons for selecting these analytes, and the adjustment factors to the concentrations are detailed below.

  • Sulphate: This is one of the major secondary components of airborne particles and at inland

Conclusions

Any chemical model of atmospheric aerosol risks being incomplete even when a very wide range of constituents is analysed, and there are problems in converting such analyses to mass since the elements oxygen and hydrogen are not analysed and have to be inferred. Additionally a chemical analysis of water, as opposed to indirect estimation of water content by other means is very difficult and therefore scarcely ever conducted. Therefore a comprehensive analysis, including a wide range of trace

Acknowledgements

The authors are grateful to the UK Department of Transport for financial support of the work through the TRAMAQ programme, and to Roger Barrowcliffe and Alex Newton of ERM who collaborated in organisational aspects of the work and the air sampling.

References (23)

Cited by (123)

  • Carbon isotope compositions and TC/OC/EC levels in atmospheric PM<inf>10</inf> from Lower Silesia (SW Poland): Spatial variations, seasonality, sources and implications

    2020, Atmospheric Pollution Research
    Citation Excerpt :

    In Poland, the distribution of pollution sources is particularly impacted by the country coal-based economy, as it represents the main combustible for households heating, with an often low-quality coal and even the use of biomass waste (Pospolita and Cholewinski, 2018). The main PM10 compounds in cities and urban areas are carbon compounds, representing up to 80% of the total mass (Chow et al., 1993; Gray et al., 1986; Harrison et al., 2003; Park et al., 2001; Rogge et al., 1993). The total carbon (TC) is constituted of: (i) organic carbon (OC), (ii) elemental carbon (EC) and (iii) carbonate carbon (CC) (Schauer et al., 1996).

View all citing articles on Scopus
View full text