4 Discussion

AEAT-3133

4.1 Nitrogen dioxide and oxides of nitrogen
Maps of annual mean concentration of NO₂ and NO_x for 1994 have been published by Stedman et al (1997b). Land cover information and NO_x emissions from major roads were used as surrogates for local emissions in this earlier work. In the maps presented here we have used estimates of low-level emissions to calculate ambient concentrations. Urban background concentrations of these pollutants are determined to a large extent by NO_x emissions from road transport. The good correlation between annual mean urban background NO₂ and NO_x concentrations and local emissions estimates indicates that the spatial variation in emissions at the scale used for the mapping is well represented by the NAEI methods.

The reliability and applicability of the NO₂ map to background locations is further examined in Figure 11 by comparison of the 1 km x 1 km grid square estimated annual mean concentrations from the map with the annual mean NO₂ concentrations for 1996 from 'urban background' (defined as being more than 50 m from any busy road and typically in a residential area) sites within the UK Nitrogen Dioxide Survey (Stevenson and Bush , 1997). The mean of the measured concentrations was 14.4 ppb; the mean of the estimated concentrations was 13.7 ppb; r = 0.63, n = 555. While this graph shows some scatter, there is no evidence of as large systematic error in the estimates of annual mean background NO₂ concentration. The estimated NO₂ concentrations in inner London are rather higher than indicated by the diffusion tube measurements. A comparison of the estimated values with automatic monitoring sites in inner London (Figure 1c) also indicates that the map overestimates concentrations at some of these sites, but not to the extent suggested by the diffusion tube measurements. It is likely therefore that concentrations in this area might be overestimated by the map, perhaps due to high emissions estimates, and/or that the diffusion tube measurements may under represent concentrations relative to automatic measurements.

Figure 12 shows a comparison of estimated annual mean NO_x concentrations and measured values for urban background and roadside and kerbside monitoring sites in London and the South East. Agreement between estimated and measured values is very good for background sites but is extremely poor for roadside and kerbside sites. This shows the influence of nearby traffic emission on NO_x concentrations in these locations. Roadside or kerbside NO_x concentrations can, in principle be calculated by adding a background value to a kerbside contribution derived from traffic activity information. This estimate of kerbside contribution could be calculated using a method such as that presented in the Design Manual for Roads and Bridges, or a simplified model based on the analysis of kerbside monitoring data and traffic activity.


4.2 Sulphur Dioxide
In our previous report we presented a 10 x 10 km resolution map of estimated SO₂ concentrations for 1994 and noted that it underestimated concentration in coal use areas, particularly in Belfast. Estimates of domestic emissions of SO₂ in Northern Ireland within the NAEI have since been revised and the estimated concentrations for automatic monitoring sites in Belfast are now much nearer to the measured values. The majority of the automatic monitoring sites used to 'calibrate' the SO₂ map are in either city centre or rural locations. The map therefore provides a reasonably good estimate of concentrations in these areas. The accuracy of the estimates of concentration in smaller urban or suburban areas is dependent on fuel use and the map may still underestimate concentrations in some areas. Figure 13 shows a comparison of estimated concentrations with measurements from Basic Urban Network sites within the UK Smoke and Sulphur Dioxide Survey. It is clear that the map underestimates concentrations at many monitoring sites. The value of the ratio of measured concentration divided by estimated concentration has been calculated for each of these sites and the average value of this ratio has been calculated for different site environments. The agreement is slightly better, on average, at sites in smoke control areas (ratio = 1.8 (all sites); 1.7 (sites in smoke control areas); 1.9 (sites not in smoke control areas)). Agreement was best at sites in city and town centre locations (ratio = 1.6) and worst at sites with high (ratio = 1.9) or medium (ratio = 2.1) density housing and industrial areas (ratio = 2.3).

There may be a number of reasons for the poor performance of the map in comparison with Basic Urban Network measurements:

• inaccuracies in the measurements;
  
• the influence of very local sources (within a few hundred metres, domestic or industrial) on concentrations at the monitoring sites;
  
• inaccuracies in the way that domestic and industrial SO₂ emissions are spatially distributed within the emissions inventory.
  

4.3 Benzene and 1,3-butadiene
The variation of emissions amount with vehicle speed is very different for VOC and NO_x. Maximum VOC emissions (in terms of g km^-1) are produced by slow moving vehicles, such as those on congested urban roads. Maximum NO_x emissions are produced by fast moving vehicles, such as free flowing traffic travelling on motorways.

Previous maps of estimated benzene and 1,3-butadiene concentrations were derived from maps of NO_x and made use of the relationship between measured ambient NO_x and benzene and 1,3-butadiene concentrations at a site where measurements are co-located. These maps therefore probably overestimated benzene and 1,3-butadiene concentrations in the vicinity of fast roads. The current benzene and 1,3-butadiene maps (Figures 4 and 5) were derived directly from VOC emission inventories and show lower concentrations at these locations. The agreement between estimated and measured concentrations (Table 3) is similar to that for our previous maps because most of the monitoring sites are in city locations where our estimates have not changed much. Estimates of concentrations in the vicinity of motorways are probably more realistic in our current maps but it is not possible to validate these estimates without additional monitoring in these areas.

The only site for which concentrations are noticeably underestimated by the map is Southampton Centre, where a nearby busy road seems to lead to higher concentrations than are predicted by the map. This is particularly noticeable for benzene and 1,3-butadiene but Southampton Centre is one of the sites for which the concentration of several of the pollutants are underestimated by the maps.


4.4 Carbon Monoxide
Figure 6b shows that the correlation between measured CO concentrations and estimated local emissions is poorer than for NO_x. It is likely that there is a larger small-scale spatial variability in traffic CO emissions than for NO_x. CO emissions per unit distance increase markedly at low speeds relative to emissions of NO_x (Figure 14). The concentrations of CO recorded at monitoring sites are to some extent dependent on emissions in the immediate locality (<< 1 km) and concentrations would therefore be expected to increase where there is significant local traffic congestion. The poorer correlation between point measurements and map values for CO than for NO_x does not necessarily mean that the map values are worse estimates of the grid square average values.

The map presented as Figure 6 should provide more realistic estimates of CO concentrations in than our previous map of estimated concentrations. Our previous map was based on an average relationship of measured ambient CO and NO_x concentrations but we noted in the report (Stedman, et al, 1997b) that this relationship between CO and NO_x was rather uncertain due to the wide range in CO/NO_x ratio observed at monitoring sites.

4.5 Particles (PM₁₀)
PM₁₀ is one of the most difficult pollutants to map due to the range of sources of both primary and secondary particles that contribute to ambient concentrations. The sources include:

• primary particles from vehicles;
  
• primary particles from stationary combustion sources;
  
• primary particles from non-combustion sources such as quarrying, demolition and wind blown dust;
  
• secondary particles.
  

Two alternative methods of estimating the secondary particle contribution were discussed in out previous report (Stedman et al, 1997b). Secondary particle concentrations were estimated from either photochemical ozone or rural particulate sulphate measurements. The secondary particle contribution to annual mean background PM₁₀ concentration in the map presented in Figure 7 was derived from particulate sulphate measurements.

Primary particle concentrations (measured PM₁₀ - estimated secondary particle concentration) is plotted against local vehicle emissions estimates from the NAEI in Figure 7b. There is a reasonably consistent relationship between these two parameters for all sites except London Brent and Kensington and Chelsea. The intercept concentration of primary particles at zero vehicle emissions represents primary particles from stationary and other sources. A spatially dissagregated emissions inventory is not currently available within the NAEI for these sources. Measurements of PM₁₀ concentrations in rural areas provide a possible method for estimating the spatial variation in concentrations derived from these stationary combustion and non-combustion sources. Measurements of PM₁₀ concentrations are now available from a limited number of rural sites. Available data is listed in Table 4 along with estimated secondary particle and vehicle derived particle contributions.

Table 4. Rural PM₁₀ concentration measurements and the estimated contributions from different source types (mgm^-3).

id	Site	Network	Period for which mean was calculated	Measured PM10	Secondary PM10	Vehicle PM10	Other PM10
15	Lough Navar	RMN	October 1996 - September 1997	9.8	5.9	0.0	3.9
25	Rochester	RMN	Annual 1996	22.0	11.3	0.1	10.7
88	Hall Farm	JEP	Annual 1995	22.9	11.2	0.9	10.8
94	Cliffe	JEP	Annual 1995	21.0	9.6	0.1	11.2
139	Bottesford	JEP	Annual 1996	21.9	10.3	0.1	11.5
162	Ratcliffe	JEP	Annual 1996	22.8	9.8	0.1	13.0
187	Narberth	RMN	March - September 1997	15.0	8.3	0.0	6.7

Note: Some 1997 data are provisional

This table indicates that there is a higher concentration of PM₁₀ from 'other' sources in England than in the west of Wales or Northern Ireland. In our previous map we assumed a constant value for this contribution across the entire country. In the map presented here we have assumed a spatially varying concentration for these sources; with a maximum value of 12 mgm^-3 in eastern England and a minimum value in the west of Northern Ireland of 6 mgm^-3. The value of this contribution was therefore calculated by multiplying the Ordinance Survey grid reference easting (in metres) by 0.00001, as indicated in Table 2.

PM₁₀ concentrations are likely to be more strongly influenced by domestic heating emissions in urban areas in Northern Ireland than in other UK cities. Concentrations of PM₁₀ in Northern Ireland have been calculated using estimates of SO₂ emissions as a surrogate for PM₁₀ emissions (see Table 2).


4.6 Lead
In most areas airborne lead concentrations are dominated by the contribution from vehicle emissions due to the use of leaded petrol. Annual mean concentrations are available for a total of ten background monitoring sites for 1996. Lead concentrations were also measured in several locations where specific industrial sources of lead emissions give rise to higher concentrations. It is not possible to map the contribution industrial lead sources to ambient mean lead concentrations using the simple empirical box modelling approach used here. The map presented here therefore represents the road transport derived background lead concentration, onto which the impact of individual industrial sources could be added in more detailed modelling studies.

Spatially dissagregated lead emissions estimates are not available from the NAEI so the map presented here was based on the relationship between lead concentrations and local NO_x emissions. NO_x emissions are taken here to be reasonably representative of emissions of lead from vehicles. An alternative approach would be to use estimated lead emissions or a different surrogate such as leaded petrol fuel use.


4.7 Ozone
The estimated ozone maps presented here in Figure 9 and Figure 10 have been derived using mapping methods which extend the work carried out jointly by NETCEN and the Institute of Terrestrial Ecology for publication in the 4th report of the Photochemical Oxidants Review Group (PORG, 1997). In contrast to the other pollutants in this report, the concentrations of ozone in urban areas are often lower than those in the surrounding rural areas. Land cover information at a 5 km x 5 km grid resolution was used by PORG as a surrogate for local NO_x emissions to estimate urban ozone concentrations from maps of rural ozone concentrations. Both 1 km x 1 km estimates of local NO_x emissions from the NAEI and 1 km x 1 km land cover information (Fuller et al, 1994) were investigated as alternative surrogate statistics from which to derive urban ozone concentrations for 1995 for inclusion in this report. Estimates of NO_x emissions were found to give the most reliable results for summer mean ozone concentration and land cover on a 1 km grid gave the best fit to the measurements for the number of days with concentrations greater than or equal to 50 ppb.

Maps of rural ozone concentration for 1995 were interpolated from measurements at RMN sites. Summer mean ozone concentrations vary with altitude (PORG, 1997), so a map cannot be interpolated directly from a network of rural monitoring sites at a range of elevations. A map of concentrations during the 'well mixed' part of the day (12-18 GMT) can, however, be interpolated with reasonable confidence because concentrations during this part of the day are not influenced by altitude. The difference between concentrations during this well mixed period and the mean over the whole day (DO3) has been found to be dependent on altitude:

DO3 = 3.4 + 7.7.exp(-4.2x10^-3.altitude)

where the average altitude is in m for the 1 km x 1 km grid square including the site location (PORG, 1997). A map of rural summer mean ozone concentrations can therefore be calculated from the 'well mixed map' and an altitude map of the UK using the above equation.

The number of days with 8-hour mean ozone concentrations greater than or equal to 50 ppb does not vary with altitude so a map of this statistic can be interpolated directly from measurements at rural sites.

For all of the pollutant mapped in this report except ozone, local urban emissions tend to increase the concentration of pollutants in urban background locations. Conversely, local emissions of NO_x tend to decrease the ambient ozone concentration and the strength of this effect has been described by PORG in terms of an 'urban influence', UI, of these emissions:

UI = ((rural ozone concentration) - (measured urban ozone concentration)) / (rural ozone concentration).

The following relationships between UI and surrogate statistics, illustrated in Figure 9b and Figure 10b, have been used to derive the ozone maps for the summer of 1995.

Summer mean ozone concentration (ppb):

UI = 0.1554 x [area + major road NO_x emissions, kTonnes per 25 km² per year].

Number of days with 8-hour mean ozone concentrations greater than or equal to 50 ppb:

UI = 0.00629 x [the proportion of land cover that is determined as urban or suburban].

INDEX