Improving Air Quality in the United Kingdom

DOUGLAS R. MIDDLETON

1 Introduction

The concentration of pollutants in urban areas from sources near the ground has become of increasing concern in the UK, particularly since the London pollution episode of December 1991. During this episode from 11 to 15 December 1991, NO2 was unusually high, with values from 350 to 400ppb recorded at several sites and reaching 423 ppb at Bridge Place. These values were well above the standard which recommended a maximum hourly average concentration for NO2 of 150ppb. A study by South East Institute of Public Health mapped contours of air pollution measurements in the form of annual average concentrations of nitrogen dioxide (NO2). During 1995, these contours in much of London were above the NO2 Standard that appears in the UK National Air Quality Strategy. This Standard is 21 ppb for an annual mean. Since in London some 76% of the emissions of oxides of nitrogen are associated with road transport, measures for improving air quality will have to address transport planning. Looking forward, the report shows a projection of NO2 contours in London for the year 2000. It suggests that by that date in central London, NO2 will still be likely to exceed 40 µg m-3 (i.e. 21 ppb).

The Environment Act 1995 has increased the powers for local authorities to manage air quality and to consult with a wide range of bodies. The decisions to be taken in managing air quality will cross traditional boundaries. It will be necessary, for example, for Environmental Health Officers and environmental scientists in local government to liaise quite closely with traffic planners and highway engineers, as well as with managers of local industry and the Environment Agency inspectors, in order to assemble the emissions databases upon which the modelling of air quality for reviews and assessments will rely. The National Strategy for Air Quality sets out Government policy with regard to improving ambient air quality in the UK. It looks to the year 2005 and is relevant to both statutory requirements and to further voluntary action.

The Environment Act 1995 requires Local Authorities to review past and assess future air quality within their areas of jurisdiction. It also requires them to identify areas where levels of pollutants are high and, if necessary, designate them to be local air quality management areas. Such areas will be defined using the Standards and Objectives which appear in Table 3.1 of the Strategy. Management areas might be based upon administrative or other boundaries, but will need to contain locales where air quality will be high to the year 2005. Computer models provide not only a means of forecasting air pollution events, but also ways to investigate the contribution of actual or planned pollution sources. Similarly, monitoring data can be used as a means of projecting forward in order to assess the extent of control that is needed. Monitoring data are costly to acquire, and do not easily lend themselves to consideration of future controls on individual sources. There will therefore be more use of emissions inventories and modelling to complement the monitoring of pollutants.

We have here an example of how the new moves to improve air quality through local management are providing a stimulus to scientific research. As well as encouraging the development of simple assessment techniques, it has led to the need to develop simple monitoring methods for pollutants such as buta-1,3-diene, along the lines of the diffusion tubes for benzene. There is also work now in progress by a number of local authority groupings funded by the Department of the Environment (DOE) to validate dispersion models against each other and against measurements. Finally, we shall see below that the UK National Air Quality Strategy presents Standards and Objectives that will serve as a yardstick against which to judge air quality improvements.

2 Pollution in Street Canyons

Much monitoring has focused on the background level of airborne substances such as Pb, NOx, CO, O3, particles and organic compounds. However, it is at kerbside locations where the general public may suffer exposures to the highest concentrations of pollutants. This is particularly true in a street canyon, a relatively narrow street between buildings that line up continuously along both sides. The combination of large vehicle emissions and reduced dispersion in these circumstances can lead to high levels of pollution. A well-trafficked street canyon therefore represents an important facet of air quality management.

Recognizing that some authorities may not have the resources to run elaborate models, but will need a quantitative method, Buckland and Middleton have produced a simple method. The result is AEOLIUS, a selection of nomograms and charts that has been devised along similar lines to volume 11 of the Design Manual for Roads and Bridges. AEOLIUS is designed with one purpose in mind: to estimate the likely maximum concentrations from traffic in urban street canyons. It does not include the additional background concentrations from sources outside the street; they should be added by the user as necessary. AEOLIUS and other models are being tested by some local authorities during the trials cited earlier.

Only a brief outline of canyon dispersion principles is made here. A fuller description of canyon models appears in Buckland and the papers cited therein. When the wind blows across a street canyon a vortex is typically generated, with the wind flow at street level opposite to that above roof level. A consequence is lower concentrations of pollutants on the windward side of the street compared with those of the leeward. The windward side is here defined as the side the roof wind blows to whilst the leeward side is the side the roof wind blows from. The quantity of pollutant that a monitor directly receives from vehicle emissions is calculated using a simple Gaussian model. The contribution from air recirculated by the vortex is calculated using a simple box model. The principle is that the inflow rate of pollutant into the volume of recirculated air is equal to the outflow rate and that the pollutants are well mixed inside this volume.

The canyon concentration is proportional to the total emission rate Q from all vehicles, which will reflect changes in the vehicle fleet emission factors. For N vehicles an hour (with all vehicle types combined) and a combined emission factor of q grams km-1 vehicle-1, the total emission rate Q µg m-1 s-1 is given by the equation: Q = Nq/3.6, which converts grams to micrograms, km-1 to m-1 and vehicles hr-1 to vehicles s-1. This equation means that in order to improve air quality by reducing motor vehicle emissions Q, it is necessary to reduce the number of vehicles N and their average mass emission factor q. When Q is used in a dispersion model to calculate air pollutant concentration, it is multiplied by road length or distance travelled S; to improve air quality this also indicates reductions in the distance S. Traffic management and public transport can be regarded as managing q, N and the distance travelled S, whilst changes in technology such as catalysts or particle traps on new vehicles, and the maintenance of existing vehicles, seek to reduce q. Other developments such as cleaner diesel or unleaded petrol also serve to reduce q (for particulates or lead, respectively). Similar principles apply to fast moving traffic on open roads, and to idling engines in congestion. When vehicles are first started, and the catalyst is cold, the value of q may be much larger than for normal driving. Some published values for q are shown in Table 1; a busy traffic flow N might be 2000 vehicles per hour. Finally, it can be seen from the Strategy that even as average q is reduced in coming years, growths in N and S beyond the year 2010 are projected to outweigh the benefits of catalysts. The Strategy, p. 46, therefore lists the principles for improving air quality, which we summarize here:

• improved technology

• tighter fleet management

• environmentally responsible use of vehicles

• policies and planning to reduce reliance on cars.

3 Motor Vehicle Contribution

The importance of motor vehicles to the urban air quality debate is manifest in figures published by the newly completed West Midlands Emissions Inventory study. This survey of all major sources of air pollutants in the seven local authorities was funded by the DOE to provide the first inventory in the UK at a very detailed local scale. Hutchinson and Clewley collected data on transport (road, rail, air), domestic and industrial, Part B (i.e. those under local authority control), and Part A (i.e. under Environment Agency control), processes. Emissions of sulfur dioxide, oxides of nitrogen, carbon monoxide, carbon dioxide, non-methane volatile organic compounds, benzene, buta- 1,3-diene and particles were mapped on 1 km squares. The database contains approximately 9000 road links (derived from the data used to run the traffic model for the West Midlands) with their start and end co-ordinates, peak hour traffic flows and vehicle speeds. They concluded that:

'The single most significant source of atmospheric pollutants in the West Midlands is road traffic. In the case of carbon monoxide, benzene and buta-1,3-diene, road traffic accounts for over 96% of emissions. ... Road traffic also accounts for 85% of emissions of oxides of nitrogen and 75% of black smoke, but only 16% of sulfur dioxide.'

It is therefore clear that improvements in air quality will be dependent upon measures to reduce the total rate of emissions (SNq) from the motor vehicle fleet.

The composition of the vehicle fleet in terms of catalysts and type of fuel will change. To reflect this, revised emission factors (Table 2) provided by NETCEN AEA Technology for the expected national fleet in future years were also used to estimate the likely fractional changes in q (Table 3). The decrease in revised emission factor for each pollutant from 1996 to 2005 is in Table 3 where the values shown are normalized to the 1996 estimates of revised emissions, e.g. for a CO calculation in the year 2001, the emissions must be multiplied by 0.636. In the absence of detailed local traffic analyses in different towns, Table 3 should provide reasonable estimates of future trends in vehicle emissions. Similar information on emission factors and their future trends appears in the Design Manual for Roads and Bridges, and in the Strategy. Finally, whilst on the subject of models, Middleton reviewed physically based models for use in air quality management, for local or distant plumes.

4 Future Air Quality Objectives

In order to improve air quality, the targets at which to aim must be made visible. The Strategy gives Standards for priority pollutants and possible Objectives for achieving them. Table 4 is derived from the Strategy and EPAQS reports. Compliance or otherwise with the proposed values will be assessed for the year 2005. Breaching of these may point to a detailed assessment being necessary and the likely declaration of an Air Quality Management Area. This will in turn invoke the need for a formal statement or Action Plan; in this the local authority will set out its programme of local action, including the pollution controls as needed to achieve compliance with the Objectives. Such plans may estimate the amount of future emission control that is indicated, and imply constraints on planning.

To identify pollution 'hot spots', air quality management areas, calculations are likely to be required at many receptors so that mapping can be carried out. Table 3.1 of the Strategy has 15 minute mean, 1 hour mean, running 8 hour mean, running 24 hour mean, annual mean and running annual mean. The list also uses percentiles of some of these quantities.

Some quantities are to be used as their means for testing likely compliance by 2005 AD, i.e. for benzene, buta-1,3-diene, carbon monoxide, lead and nitrogen dioxide. Others involve their percentiles (on percentiles, see Spiegel; on log-normal and Weibull distributions to describe air quality data, see Seinfeld). It is then their concentrations at the various percentiles that are tested for compliance. They include 97th, 99th and 99.9th percentiles, which need additional processing of model results to generate the cumulative frequency distribution and derive the relevant percentile, i.e. ozone (not a local authority modelling task, cf. the DOE), particles as PM10 and SO2. After the calculations, running means and percentiles can be evaluated (using spreadsheets, or routines within some models). Surrogate statistics might also be considered, such as appear in the EPAQS standard for the maximum 15 minute means of SO2versus the hourly means. In general, the modelling will need to calculate hourly averages as sequential means and then apply the relevant running means. For mapping air quality and the publication of results, annual averages are often convenient, such as in SEIPH. They give a clear overview of the variations in air quality across an area. Maps of the Standards and Objectives in the Strategy may need special treatment, as outlined below.

5 Mapping an Air Quality Management Area

The Environment Act 1995 seeks to establish improvements in air quality through the identification of air quality management areas. At the time of writing, the manner in which such areas will be defined has yet to be announced; it is likely to be part of the regulations or guidance. Nevertheless, it is important to analyse ways in which such areas might be mapped. Future planning by local authorities may be strongly influenced by the projected extent of such areas, and public decision making in this area will need to be transparent and accountable. The following suggestions give an idea of the special nature of this mapping of regions that might exceed air quality Standards or Objectives.

Mapping Principles

In pollution modelling the phrase 'long term' is often used to convey the notion that the average concentration is to be obtained for some very long time period, such as one year or even 10 years. Annual average concentration is a typical quantity of this sort and would average some 8760 hourly values in a year. For reliable statistics, a 10 year run which requires 87 600 iterations might model the plumes from all sources, and generate a time series of results that would then be averaged. To map the likely pollution 'hot spots' using the annual average concentration, each hourly run would have to be repeated at each receptor position over the area to be mapped. Where the criterion for compliance to identify a problem area is based upon a concentration at some percentile, then instead of the average being found at each receptor, the results must be used to obtain the percentile at each receptor. The mapped 'hot spot' is the area enclosed within a contour that follows the positions where the percentile either exceeds or is less than the standard. For NO2 a mapped area may enclose receptor points with maximum values greater than the standard 150 ppb (Table 4) and mark these as exceeding the Objective. Receptors with maximum values less than 150ppb would be judged as complying with the Objective. Alternatively, there is scope in the Strategy to map an area where annual average concentration of NO2 may exceed 21 ppb.

On a map to show exceedance of the Objective for Fine Particles as PM10, the exceedance area would be the region where the receptors have 99th percentile cP > 50 µg m-3. In the case of Particles, the concentrations must be expressed in the form of running 24 hour means before the percentiles of the distribution are evaluated at each receptor.

Modelling

The input data must be formatted to suit the chosen model. Although broadly the same emissions inventory information are needed to model a given stack or line source, each computer code is likely to have its own sequence of data entry and slight variations on the parameters that are needed. For example, effluents from a chimney require stack height, stack diameter, exit momentum, exit buoyancy flux and emission rate of pollutant. Some models may require details of nearby buildings, whilst others will not.

The next step will be to prepare the meteorological data file. Either a series of short-term hourly observations of meteorology will be needed, or a frequency table sorted into joint categories by wind direction, stability class and wind speed. This choice of sequential versus statistical data can be decided using Table 4 for each pollutant in turn.