Historically the NAEI’s main aim has been to calculate historical and projected emissions trends at a national level for significant air pollutants. There is now increasing demand for emission estimates for small areas and emission mapping for deposition and atmospheric concentration modelling. While national emission mapping will not replace detailed local inventories it can provide a valuable basis from which a more detailed local inventory can quickly be built and validated.

This appendix gives an overview of the methodology used to generate the 1999 UK emission maps for different pollutants. This effectively constitutes an update on Goodwin et al (1997) and Goodwin and King (2000), but much of the detail is not included here.

The method used to map emissions in the UK is shaped by the data that are available. Purpose built surveys of the UK that collect data necessary for a detailed inventory featuring every significant local source in its exact location are too time consuming and expensive to be undertaken. There are, however, generic data that can be used to distribute national totals of emissions, such as the population census, employment surveys and land use classifications. There are a number of problems associated with using these types of data. Data are often disparate, rarely do they cover the whole UK and costs can be prohibitive. However, by spending relatively little time and effort searching out this generic data, resources can be made available for the detailed "survey" method in key areas.

The sources contributing to the UK emissions can be represented as one of three categories: points, lines or areas. Sectors such as power stations, refineries and large industrial plant can be represented by points. Their locations are known and data to estimate emission contributions are available. Major roads, railways and shipping are sectors that can be represented by lines if data are available. Other diverse and numerous source sectors such as agriculture, domestic and commercial are represented by areas.

The emission maps combine a number of different source sectors represented by combinations of points, lines and areas. In order to map this combination, the UK is divided into a grid of 1km squares. Emissions are then represented in terms of tonnes per 1km grid square.

A point source is a site where emissions are attributed to a single point located by a grid reference. Point source emissions are derived in a number of ways depending on the availability of data. Power stations, for example, report emissions of some pollutants but also provide fuel use data so that other pollutant emissions can be calculated, providing a reliable "bottom up" inventory for the sector.

There are other sectors for which it is preferable to calculate a national total estimate, which is then distributed by using a related surrogate statistic that is known for each site. For example this surrogate takes the form of fuel use statistics for industrial combustion and acid production capacity for acid plant.

Sectors such as road transport and railways are represented by a number of lines. Road traffic emissions are calculated for each link based on traffic count data compiled by The Department of Environment Transport and the Regions. Railway emissions are mapped using track and vehicle kilometres data compiled by London Research Centre.

Source sectors represented by areas include domestic, small industrial combustion and processes, agriculture and commercial emissions. National totals are distributed over the UK based on surrogate statistics represented in area grids. These grids are compiled from available geographical statistical data such as land use, population and employment. Often other surrogate statistics are derived from combinations of these basic data sets to produce a more appropriate distribution surrogate.

Maps of emissions for various pollutants have been included in the main body of this report.

Emissions from point sources represent sectors of the UK inventory either fully (such as power stations where the sector is made up of large sites for which emissions reporting is mandatory) or in part (such as combustion in industry, for which only the large sites within the sector are required to report emissions). In the latter case, the remainder of the emissions for the sector are mapped as an area source.

Emissions for the point sources have been compiled using a number of different data sources and techniques. The primary data sources for most of the emissions data have been:

• Local Authority held emissions, location, operation data for significant NMVOC emitting part "B" installations.

Where reported emissions data are available for 1999, these data have been used. However, where reported data are not available, estimates are used. Two types of estimates are made: (a) for pollutants not reported at sites where some emissions data is available, or (b) for sites that do not report emissions.

The road transport mapping combines both line and area source emissions. Where traffic flow data are available (i.e. for major roads) emissions are calculated for each road link. Where traffic flow data are not available on a road link basis (i.e. for minor roads) emissions are estimated at a national level and distributed based on surrogates derived from OS minor road maps, land cover data and regional average flows.

Emissions for the major roads are estimated for each road link by vehicle type based on traffic flow data provided by DETR and the Department of Environment for Northern Ireland (DoE NI). The dataset for 1999 proved to vcontain significantly more data than the 1998 dataset. Consequently updating to the 1999 numbers has provided a more extensive coverage of traffic flow data. This data has also been combined to generate a major road traffic flow map for the whole UK.

The traffic flows from the DETR’s 3 year rolling survey are assigned to the new road links to provide a flow estimate for each stretch of the GB A-road and Motorway network. The more extensive 1999 dataset was used.

Average road speed classes are assigned to each road based on four main attributes:

• Road Type
• Road setting: 14 classifications from Central London to Rural. The M25 and nearby roads are treated as a special case with a reduced average vehicle speed.

Traffic flow data from the Annual Traffic Census Report 1999 (Roads Service) and Vehicle Kilometres of Travel Survey of Northern Ireland 1998 (Oscar Faber for the Roads Service) are used to populate the major road links. The basic vehicle mix for some count points are also available in the Vehicle Kilometres of Travel Survey report. For roads without data, a generic vehicle mix and average vehicle split is applied.

The Northern Ireland roads do not have any predefined road attributes. In order to assign appropriate speeds they are classified into Urban and Rural roads. A satellite land use map is used to classify the major roads as built up or non-built up. Roads that intersect with a grid square containing greater than 50% urban land are classified as a ‘built up’ and assigned average speeds appropriately. It is recognised that this results in an overestimate of urban links because of the length of some links extending into non-built up grid squares. Similar average vehicle speeds are assigned to the roads as for the GB data set.

The emissions for each major road are calculated by applying a speed emission equation to the traffic flows and derived speeds for each road link. The speed emission equations are derived from the NAEI road transport model (Goodwin et al, 1999) and are representative of the average age mix of vehicles across the UK. These equations also take account of assumed catalyst failure rates of 5% per year.

The emissions, vehicle kilometres and fuel use estimates by road link are aggregated into 1x1km cells to be added to the UK emission maps or added to the minor road grids and used to distribute other emissions such as break and tyre wear.

Traffic flow data for the minor roads are not available at a level where they can be used to calculate road link emissions as done for the major roads. Thus, a top-down approach is used. National total emissions from road transport are calculated using a comprehensive transport model including data on the fleet composition, vehicle kilometres and speed-emission profiles (Goodwin et al, 1999). The minor road component is calculated for each pollutant as the difference between the UK total road transport emissions and the major road emissions (calculated as described above). This minor road ‘remainder’ emission amount is then treated as an area sources and distributed across the UK using the surrogate data sets described below.

The OS "Meridian" and OS Northern Ireland road maps are used as the basis for the UK minor road distribution. The total lengths of each type of minor road within each 1x1km grid square is calculated. An appropriate regional average traffic flow, provided by DETR and DoE NI, is applied to this total length per cell to estimate the total vehicle kilometres per cell. Vehicle kilometre grids are calculated in this way for each of the minor road classes (B, C and U) for urban (U) and rural (R) areas.

The vehicle kilometre grid maps for each minor road class are then used to generate grids of fuel use for petrol and diesel road transport assuming uniform vehicle mix for each BU, BR, CU, CR, UU and UR road type.

These petrol and diesel distribution grids are used to map the minor road component of the hot exhaust road transport emissions.

Cold start kilometres for Great Britain are available for cars by average trip length and trip type. This enables accurate estimates of the associated emissions to be determined. Cold start emissions were assumed to have similar characteristics in Northern Ireland. Evaporative emissions have been distributed using a normalised population grid. UK Total emissions from break and tyre wear are distributed using 1x1km surrogate grids prepared from vehicle km estimates on major and minor roads.

Industrial combustion includes a number of different source sectors:

• Refineries,
• Iron and Steel,
• Other Industry,
• Auto-generators (industrial sites generating their own electricity),
• Power Stations.

Much of the emissions form the Refineries, Iron and Steel and Power Stations are represented as point sources. These sectors contain mostly large sites for which emissions can be estimated on a site by site basis.

However, there are significant industrial combustion emissions from smaller combustion plant that are not reported, or reported data is inaccessible. As the NAEI industrial combustion emissions are calculated as a national level, the sum of all reported industrial emissions is therefore less than the UK total for this sector. The reported point source emissions are therefore subtracted from the UK total, and the remainder is assumed to be from small industrial combustion and distributed as an area source using surrogate statistics.

To develop a distribution for small industrial combustion sources the following data sets are used:

• Office of National Statistics (ONS) Inter-Departmental Business Register (IDBR) which provides data on employment at unit level database; and
• ETSU functions for calculating energy demand by employee for different industrial sectors (ETSU, 1999)

The employment numbers for each industrial classification are manipulated using algorithms developed by ETSU. These algorithms model the energy demands per employee by sectors and the propensity for use of solid/liquid or gas fuels. An urban mask has been used to remove solid and liquid fuel emissions from the large urban centres where it is assumed that gas is burned.

Different distribution grids were developed for different industrial fuel use sectors using relevant sectors from the IDBR employment database.

Process emissions are divided across numerous sectors. Many of these sectors are exclusively mapped using point source data which, in most cases, have been generated as part of the procedure to produce national totals i.e. estimates are made for each site and summed to give the national total. In a few cases where these type of data are not available, it has been possible to divide the national total, generated from the use of an emission factor and activity statistic, between known sites so as to produce a set of point source data for mapping. Sectors or parts of sectors that are not accounted for by point source emissions are distributed using employment statistics from the ONS IDBR database. Process emissions are assumed to be proportional to employment within the limitations of sectors selected from the IDBR database.

The BGS (British Geological Survey) has provided a database of mines and quarries in the UK. This data set includes the location of the site and a brief description of products and commodities. Neither production or any indicator of the size of the sites are given. In order to provide some estimates of the relative contributions for each site, employment data for quarrying industries from the ONS IDBR database has been used.

National emissions to a ceiling of 1000m are calculated from the number of aircraft movements at UK airports on an individual airport basis according to the assumed mix of aircraft. The methods incorporate the results of detailed urban inventory studies. Aircraft emissions are treated as area sources for the 1x1km grids covering the airports. For a number of airports take off and landing pattern data are available, enabling a spatial distribution to be determined.

UK total emission estimates from support vehicles are distributed over the physical airport area based on aircraft arrival statistics for each airport.

UK shipping totals are calculated at a national level. These estimates include:

• Coastal shipping (Fuel sales for ships confined to UK waters),
• Fishing vessels,
• Other UK shipping (UK bunker fuel sales corrected for estimation of emissions in port and cruise in UK waters only).

These emissions are distributed over the UK as follows. Emissions are assigned to port areas and routes according to ship arrival data provided by DETR (Marine Statistics Yearbook) The current routes have been digitised and represent the central path through estuaries and then to an approximate 12 mile offshore boundary. The port areas are also approximate to the real port locations. This gives a rough approximation of where shipping is likely to be.

Nationally estimated agricultural emissions are distributed using employment statistics for agricultural activities aggregated into 1x1km grids.

To develop a distribution for agricultural combustion sources the following data sets are used.

• ONS IDBR database of employment at unit level
• ETSU energy demand by employee for different industrial sectors (ETSU,1999)

The employment numbers for each agricultural classifications were modified using the algorithms developed by ETSU. These algorithms model the energy demands per employee by sectors and the propensity for use of solid/liquid or gas fuels. An urban mask has been used to remove solid and liquid emissions from the large urban centres where it is assumed that gas is burned.

Emissions from offshore installations are provided at an installation level by UKOOA. These include:

• Offshore flaring,
• Offshore loading,
• Offshore own gas use,
• Offshore oil and gas operations.

These estimates are aggregated for the UK totals. For the UK emission maps, the installation reported emissions are assigned to locations provided by the UK Hydrographic Office based on the Company Name and field location.

Diesel and Gas Oil fuel use is incorporated in the NAEI coastal shipping sector as the majority of fuel burned by offshore operations is for shipping. As a result pollutants that have emissions from liquid fuels only will not feature emissions at the offshore installations.

Maps of the spatial distribution of atmospheric emissions are a key input to any air quality assessment. The reliability of emissions maps should be verified if they are to be used to model potential exceedances of air quality objectives. It is helpful to draw a distinction between emission inventory validation and verification. Validation is the process of checking that emissions have been estimated using the appropriate protocols, while verification involves comparison with independently derived data such as ambient monitoring results.

The use of monitoring data to verify the spatial pattern of atmospheric emissions has been described by Goodwin et al (1997). Measured annual mean background concentrations can be considered to be made up of two parts:

• A contribution from relatively distant major point and area sources such as power stations or large conurbations. Measurements from monitoring sites well away from local sources, from rural sites within the UK Acid Deposition Secondary Network, for example, provide good indications of the spatial variation of concentrations due to distant sources.
• A contribution from more local emissions. The spatial scale at which local emissions seem to influence ambient air quality has been studied in earlier work on the estimation of air pollutant concentrations from emission related parameters (Stedman et al, 1997). It was found that estimates of emissions in an area of 25 km² centred on a background monitoring site provide the most robust relationships.

The difference, diff, between measured ambient concentrations at urban automatic monitoring sites (not roadside, industrial sites or urban or suburban sites that are significantly influenced by emissions from a nearby road) and an underlying rural concentration field is calculated where monitoring data are available.

Comparisons between this difference for annual mean NO_x and the total of area NO_x emission (in the area of 25 km² centred on background monitoring sites) indicate a strong relationship.

Local emissions within an area of 25 km² have been used to estimate the contribution that urban emissions make to ambient NO_x concentration in urban background locations. It would be reasonable to assume, however, that for urban areas larger than 25 km² there should also be a contribution from more distant emissions within the same urban area. This has been investigated in some detail but it was not possible to find robust relationships between measured concentrations and emission rates for concentric regions larger than 25 km². Sites in large urban areas are noted to give a differing relationship, which is postulated to arise from a sizeable contribution from more distant grid cells within the urban area.

The determined relationships can be used derive maps of estimated background air pollutant concentrations as described by Stedman et al (1997). These maps can then be used to subtract this background component from measured annual mean NO_x concentrations at roadside monitoring sites:

For NO_x there is clearly a relationship between the roadside increment and estimates of NO_x emissions i.e. the dominant influence of emission strength is traffic flow.

The verification of the spatial distribution of other pollutants can also be carried out using similar methods to those described above. Inventory verification for pollutants such as PM₁₀ is, however, more problematic due to the diverse nature of PM₁₀ and the range of sources of primary combustion, secondary and mechanically generated coarse particles.