Annual average sulphur dioxide concentration maps derived by dispersion modelling
| Authors | John Abbot |
| Keith Vincent |
Copyright AEA Technology plc
Annual average concentrations maps of sulphur dioxide for 1996 and 2005 throughout the United Kingdom have been prepared using the dispersion model ADMS-2 Version 2.2. These maps were prepared to support the review of the National Air Quality Strategy (NAQS). The study is the first part of a two part modelling study series which considers how sulphur dioxide emissions from both Part A emission processes and area sources contribute to exceedance of the NAQS for sulphur dioxide. The second part of the study provides maps showing the extent of the exceedence of the National Air Quality objective for sulphur dioxide in England, Scotland, Wales and Northern Ireland for 1996 and 2005
Part A processes with emissions greater than 500 tonnes per annum, and area sources, have been included in the study. The model results have been calibrated to correspond to measured concentrations at rural monitoring sites. The model has then been used to prepare calibrated maps of sulphur dioxide concentrations. The map estimates have been validated by reference to measured concentrations from all available sulphur dioxide monitoring stations throughout the United Kingdom.
Predicted annual average concentrations in the new mapping procedure are not very different from those produced using a combination of a rural SO2 concentration map and empirical modelling of area source contributions. The key advantage of the current modelling work is that the contribution to ambient SO2 concentrations from different source sectors are treated independently. Annual average concentration for 2005 for a range of scenarios were also produced- the maps show concentrations in the more populated areas of the country substantially lower than the 1996 concentrations.
This report was prepared to support the review of the National Air Quality Strategy (DETR, 1999a). The aim of this study was to provide maps estimating the annual average concentration of sulphur dioxide in England, Scotland, Wales and Northern Ireland for 1996 and 2005. The study is the first part of a two part modelling study which considers how sulphur dioxide emissions from both Part A emission processes and area sources contribute to exceedance of the NAQS for sulphur dioxide. Where the objective is defined as 100 ppb measured as the 99.9 percentile of 15 minute averaging periods for a year to be achieved by 2005 (DOE, 1997). Annual mean concentration maps are further required for analysis of cost and benefits of the Air Quality Objectives (DETR, 1999a, DETR, 1999b and Stedman et al., 1999). The second part of the study provides maps showing the extent of the exceedence of the National Air Quality objective for sulphur dioxide in England, Scotland, Wales and Northern Ireland for 1996 and 2005 (Abbott and Vincent, 1999).
The mapping method involved dispersion modelling of emissions from both Environment Agency regulated Part A processes with an emission greater than 500 tonnes per annum and from area sources. Linear regression was then used to calibrate the predicted concentrations based on concentrations measured at automatic monitoring stations in rural locations. The resulting coefficients obtained were then used to predicted annual concentrations throughout the modelling domain. Concentrations fields in 2005 were derived for several emission scenarios.
The maps make use of:
The maps were calibrated by comparison with monitoring data from:
The maps were validated by comparison with monitoring data from:
All Part A processes within the 1996 National Atmospheric Emission Inventory with emissions greater than 500 tonnes per annum were included in the dispersion modelling study.
The following parameters were used to characterise the emissions from each of the modelled Part A processes:
Emissions data for 1996 were taken from the NAEI (Salway et al. 1999) based on reported emissions. A range of predictions were made for 2005 based on DTI estimates for alternative emission scenarios [DTI, 1995; Abbott et al, 1996; Abbott and Watterson, 1996]. These are shown in Table 1.
Point source emissions for 2005 were scaled from the 1996 estimates to ensure the total emissions corresponded to the estimates in Table 1.
Environment Agency regional offices were contacted and asked to provide details of the discharge conditions from Part A sources in each region. Much useful information was provided for many sources and was used in the modelling study. Other information was obtained from recent studies carried out at NETCEN. Where no information was supplied to us within the timescale of the project concerning discharge conditions, we have made our best engineering judgement to provide suitable estimates. Our engineering judgement has included consideration of the likely sulphur content and calorific values of the fuel used in order to estimate discharge velocities and the heat content of the discharging plume and use of HMIP's Technical Guidance Note D1 to estimate stack heights.
Table 1: Sector totals used for the emission scenarios in 1996 and 2005
|
|
1996 Emission (Mt) |
Energy Paper 65 (Mt) |
*CL with total upgrade and Sulphur Content of Liquid Fuel Directive (Mt) |
CH with total technical upgrade (Mt) |
CL with total technical upgrade (Mt) |
|
Power stations |
1.318 |
0.58 |
0.40 |
0.40 |
0.40 |
|
Transport |
0.083 |
0.09 |
0.08 |
0.08 |
0.09 |
|
Domestic |
0.069 |
0.06 |
0.06 |
0.06 |
0.06 |
|
Services |
0.059 |
0.07 |
0.06 |
0.06 |
0.07 |
|
Refineries |
0.122 |
0.13 |
0.07 |
0.09 |
0.12 |
|
Other industry+ agriculture |
0.377 |
0.54 |
0.35 |
0.40 |
0.48 |
|
Total |
2.028 |
1.47 |
1.02 |
1.09 |
1.22 |
Where:
* The CL with total upgrade and Sulphur Content of Liquid Fuel Directive scenario was used in the cost benefit analysis study (Stedman et al. 1999).
CL is defined as central GDP growth and low fuel prices
CH is defined as central GDP growth and high fuel prices
Many of the Part A processes emit sulphur dioxide through multiple stacks. We have ignored the possible enhancement of plume rise that may occur when plumes combine from stacks that are located near to each other. Where there are many stacks at the same site with different discharge conditions, for example at refineries or chemical works, we have, in some cases, grouped stacks of similar height and discharge characteristics together and assumed that the emission from each group of stacks may be represented by a single stack with characteristics of the stack with the largest emission or by a stack with an emission-weighted "average" stack.
2.2 AREA SOURCESEstimates of emissions from area sources were obtained from the NAEI for each 1 km square area. The emissions from each source were assumed to be distributed uniformly through an initial height of 10 m: i.e. each 1 km source was represented by an emitting volume 1 km
´ 1 km square and 10 m high. The estimate of 10 m is based on the height of a typical house and assumes that emissions will be entrained in the building wake. It is however recognised that this may lead to overestimates of ground level concentrations near small industrial operations, where the emissions are discharged through stacks typically 30 m tall and with significant plume buoyancy. A possible refinement of the procedure would allow industrial sources to be modelled separately from domestic sources.Emissions for 2005 for a range of alternative scenarios were estimated by scaling the 1996 estimated emissions to ensure that the sector total emissions corresponded to those shown in Table 1.
The dispersion model, ADMS-2 version 2.2 was used to calculate annual average concentrations.
A number of receptor areas were defined, which together covered the majority of the country. Each receptor area was 100 km square, extending out to 150 km to cover coastal areas where appropriate. All sources within the receptor area and in sources in the adjoining 100 km square areas were included when modelling the combined effect of Part A sources on the concentration in the receptor area. Concentrations were calculated on a regular 5 km grid throughout the receptor areas.
Meteorological conditions throughout the whole country were represented using statistically analysed data for Elmdon, 1986-95, obtained from the Meteorological Office. The statistical analysis gave the frequency of all combinations of the following:
A uniform surface roughness of 0.1 m, typical of agricultural areas was used for the whole country. A dry deposition velocity of 0.01 m/s was assumed. Wet deposition was not calculated.
3.2 AREA SOURCESThe annual average contribution from area sources was calculated on a 1 km receptor grid covering the country using the dispersion model ADMS-2 Version 2.2 . Each source was represented as a volume source 10 m high and 1 km square. Contributions at each receptor from sources at distances greater than 15 km in the north-south or east-west directions were ignored.
Meteorological conditions throughout the country were represented using statistically analysed data for Wyton, 1975-86. The statistical analysis gave the frequency of all combinations of 12 wind direction classes and 6 Pasquill stability classes and assigned an average wind speed to each combination.
A uniform surface roughness of 1 m, corresponding to typical urban areas was used for the whole country. Wet and dry deposition were ignored.
The concentrations of sulphur dioxide at urban monitoring stations are affected both by emissions from local sources and by larger sources some distance away. Furthermore, the top down methods used by the NAEI to disaggregate the emissions from domestic and small industrial sources are inaccurate at the local scale. It is therefore difficult to calibrate the area source model using sulphur dioxide monitoring data. However, NAEI emissions estimates for oxides of nitrogen, mainly from road transport sources, are considered much more reliable and allow calibration of the area source model.
The area source dispersion model was used to estimate annual average oxides of nitrogen concentrations at Automatic Urban Network sites across the country. An allowance was made for background contributions by adding 10
m g m-3 to the modelled concentrations. Figure 1 shows 1995 modelled and measured concentrations. It was concluded that the model provides a reasonable estimate of the contribution to annual average concentrations from area sources. In other words, annual average concentrations were predicted directly from the area source model and calibration was not required. 4.2 ANNUAL AVERAGE CONCENTRATIONSMonitoring data from the Rural SO2 monitoring network and the power generators Joint Environmental Programme were used to calibrate the model to ensure agreement between the 1996 map and measured data.
Linear regression analysis of modelled and measured concentrations at rural monitoring sites was carried out to establish the values of constants, A and B in:
Measured annual average = A + B
´ Modelled Part A + Modelled Area SourcesRegression analysis gave the following values:
A = 3
m g m-3B=1.1
The value of A is small and the value of B is close to unity so that the mapping model we have developed is providing results very close to the measurements. Figure 2 shows the data used to calibrate the model.
The residual concentrations were then calculated at each monitoring site:
Residual = Measured - Regression Model
These residual contributions are associated partly with errors in the model and partly with the contributions from more distant sources, not modelled in this study. They include, for example, contributions from emissions from sources on continental Europe.
The residual concentrations were interpolated across the country to provide a map of residuals using simple kriging. The final map was calculated from:
Mapped Value = A + B
´ Modelled Part A + Modelled Area Sources + ResidualThe inclusion of the factors A and B and the residual concentrations in the mapping procedure ensures that the map produced is based as closely as possible on measurements rather than on dispersion modelling alone.
It was anticipated that a reduction in emissions throughout Europe in the years to 2005 will result in a reduction in the residual concentrations. The HARM model version 11.3 was run for the years 1995 and 2005, with emissions over the EMEP grid area for 2005 scaled to match reference scenario country totals listed in IIASA's 5th Interim Report (Amman et al., 1998) and a UK national total of 1160 kt [Whyatt, 1998]. Estimates of the reduction in the residual concentration were made by scaling the 1996 residuals at the rural monitoring network sites in proportion to the reduction in the HARM model results between 1995 and 2005.
Modelled annual average concentrations are compared with the values observed at Automatic Urban Monitoring sites in Figure 3. At many of the sites the map provides a reasonable estimate of the annual average concentrations. However, there is a group of sites including, Middlesbrough, Manchester, Leicester, Liverpool and Edinburgh, where the map significantly overestimates the observed concentrations. In each case, the presence of a relatively large industrial source within the area inventory in the vicinity of the city has contributed significantly to the mapped concentration. Emissions from industrial sources at some locations may be overestimated in the emissions inventory. The simple area model may also overestimate concentrations close to small industrial sources by neglecting stack effects.
The model underestimates observed concentrations in Barnsley significantly. This may be because the NAEI disaggregation procedure underestimates the emissions from domestic sources in this area.
Mapped annual average concentrations are compared with values at Smoke and SO2 monitoring sites in Figure 4. Clearly there is no correlation. There are two possible explanations:
Map 1 shows the annual average concentration for 1996. The annual average concentrations in the new map presented in Map 2 are not very different from the SO2 map estimated by Stedman (1998) from a combination of a rural SO2 concentration map and empirical modelling of area source contributions.
The key advantage of the current modelling work over that carried out by Stedman (1998) is that the contribution to ambient SO2 concentrations from different source sectors are treated independently. This more detailed modelling approach enables predictions for 2005 to be calculated.
Map 3, Map 4, Map 5 and Map 6 show the estimated annual average concentration for 2005 for a range of scenarios. The maps show annual average concentrations in the more populated areas of the country substantially lower (approximately two thirds of) the 1996 concentrations.
Amann M., Bertok I., Cofala J., Gyarfas F., Heyes C, Klimont Z., Makowski M., Sch
ö pp and Syri S. (1998). Cost Effective Control of Acid and Ground-Level Ozone. Part A: Methodology and Databases. Fifth Interim Report to the European Commission, DG-XI. May 1998.JA. Abbott, P Coleman, M Cupit, C Johnson, K Wigley and A Markandya (1996). Sulphur Dioxide Ambient Air Quality Study-Final Report. AEA Technology AEA/WMES/20077002/002/ISSUE2, September 1996. Culham Science Centre, Culham, OX14 3ED, Oxfordshire.
Abbott and Watterson, 1996. Sulphur Dioxide Ambient Air Quality Study-Sulphur Content of Liquid Fuels. AEA/WMES/20077001/003/ISSUE1, December 1996. Culham Science Centre, Culham, OX14 3ED. Oxfordshire.
Abbott J.A. and Vincent K.J (1999). Dispersion Modelling of SO2 Concentrations in the United Kingdom for Comparison with the National Air Quality Strategy. AEA Technology, Report Number AEAT-5120. Culham Science Centre, Culham, OX14 3ED, Oxfordshire.
DoE (1997). Department of the Environment. The United Kingdom National Air Quality Strategy. The Stationary Office, March 1997, CM 3587.
DETR (1999a). Department of the Environment, Transport and the Regions. Review of the United Kingdom National Air Quality Strategy. A Consultation Document. Product Code 98EPO541/A
DETR (1999b). Department of the Environment, Transport and the Regions. An Economic Analysis of the National Air Quality Review Strategy Objectives. An Interim Report of the Interdepartmental Group on Costs and Benefits. Product Code 98EPO541/B
Dti, 1995. Energy Projections for the UK: Energy Use and Energy-Related Emissions of Carbon Dioxide in the UK, 1995-2020. March 1995.
Salway, A. G., Eggleston, H. S. Goodwin, J. W. L, Berry, J.E. and Murrells, T. P. (1999). UK Emissions of Air Pollutants 1970-1996. National Atmospheric Emissions Inventory, AEA Technology, National Environmental Technology Centre. Report AEAT-3092.
Stedman J.R. (1998). Revised High Resolution Maps of Background Air Pollutant Concentrations in the UK: 1996. AEA Technology, Report Number AEAT-3133. Culham Science Centre, Culham, OX14 3ED, Oxfordshire.
Stedman J.R, Linehan E and King (1999). Quantification of the Health Effects of Air Pollution in the United Kingdom for the Review of the National Air Quality Strategy. AEA Technology, Report Number AEAT-4715. Culham Science Centre, Culham, OX14 3ED, Oxfordshire.
Whyatt, 1998. J.D Whyatt, Lancaster University, personal communication, 4 September 1998.
This work was funded by the United Kingdom Department of the Environment Transport and the Regions as part of their Air Quality Research Programme.
Data for monitoring sites with the Joint Environment Programme of National Power, PowerGen and Eastern Electricity were provided by Alan Webb.
We are indebted to the regional offices of the Environment Agency for the provision of emission data relating to Part A processes.
Appendix A Figures
|
Comparison of modelled and observed annual average concentrations, NOx at AUN sites, 1995 |
|
|
Figure 2 |
Comparison of modelled concentrations with observed concentrations at rural and JEP sites. |
|
Figure 3 |
Comparison of modelled and observed concentrations at Automatic sites |
|
Figure 4 |
Comparison of concentrations at Smoke and SO2 network sites |
|
|
|
|
Annual sulphur dioxide concentration (ppb), 1996. Derived by dispersion modelling |
|
|
Map 2 |
Annual sulphur dioxide concentration (ppb), 1996. Derived by empirical modelling of monitoring data (Stedman, 1998) |
|
Map 3 |
Annual sulphur dioxide concentration (ppb), 2005. Concentrations predicted for the emission scenario provided by EP 65 |
|
Map 4 |
Annual sulphur dioxide concentration (ppb), 2005. Concentrations predicted for emission scenario CL + LFSD + upgrade |
|
Map 5 |
Annual sulphur dioxide concentration (ppb), 2005. Concentrations predicted for emission scenario CH + upgrade |
|
Map 6 |
Annual sulphur dioxide concentration (ppb), 2005. Concentrations predicted for emission scenario CL + upgrade |
|
|
|
|
|
|
|
|
|
