1.1 Sources of NMVOCs

Figure 1 shows source data from the National Atmospheric Emissions Inventory (NAEI) for Non-Methane Volatile Organic Compounds (NMVOCs) from 1990 to 2020, ordered by their 2020 values. The plot can be zoomed in on by clicking and dragging, and categories can be added and removed from the plot by clicking on the legend. VOC emissions from road vehicles and fuel use predominated emissions until around 2001. However, the predominant source today is solvent use.

1.2 Regulatory background

1.3 Network background and methods

The UK Hydrocarbon Network is one of several air quality monitoring networks operated by the Environment Agency on behalf of Defra to fulfill its statutory reporting requirements and policy needs. These include the Automatic Urban and Rural Network, which measures particulate matter, nitrogen dioxide (NO₂), carbon monoxide (CO), sulphur dioxide (SO₂) and ozone (O₃), Heavy Metals Network and Polycyclic Aromatic Hydrocarbon Network, which meet the requirements of the AQD and Fourth Daughter Directive (DD4). Other monitoring programmes including the Particle Concentrations and Numbers Network, Black Carbon Network and UK Eutrophying and Acidifying Pollutants Network exist to meet other requirements including those set out in the Air Quality Strategy.

2.1 Monitoring stations during 2021

The monitoring stations operating in the UK Hydrocarbon Network during 2021 are shown in Figure 2. Full names of the monitoring stations and their coordinates can be found in Appendix A1. Further details on the sites can be found on UK-AIR.

2.2 Equipment Maintenance and Audits

All non-automatic benzene monitoring stations were visited by Ricardo field engineers in April and October 2021 in order to carry out site audits and to undertake routine maintenance of the equipment. The main functions of these visits are to:

• Carry out certified flow measurements and calibration using a low flow BIOS instrument (UKAS accredited)
• Ensure no blockages or leaks in the system
• Clean or replace dirty filters and inspect/replace the sample inlet
• Replace O-rings and leak test all connections
• Carry out electrical Portable Appliance Testing (annually)
• Review the site infrastructure and surroundings
• Review health and safety risks at the site
• Replace or refurbish non-automatic sampler pumps

Routine flow measurements have been used to calculate sample volumes for the 2021 data set by means of interpolation. The calibration data from these audits have been used to rescale the benzene concentrations during the ratification process.

The automatic monitoring stations are serviced annually by the Equipment Support Unit (Perkin Elmer) where the following routine tasks are undertaken.

Annual preventative maintenance visits:

• Leak check all pneumatic systems
• Replace all consumables such as filters, gaskets
• Replace the cold trap
• Check and condition columns, trimming or replacing as necessary
• Checking and replacing transfer line if necessary
• Checking and replacing fused silica lines if necessary
• Replacing the nafion dryer if necessary

The Central Management and Co-ordination Unit (Ricardo Energy & Environment) provides an annual reference gas audit in addition to the automatic on site calibrations. These audits use the instrument sample port as opposed to the analyser calibration port. The sample line is inspected and cleaned/replaced annually.

The operational performance and stability of these types of automated chromatography systems can be affected for a period of time following ad-hoc repairs or power cuts. This means that an analyser that was only off for an hour might produce poor chromatography for a few days subsequent to that issue. Data obtained when the instrument is stabilising following repair will not be representative of ambient concentrations at the monitoring location. The ratification team will remove any such erroneous data up until the period when the data demonstrates that the instrument has stabilised and is producing meaningful data.

Ancillary equipment failure is the cause of most prolonged downtime. A spare hydrogen generator, Total Organic Carbon (TOC) zero air generator and air compressor is kept by the equipment Support Unit (ESU) such that equipment can be swapped quickly if necessary.

3.1 Estimation of Uncertainty

Calculated uncertainty for the Non-Automatic Hydrocarbon sites in 2021 for benzene is 15%, expressed at a 95% level of confidence. This includes contributions from Ricardo’s flow measurements, desorption efficiency and analysis uncertainty. The measurement uncertainty calculations for the individual hourly measurements are as defined in the standard operating procedures adopted under the ACTRIS project.

The requirement for benzene measurement uncertainty from an automatic hydrocarbon analyser is 25%, expressed at 95% confidence limit. The Perkin-Elmer analyser used in the UK network has not been type tested, as there is no reference method comparator so an estimate of the various contributions has been made to assess compliance with the DQO requirement. The main contributions are:

• Repeatability and lack of fit – derived if possible from the manufacturers specifications
• Variation in sample gas pressures, surrounding temperature and electrical voltage – derived if possible from the manufacturers specifications
• Interference from ozone – derived if possible from the manufacturers specifications
• Memory effects – derived if possible from the manufacturers specifications
• Differences between the sample and calibration port – these differences are negligible; the sample and calibration port are in contact with 90% of the same valve. Removing the calibration cylinder to evaluate this will disturb the system and affect sample measurements for some considerable time afterwards.
• Uncertainty in calibration gas – from NPL cylinder certificate
• Reproducibility under field conditions – this could be estimated from the manufacturers specifications
• Long term drift – corrections are made such that this is not applicable to the expanded uncertainty.

The largest components in the uncertainty budget are lack of fit and calibration gas uncertainty, although the calibration gas used is of the highest available quality. In the absence of data from type testing, the maximum permissible values stated in the EN Standard have been used as a worst case scenario. Using these values and the known values from the calibration cylinder the uncertainty budget has been calculated. The uncertainty of benzene measurements using a Perkin-Elmer analyser is estimated to be < 24%. Measurements reported to EMEP and ACTRIS are required to be reported with hourly uncertainty figures for each hour. The calculation used is provided in ACTRIS standard operating procedures:

\[ \chi_{sample, i} = (A_{sample,i} - A_{blank,i})/{V_{sample}} * f_{cal,i} \]

Equation 1: Calculation of mole fractions.

\[ f_{cal,i} = (V_{cal,i} * \chi_{cal,i})/(A_{cal,i} - A_{blank,i}) = 1/(C^i_{num} * C_{resp,i}) \]

Equation 2: Calculation of calibration factor.

Where

• \(A_{sample,i}\) = peak area of sample measurement of compound \(i\)
• \(A_{cal,i}\) = peak area of calibration gas measurement of compound \(i\)
• \(A_{blank,i}\) = possible blank value of compound \(i\) determined in zero gas measurements
• \(\chi_{cal,i}\) = certified mole fraction of calibration gas standard
• \(V_{cal}\) = sample volume of calibration gas
• \(V_{sample}\) = sample volume of sample
• \(C^i_{num}\) = Number of C atoms in the molecule \(i\) (e.g. for \(i\) = n-Pentane, \(C_{num}\) = 5)
• \(C_{resp,i}\) = mean C-response factor of compound \(i\)

3.2 Determination of precision

Precision covers random errors of peak integration, volume determination and blank variation.

Precision (\(\delta_{\chi_{prec}}\)) is determined as the standard deviation of a series of sample measurements (\(\sigma_{\chi_{sample}}\))

\[ \Delta_{\chi_{prec}} = \sigma\chi_{sample} \]

Equation 3: Definition of instrument precision.

This represents the instrument precision at the concentration level and complexity of the sample gas. Because the instrument is subjected to variable peak areas in ambient air, a more general description is applied:

\[ \Delta_{\chi_{prec}} = \sqrt{(1/3DL)^2 + (\chi*\sigma^{rel}\chi_{sample})^2} \]

Equation 4: Formula for calculating instrument precision.

Where

• \(DL\) = detection limit (described below)
• \(\chi\) = mole fraction of the ambient peak
• \(\sigma^{rel}\chi_{sample}\) = relative standard deviation of the sample (from the reference standard)

The detection limit is the dominant factor for small peaks, while reproducibility becomes more important for larger peaks.

3.3 Uncertainty calculations

The total uncertainty \(\Delta_{\chi^2_{unc}}\) of a single measurement does not only include the random errors described by the precision but also systematic errors \(\Delta_{\chi^2_{sys}}\) of the measurement.

\[ \Delta_{\chi^2_{unc}} = \Delta_{\chi^2_{prec}} + \Delta_{\chi^2_{sys}} \]

Equation 5: Calculation of total measurement uncertainty.

Errors included in the UK network calculations include:

• Uncertainty in the standard gas mole fraction \(\delta_{\chi_{cal}}\)
• Systematic errors (for example peak overlay/poor separation) \(\delta_{A_{int}}\)
• Sample volume error \(\delta\nu\)
• Other (system artefacts) \(\delta_{\chi_{instrument}}\)

The overall systematic error is described as:

\[ \Delta_{\chi^2_{sys}} = |\Delta_{\chi^2_{cal}} + \Delta_{\delta^2_{int}} + \Delta_{\chi^2_{vol}} + \Delta_{\chi^2_{instrument}}| \]

Equation 6: Calculation of overall systematic error.

Referring to Equation 1 and Equation 2, the single error contribution is determined for each measurement:

\[ \Delta_{\chi_{cal}} = (A_{sample} * \nu_{cal})/(\nu_{sample} * A_{cal}) * \delta_{\chi_{cal}} \]

Equation 7: Calculation of calibration error?

Where \(\delta_{\chi_{cal}}\) includes the certified relative uncertainty of the standard gas and possible drift.

\[ \Delta_{\chi^2_{int}} = (f_{cal}/\nu_{sample} * \delta_{A_{int.sample}})^2 + (A_{sample} * \nu_{cal} * \chi_{cal}/(\nu_{sample} * A^2_{cal}) * \delta_{A_{int.cal}})^2 \]

Equation 8: Calculation of relative uncertainty.

Where \(\delta_{A_{int.cal}}\) represents the reference gas standard deviation and the \(\delta_{A_{int.sample}}\) represents the sample gas standard deviation respectively.

Sample volume error cancels from Equation 1 as sample and calibration volume are identical.

3.4 Standard Methods

The AQD states that automatic measurements of benzene should be compliant with European Standard EN14662-3:2015 – Part 3 (CEN, 2015): Automated pumped sampling with in-situ gas chromatography which is determined as the Ambient Air Quality Standard method for the measurements of benzene concentrations. This Standard is for the determination of benzene in ambient air for the purpose of comparing measurement results with annual mean limit values. It describes guidelines for measurements with automated gas chromatographs, between 0 and 50 µg m^-3. Measurements undertaken by the Automatic Hydrocarbon Network are carried out in accordance with this Standard.

The Standard Method for measurement of benzene using an automatic analyser is in the process of review by CEN Working Group 12. Ricardo has a presence at CEN meetings, comments of which are summarised and sent to Defra following each meeting. At the time of publication of this report, the proposed revisions include a requirement for more rigorous linearity tests. The proposal states the linearity tests will be performed using at minimum the following concentrations: 0 %, 10 %, 50 % and 90 % of the maximum of the certification range of benzene or the user-defined range. At each concentration (including zero) at least 3 measurements shall be performed, the result of the first shall be discarded. The test shall be repeated at the following intervals:

• Within 1 year of the test at initial installation; subsequently:
• Within 1 year after test if the lack-of-fit is within 2.0 % to 5.0 %;
• Within 3 years if the lack of fit is ≤ 2.0 %;
• After repair

Currently the first three of these were already implemented on the UK network.

The AQD states that non-automatic measurements of benzene should be compliant with European Standard EN14662-1:2005 the Ambient Air Quality Standard method for measurement of benzene concentrations – Part 1: Pumped sampling followed by thermal desorption and gas chromatography. This Standard gives general guidance for the sampling and analysis of benzene in air by pumped sampling, thermal desorption and capillary gas chromatography. The pumped sampler was developed by the National Physical Laboratory in compliance with this standard. Ricardo contract SOCOTEC to analyse the samples in accordance with this standard. The non-automatic samplers were built specifically to meet the standard.

The AQD does not specify a standard method for the measurement of ozone pre-cursors (including formaldehyde), with the exception of benzene, as described above.

3.5 Limit of Detection

The Limit of Detection for the mass of benzene on a desorption tube from the Non-Automatic Hydrocarbon Network is approximately 5 ng. This is equivalent to about 0.05 µg m^-3 from a 14-day sample period.

The Limit of Detection for each of the 29 species measured by the Perkin Elmer Ozone Precursor Analysers is 10 ppt. These are converted to µg m^-3 and shown in Table 4.

4.1 Comparison with Limit Values and Objectives

The annual mean concentrations of benzene measured using non-automatic samplers over the calendar year 2021 are summarised in Figure 3, alongside the associated data capture rate. Figure 4 provides similar statistics for the automatic monitoring of benzene and 1,3-butadiene. Figure 5 shows the data capture rate for other automatically measured VOC species. Summary statistics for other measured pollutants can be seen in Appendix A2.

The increased maximum concentrations seen at London Eltham are due to two short episodes where air masses travelling to the station from a southerly direction on 30^th October and 11^th December showed an unusually high concentration for 1,3-butadiene. For benzene, the same episode on 11^th December resulted in the maximum value recorded.

Figure 3: Non-Automatic Benzene annual statistics for 2021. Range shows the annual minimum and maximum nominal fortnightly measurement.

Figure 4: Benzene and 1,3-butadiene annual statistics for 2021. Range shows the annual minimum and maximum hourly measurement.

Figure 5: VOCs annual statistics for 2021. The dashed lines show the objective data capture rate; any VOC below the objective are highlighted in red.

Annual time weighted mean concentrations at all monitoring stations were below the Limit Value of 5 µg m^-3 for benzene set by the European Ambient Air Quality Directive as well as the UK Air Quality Objectives as defined in the Air Quality Strategy 2007.

The data flags used in the Implementing Provisions Regulations (IPR) are applied using a program, written by Ricardo.

4.2 Instrumental Impacts on Data Capture

The automatic system comprises several components listed below:

• Turbomatrix Thermal Desorber (TD)
• Sample vacuum pump
• Clarus 500 Gas Chromatograph (GC)
• Zero Air generator
• Air Compressor
• Hydrogen Generator
• High Volume Flow Inlet (including a fan)
• Site PC including Totalchrom software

These components are checked by local site operators on a fortnightly basis. The system manufacturer (Perkin Elmer) carries out annual preventative maintenance. The data from the system is checked Monday to Friday by Ricardo’s daily data checking team. If there is an instrument failure Perkin Elmer engineers are called out to the site to repair the problem. There are no hot spare Thermal Desorbers or Gas Chromatographs, so some considerable downtime is possible if the instrument fault cannot be diagnosed and/or repaired quickly.

Further data loss is likely due to instrument detector stability following power cuts, preventative maintenance visits and instrument faults. It can take several days for the instrument to stabilise. This problem is unavoidable with chromatography, we ensure all faults are diagnosed within 48 hours (excluding weekends and public holidays), and all faults are repaired following diagnosis unless this is not possible, for example where a component has failed that needs to be ordered. In addition, specific peaks can be affected by co-elution with a system artefact or other unknown hydrocarbons resulting in reduced data capture for individual species.

4.3 Long term trend of benzene concentration

Figure 6 shows the trends in benzene concentration averaged across four main site types with a smooth trend line fitted. The plot reveals that the highest concentrations were generally observed at roadside sites until 2008. Note that there are only two industrial sites, so these trends are noisier than for other site types. What is clear from the trend analysis is that concentrations of benzene decreased sharply from 2002 to 2008, which reflects better emissions control on vehicles (exhaust and evaporative emissions). For 2021, roadside concentrations are on average much closer to background concentrations than they were in the early 2000s.

Figure 6: Average Non-Automatic network benzene means by site type from 2002 to 2021.

Trend estimates using robust statistical techniques in openair provide a way of quantifying the trends over time as a percentage change in benzene concentration per year. The trends have generally shown two characteristics: a decrease from 2002 to 2008 and then a period of stabilisation from 2008 to 2021. Figure 7 separately considers the trends for these two periods. While the post-2008 trends are significant, the magnitude of their slopes are notably smaller than pre-2008, suggesting more stable benzene concentrations. 2020 and 2021 are likely lower than expected due to Covid lockdown restrictions resulting in reductions in road transport and therefore vehicle emissions in urban areas.

Figure 7: Non-automatic benzene rate of change for 2002-2007 and 2008-2021.

To compare trend in benzene concentrations across all measurement sites, Figure 8 shows the trend in annual average benzene concentrations at non-automatic monitoring sites from 2002 to 2021. To help with interpretation the trends are ordered and a dashed line is shown for zero change.

The plot below shows that all sites showed a decrease in benzene concentration over the period 2002 to 2021. The magnitude of the decrease is larger between 2002 and 2007 compared to more recent years. Note that the error bars relate to the 95% confidence intervals, which reveals that the trends at some sites are relatively uncertain.

Figure 8: Benzene concentration trend (μg m^-3 / yr) at non-automatic sites from 2002 to 2021.

Data obtained from the National Atmospheric Emissions Inventory (NAEI) can be used to see if there is a long-term relationship between emissions and measurements. Normalisation involved dividing the annual means by the mean value for the whole time series. Figure 9 shows that the NAEI urban benzene emissions data agrees with the monitoring data, where a sharp decline can be seen from 1995 up to the year 2000 where petrol benzene additives were replaced by toluene for the purposes of maintaining the required octane value. The emissions data is steadily decreasing since 2000, but the monitoring data has stabilised, possibly due to additional urban sources of benzene such as use of wood burning appliances for domestic and commercial space and water heating. NAEI benzene data for 2021 is not yet available.

Figure 9: Normalised annual average benzene concentration and NAEI benzene emission estimate (1995 - 2020).

4.4 Impact of VOCs on regional O₃ formation

4.4.1 Regional O₃ increment

The UK measures different species of VOCs due to their ozone creation potential. To understand the contribution of individual VOCs to the O₃ production at the UK (regional) scale, we first need to quantify the O₃ increment at the UK level. This section presents evidence of the regional ozone formation and the relative contribution from each VOC. The methodology of the analysis in this section follows that set out by Malley et al. (2015).

In this report, the regional O₃ production is defined as the O₃ concentration measured at Chilbolton minus the northern hemispheric background O₃ concentration. Chilbolton is a rural monitoring site generally representing the regional background O₃ in southern England. The hemispheric background O₃ concentrations were derived from measurements at Mace Head coupled with air-mass back trajectory analysis. 96h air-mass back trajectories arriving daily at Mace Head were modelled with the Hysplit model and were grouped into 4 clusters based on the similarity of the angle of each trajectory from the origin (Mace Head). These clusters are visualised in Figure 10. The hours corresponding to trajectory clusters from the west and southwest (C2 and C4) are considered to be “clean” air masses, and therefore representing the hemispheric background O₃ concentration.

Figure 10: Air mass back trajectories arriving at Mace Head in 2021

To quantify the regional O₃ increment, 288 month-hourly average O₃ concentrations were calculated at Chilbolton and for the “clean” air masses at Mace Head. The difference between the two is shown in Figure 11 and Figure 12. In general, we would expect higher concentrations for the hemispheric background O₃, i.e. negative values in Figure 11 and Figure 12. However, elevated regional background O₃ concentrations due to photochemical reactions were evident especially during the summer day hours.

2021

Figure 11: Month–hourly average differences between regional and hemispheric background O₃ for 2021 at Chilbolton.

2020

Figure 12: Month–hourly average differences between regional and hemispheric background O₃ for 2020 at Chilbolton.

4.4.2 Photochemical ozone creation potential

Multiple studies have tried to quantify the propensity of each VOC to produce O₃ and derived the photochemical ozone creation potential (POCP) (Derwent et al., 2007; Hakami et al., 2004; Luecken and Mebust, 2008). The POCPs derived by Derwent et al. (2007) are used in this analysis as they were calculated under simulated north-western European conditions, which is most relevant in the UK context. In Derwent et al. (2007), a VOC POCP was defined as the ratio (multiplied by 100) of the increase in O₃ due to increased emissions of the VOC simulated in a Lagrangian model along a trajectory traversing from central Europe to the UK, relative to the modelled increase in O₃ from the same mass increase in emissions of ethene (the reference POCP VOC assigned a value of 100). Hourly VOC concentrations were multiplied by the corresponding POCPs to weight their potential to create O₃. Diurnal variation of the weighted VOC concentrations at Chilbolton represents the extent of photochemical depletion that occurred in the southern England, and in turn leads to the regional O₃ creation. It is acknowledged that part of the diurnal variation in VOC concentrations is due to changes in the boundary layer mixing depth. To eliminate this effect, hourly POCP-weighted VOC concentrations were normalised by the corresponding POCP-weighted ethane concentrations. Ethane was chosen because it has the second smallest POCP of the measured VOCs (see Table 3) and has previously been used to estimate photochemical loss of VOCs (Malley et al., 2015; Yates et al., 2010). The VOC diurnal photochemical depletion is then calculated as the difference between the average POCP-weighted VOC/ethane ratio at night (00:00 – 04:00) and in the afternoon (12:00 – 16:00). A positive value indicates daytime photochemical depletion of the VOC relative to ethane. The median of the VOC diurnal photochemical depletion for each month is summarised in Figure 13. The seasonal pattern of VOC diurnal photochemical depletion matches the seasonal pattern of regional O₃ increment as observed in Figure 11, i.e. largest VOC diurnal photochemical depletion and regional O₃ increment in the summer months. The magnitude of the VOC diurnal photochemical depletion indicates relative contribution of each VOC to total VOC photochemical depletion, which in turn represents the relative contribution to the production of regional O₃.

Figure 13: Median of the diurnal VOC photochemical depletion at Chilbolton for each month in 2021. VOCs are ordered by their average diurnal photochemical depletion.

To compare the relative contribution of each VOC to the regional O₃ increment, Figure 14 shows the median VOC diurnal photochemical depletion values in descending order for the months when regional O₃ increment was observed. Ethene consistently showed the largest contribution to regional O₃ increment during these months. M-xylene, p-xylene, toluene, 1,2,4-trimethylbenzene, o-xylene, propane, n-butane, and propene also had a notable contribution to regional O₃ creation and were comparatively larger than other VOC species. The result might suggest that, of the measured VOCs, reduction in emission of Ethene, m-xylene, p-xylene, toluene, 1,2,4-trimethylbenzene, o-xylene, propane, n-butane, and propene emissions would be most effective in reducing regional O₃ increment.

Figure 14: Median of the VOC diurnal photochemical depletion at Chilbolton between May and September in 2021. Error bar indicates the 25 and 75% percentile of the VOC diurnal photochemical depletion for a that VOC in the relevant month.

4.6 Benzene Modelling

Ricardo undertakes compliance modelling activities on behalf of Defra under contract AQ0650. These activities use the national hydrocarbons network measurements to support the benzene model. The model results, in combination with the measurements from the hydrocarbons network, then form the basis of the annual compliance assessment for benzene under the AQD.

The latest report available detailing the modelling methodology and compliance results is presented on Defra’s UK-AIR website for 2018, subsequent annual update reports will follow.

There were no exceedances of the 5 µg m^-3 annual mean limit value modelled in 2020. The highest estimated background concentration was 2.45 µg m^-3 at South Killingholme which is predominantly caused by emissions from combustion in industry.

5.1 EN14662-1:2005

European Standard EN14662-3:2005 is currently under revision by CEN Working Group 12. The replacement refers to a sequential benzene sampler as opposed to a simple single tube system. The replacement document is likely to be released in 2023. The existing samplers on the non-automatic hydrocarbons network would need to be replaced with new samplers once manufacturers have been able to design, test and manufacture new instrumentation to meet the requirements of the new standard method.

5.2 Standard method for ozone precursors

In Europe, there has never been a standard method for the measurement of ozone precursors to date. Under Working Group 12, funding from the European Commission had been requested in order to start a five-year process for development of such a standard. A mandate for this has now been issued, however the funding requested was costed for the preparation of a single standard method, the EC issued a mandate for six standard methods to be developed. These include:

• automatic pumped sampling, pre-concentration and on-line gas chromatography with flame ionisation detector (FID) and/or mass spectrometer detector (MSD);
• manual or automatic canister sampling followed by off-line gas chromatography with FID and/or MSD;
• manual or automatic pumped sampling followed by off-line thermal desorption and gas chromatography with FID and/or MSD;
• diffusive sampling followed by thermal desorption by off-line gas chromatography with FID and/or MSD;
• manual or automatic pumped sampling of formaldehyde on dinitrophenylhydrazine (DNPH) followed by off-line high performance liquid chromatography (HPLC)/ ultraviolet (UV) detection;
• diffusive sampling of formaldehyde on DNPH followed by off-line HPLC/UV detection.

A first draft work programme has been provided to the EC in August 2019. Working Group 13 has been resumed in order to cover most parts of this, with Working Group 11 covering the diffusive methodologies. Ricardo maintain one member within working group 13.

5.3 Requirement for network changes

From 2017, modelled benzene data (from Defra’s Pollution Climate Mapping model) suggests the highest background concentrations of benzene are in South Killingholme, North Lincolnshire. Modelled concentrations of benzene were 3.4 µg m^-3 in 2017, 3.2 µg m^-3 in 2018, 3.25 µg m^-3 in 2019 and 2.45 µg m^-3 in 2020. The model has predicted higher levels in this area previously as a result of the emissions inventory. However, measurements made in the area in 2008 (Butterfield et al., 2009) confirmed no exceedances of the limit value, therefore no repeat campaign would be necessary.