Spatial correlation of automatic air quality monitoring at urban background sites: Implications for network design

Measurements of carbon monoxide, nitrogen oxides, PM10, sulphur dioxide and ozone from four UK National Automatic Urban Network Stations have been analysed for the year 1994. All of the sites are urban background, with two in the same city (Birmingham Centre and Birmingham East) and the other two also in cities in the centre of England, Leicester and Leeds, respectively 58 and 154 km from Birmingham. Expressed as cumulative frequency distributions and median concentrations, the pollution climates of the four sites are relatively similar, with differences between Birmingham Cathe and Birmingham East as large as those between Birmingham and the other cities. Sulphur dioxide, which does not have an appreciable road traffic source, behaves differently, especially in Leeds, which is adjacent to a number of major point sources. However, when monthly hourly maxima and exceedences of health-related threshold concentrations are examined, the sites behave differently and relationships between them are much poorer, even for the two sites within the same city. It is concluded that network design needs to compromise between two extreme views. On the one hand, very high site density would be necessary to reflect all local conditions and provide adequate public information. On the other hand, concentrations are so spatially variable that the view may be taken that it is impossible to provide sufficient stations to monitor adequately, and that very few stations are all that is needed to give a general view of the pollution climate. Provided an adequate safety margin is built into ambient air quality standards, a high network density is not essential due to the generally strong correlation of measurements at adjacent sites.