The answer, my friend

The vast majority of UK air pollution episodes in recent years have been of two main types. In the first, traffic-generated pollutants are trapped close to the ground in stagnant winter conditions, leading to high urban concentrations of nitrogen oxides, carbon monoxide and particulate matter. The severe London episode of December 1991 was a classic case of this type. The second type is a phenomenon in which ozone, nitrogen dioxide and particles are produced in increased amounts through atmospheric photochemical reactions and build up in summer anticyclonic conditions, particularly in association with air massess reaching the UK from mainland Europe. Occasionally, however, episodes of exceptional pollution levels occur which are not associated with either of these patterns; concentrations of sulphur dioxide, nitrogen oxides and particulate matter become elevated in a way which is not associated with local ground level sources. The event of September 2 1998, which affected the Midlands and South Yorkshire, was such an example.

Atmospheric mechanism

Industrial processes release waste gases into the atmosphere through chimneys with a height normally sufficient to disperse the pollutants. Under certain weather conditions, however (see below), the pollutants can fail to disperse and so accumulate, trapped in the local atmosphere producing high ground-level concentrations that exceed air quality standards.

Because the DETR monitoring station network is limited, it cannot always track individual air pollution episodes and supply detailed information on them for analysis. The record of complaints and the results from sulphur dioxide monitoring stations indicate the progress of this air pollution episode across the Midlands and South Yorkshire. However, these fail to identify the sources of the pollution or confirm the atmospheric mechanism that caused the episode.

Numerical modelling of the plumes from the likely sources, i.e. those roughly upwind, was carried out to gain some insight into the dispersion process, help determine source culpability and to predict the likely concentrations in areas where no monitoring stations were actually located. This process was carried out using NAME (Nuclear Accident ModEl), initially developed by the Met. Office to model the dispersal of radionuclides in the atmosphere following a nuclear accident, but which is equally suited to modelling the everyday release of non-radioactive pollutants by industrial processes. The actual emission of pollution from a source is represented using large numbers of ‘pollution particles’ that are released into a ‘model atmosphere’. These particles represent a quantity or amount of pollution within a small volume of air (and must not be confused with particulate matter pollution (PM₁₀)). The model atmosphere, represented by the winds and other meteorology, is obtained from the Met. Office’s operational numerical weather prediction model, known as the Unified Model (UM). Given an initial meteorological condition, the model predicts all of the meteorology over the whole globe and over the time period of interest.

Release data

In order to numerically model the pollution episode, however, release data for the period leading up to and through the episode is required from those sources identified as being potential contributors. Sources include those which may be regulated by the Environment Agency under IPC, or by local authority under LAAPC or the Clean Air Act 1993, or which do not come under direct regulation (such as traffic or residential).

Clearly, power generation, refineries, iron and steel and other major combustion processes (such as cement works and large industrial heating systems) are significant sources, accounting for nearly 90% of the annual sulphur dioxide emitted in the UK. Other sources of sulphur dioxide are much smaller and geographically dispersed. Whilst these sources will have contributed to the air pollution episode, collecting the data from the major sources was believed sufficient for modelling purposes by the Agency.

(Consideration was given to the potential presence of a LAAPC process or smaller heating system located close to and upwind of a monitoring station. If this were the case, then the recorded sulphur dioxide peak may have been disproportionately influenced by a relatively small source. No such LAAPC or heating systems could be identified, however. The City of Nottingham Environmental Services informed the Agency that the nearest LAAPC process authorised to release sulphur dioxide was some 5km north of the Nottingham Centre monitoring station. At such a distance, and direction, it is improbable that this source would have contributed significantly, if at all, to the peak. Similarly, the local authorities in Nottinghamshire as a whole have prepared an inventory of releases, which includes small heating systems. A search of this inventory did not reveal any sulphur dioxide sources that were close to and upwind of the monitoring site.)

The release data available on the Environment Agency’s Pollution Inventory takes the form of annual totals. In order to carry out a detailed study for 2 September, hourly emission rates were required. In selecting possible processes that may have contributed to the episode, consideration was given to the size of the process and the distance upwind (towards the north-east). In practice, data was collected from processes releasing sulphur dioxide within a box extending from Birmingham in the south-west to Malton in the north-east – 16 operators were asked for the following information:

• for each stack, its height, diameter and number of flues, flue diameter and National Grid Reference;
• for each stack, the average release rate of SO₂, NO₂, NO and particulates, temperature and release velocity (or volumetric flow for each hour of the day for 2 September);
• a commentary of plant performance, configuration or anything unusual on that day.

None of the operators indicated any unusual operating conditions or informed the Agency of a breach of any IPC Authorisation condition on this day. (There is no evidence that any of the sources emitted more sulphur dioxide than reported. The magnitude of the difference betwen measured and predicted values suggests that no individual process could realistically emit the quantity of sulphur dioxide required to cause the peak excess.) The release data supplied was collated and passed to the Met. Office for NAME modelling. NAME displays the information on what are known as ‘plume plots’, which show particle position and age.

Although such modelling conclusions cannot be confirmed, the Agency currently considers, on the balance of probabilities, that the peak at Nottingham Centre can be explained by the division of culpability given in the above table, and that the contribution from all other low-level sources is so small as to be insignificant.

Overall, the Agency concluded that the widespread nature of the incident indicated a regional-scale air pollution episode that no single regulated process could have caused. Rather, the incident was driven by the weather of 1 and 2 September 1998. A period of low wind speed allowed pollutants from a range of industrial sources to accumulate in the atmosphere to a high concentration before dispersal.