3.2.1 Annual mean background NO_X and contributions from local area sources

Dispersion kernels generated with ADMS 5.2 were used to calculate the contribution of local area source emissions to ambient NO_X concentrations at a central receptor location from area source emissions within a 33 km x 33 km square surrounding each receptor. 2020 hourly sequential meteorological data from the Weather Research and Forecasting (WRF) numerical weather prediction modelling system, were used to construct the dispersion kernels. More information on the meteorological input data to the model can be found in Appendix 4 of the 2020 UK modelling report (Pugsley et al. 2022).

The modelled NO_X contributions from point sources, combustion in industry area sources and regional rural NO_X sources have been subtracted from the measured annual mean NO_X concentration at background sites, to derive a “Corrected” measured annual mean NO_X for calibration purposes. The uncalibrated modelled local area source contribution to ambient NO_X concentrations was compared to these “Corrected” measurements from eleven background NO_X monitoring sites within the Scottish network to calibrate the model.

Figure 3 shows the calibration plot for the Scotland-specific NO_X local area source model. The gradient of the line of best fit forced through the origin was 1.506, and this value was used as the Scotland-specific NO_X area source calibration coefficient. The regression line has been forced through the origin to provide a reasonable model output without imposing an unrealistically high residual on the area source contribution. For each grid square in Scotland the uncalibrated modelled local area source contribution was multiplied by the Scotland-specific calibration coefficient to calculate the calibrated area source contribution. Point source, combustion in industry area source and regional rural contributions from distant sources were then added, resulting in a map of total background annual mean NO_X concentrations for Scotland for 2020 as shown in Figure 22. A detailed description of the background modelling approach for NO_X and NO₂ can be found in Section 4 of the 2020 UK modelling report (Pugsley et al. 2022).

Figure 3: Calibration of the Scotland-specific background NO_X model for the year 2020.

As noted above, background area source NO_X model calibration coefficients applied in the Scotland-specific and UK PCM models are compared in Figure 4. This presents the calibration coefficients from 2008 to 2020 and highlights:

1. the year-to-year variation in the Scotland-specific background area source NO_X model calibration coefficient;
2. the extent to which these values vary in both models.

From Figure 4 it can be seen that the 2020 Scotland-specific background area source NO_X calibration coefficient was greater than the equivalent value used in the UK PCM model in 2020 due to the different mix of sites included in the UK and Scotland-specific PCM model calibrations. Overall the Scotland and UK coefficients are quite similar for each year.

Figure 4: Background model calibration coefficients applied in the Scotland-specific and UK models for NO_X for 2008 to 2020.

3.2.2 Roadside NO_X concentrations

It is assumed that the annual mean NO_X concentration at roadside locations (NO_{X, roadside}) is made up of two components: the background concentrations as described above (NO_{X, background}) and a local roadside increment (NO_{X, roadside − increment}):

• NO_{X, roadside} = NO_{X, background} + NO_{X, roadside − increment}

The NAEI provided estimates of NO_X emissions for major road links in Scotland for 2019 (Churchill et al. 2022). In 2020 the roadside increment was calculated using the PCM Roads Kernel Model (PCM-RKM) which was introduced into the UK PCM model in 2014. The PCM-RKM uses the ADMS-Roads 5.0 dispersion model to calculate the roadside increment to NO_X concentrations on urban major roads. Individual model runs were carried out for 950 census points covering urban major roads in Scotland. Each model run is parameterised using specific input data for the census point:

• Road geometry
• Traffic speeds, emissions and traffic counts
• Meteorology
• Receptor locations

The input data used for this Scotland-specific modelling is consistent with the data used for UK PCM modelling and is described in detail in Appendix 8 of the 2020 UK modelling report (Pugsley et al. 2022). Modelled concentrations were derived from the ADMS-Roads model outputs at a distance of 4 m from the kerb, averaged across each side of the road for all road links modelled. The PCM-RKM represents a more process-based approach than the previous empirical method. It provides a more robust assessment, whilst retaining the link with measurement data provided by the use of measurement data to calibrate this component of the model.

The PCM-RKM modelled roadside increment is calibrated by comparison with an empirically calculated roadside increment (measured roadside NO_X concentration minus modelled background NO_X concentration) at Scottish roadside or kerbside monitoring sites. Calibration of the PCM-RKM is shown in Figure 5. The background NO_X component at these roadside monitoring sites is taken from the background NO_X map described in the previous section. The calibration coefficient derived is then applied to the roadside increment calculated using the PCM-RKM for each road link. The background NO_X is added to the calibrated roadside increment, resulting in a map of total roadside annual mean NO_X concentrations for Scotland for 2020.

Of the seventy one roadside NO_X monitoring sites within the Scottish network, thirty seven had sufficient data capture for NO_X (and NO₂) in 2020 and were located on modelled major road links close to census points at which the traffic flow along the road was measured. The roadside NO_X measurements from seventeen of these sites were used to calibrate the model. Scottish NO_X air quality monitoring data from twenty roadside sites was used to verify the model output.

A total of nine roadside NO_X monitoring sites were excluded on the basis of low annual data capture (annual percentage data capture <75%). A further twenty one roadside sites, with sufficient data capture, were excluded because they were not located on a modelled urban major road link. A total of four roadside sites were excluded because they were not located on a modelled urban major road link, and also had low data capture. The air quality monitoring sites outside of the AURN are typically installed at kerbside or roadside for AQMA purposes and may not be located on modelled major road links.

Figure 5: Calibration of the Scotland-specific roadside increment NO_X model for the year 2020.

Figure 6 provides a comparison of the 2020 Scotland-specific and UK roadside NO_X calibration coefficients. The Scotland-specific and UK roadside NO_X calibration coefficients from recent previous years are also presented. From Figure 6 it can be seen that the 2020 Scotland-specific roadside NO_X calibration coefficient was greater than the value used in the 2020 UK PCM model, which as noted in relation to the background model is due to the different mix of sites included in the UK and Scotland-specific PCM model calibrations.

Figure 6: Roadside model calibration coefficients applied in the Scotland-specific and UK models for NO_X for 2014 to 2020.

3.2.3 NO₂ modelling

Maps of the estimated 2020 annual mean background and roadside Scottish NO₂ concentrations were calculated from the modelled NO_X concentrations using a calibrated version of the updated oxidant-partitioning model (Jenkin 2004, Jenkin 2012; Murrells et al.  2008). This model uses representative equations to account for the chemical coupling of O₃, NO and NO₂ within the atmosphere. A key advantage of this approach for modelling NO₂ concentrations is that different emission scenarios can be addressed by varying regional oxidant levels and/or primary NO₂ emissions.

This approach is discussed in detail in Section 4.4 of the 2020 UK modelling report (Pugsley et al. 2022). Briefly, the oxidant-partitioning model enables NO₂ concentrations to be calculated using the following equations:

• [NO₂] = [OX].f(NO_X)
• [OX] = fNO₂.[NO_X] + [OX]_B

Where f(NO_X) expresses [NO₂]/[OX] as a function of [NO_X] and hence allows the calculation of the concentration of NO₂ from the concentration of NO_X when combined with [OX], the total oxidant (the sum of NO₂ and O₃ concentrations). Here fNO₂ is the primary NO₂ emission fraction (defined as the proportion of, or volume ratio, of NO_X that is directly emitted as NO₂), and [OX]_B is the regional oxidant. The regional oxidant component of the NO_X-NO₂ model remains unchanged from the UK PCM modelling (see Pugsley et al. 2022) as this is not affected by the Scotland-specific changes to the UK PCM model noted in Section 2.2. NO_X, NO₂, O₃ and OX are all expressed as ppb in these equations: 1 ppb of O₃ = 2 μg m^-3; 1 ppb of NO₂ = 1.91 μg m^-3. By convention when NO_X is expressed in μg m^-3 it is expressed as “μg m^-3 as NO₂” therefore 1 ppb of NO_X = 1.91 μg m^-3 of NO_X as NO₂.

In Jenkin (2004), [NO₂]/[OX] was calculated using two equations, one of which represented background locations and the other roadside locations. Updated equations for [NO₂]/[OX] were subsequently developed in Murrells et al. (2008). More recently, Jenkin (2012) found that short term variability in NO_X concentrations is a major cause of the scatter in the relationship between [NO₂]/[OX] and [NO_X]. The ratio of the upper to the lower quartile of hourly concentrations is a good indicator of this variability and this ratio increases with decreasing NO_X concentrations at roadside and background sites. This dependence has been used to interpolate between equations based on a constant NO_X quartile ratio. This led to two equations for calculating [NO₂]/[OX], one of which represents background locations and the other roadside locations, derived from background and roadside monitoring sites respectively. The Jenkin (2012) equations are as follows:

• For background locations: [NO₂]/[OX] =  − 2.5124 × 10^− 13[NO_X]⁶ + 1.5805 × 10^− 10[NO_X]⁵ − 4.1429 × 10^− 8[NO_X]⁴ + 5.8239 × 10^− 6[NO_X]³ − 4.8076 × 10^− 4[NO_X]² + 2.5916 × 10^− 2[NO_X]
• For roadside locations: [NO₂]/[OX] =  − 2.0901 × 10^− 13[NO_X]⁶ + 1.5001 × 10^− 10[NO_X]⁵ − 4.2894 × 10^− 8[NO_X]⁴ + 6.2659 × 10^− 6[NO_X]³ − 5.0720 × 10^− 4[NO_X]² + 2.5322 × 10^− 2[NO_X]

Where [NO₂]/[OX] and [NO_X] are in ppb. These are the equations that are used in the modelling for 2020 and are an improvement over the equations presented in Murrells et al. (2008) because the background equation requires a smaller calibration and the roadside equation enables the use of a linear calibration for the roadside model. The equations for the Scottish maps are the same as are used for the UK PCM model for 2020 but the background and roadside calibration coefficients for NO₂ are specific to the Scottish maps.

Detailed description of how the NO₂ modelling approach is applied can be found in the 2020 UK modelling report (Pugsley et al. 2022), this includes:

• Derivation of the value for regional oxidant ([OX]_B) for the UK
• Calculation of the local oxidant (fNO₂.[NO_X]) at background and roadside locations
• Calculation of f(NO_X) (i.e. [NO₂]/[OX]) in the PCM model
• How the updated oxidant-partitioning model has been applied in the UK to background and roadside locations

The resultant maps of background and roadside annual mean NO₂ concentrations for Scotland for 2020 are shown in Figure 23 and Figure 24, respectively. In terms of roadside locations, the roadside concentrations included in this report correspond only to the urban major road links (A-roads and motorways) that are reported as part of the annual compliance assessment for the UK as a whole.

3.3.1 Annual mean background PM₁₀ and contributions from local area sources

The dispersion kernels applied in the background NO_X modelling (as described in Section 3.2), have also been applied to calculate the contribution of local area source emissions to ambient PM₁₀ concentrations for the year 2020. This provided the uncalibrated modelled Scotland-specific PM₁₀ area source contribution.

The modelled PM₁₀ contributions from point sources, combustion in industry area sources, secondary inorganic aerosol (SIA), secondary organic aerosol (SOA), iron and calcium-rich dust, long range transport (LRT) of primary PM₁₀, sea salt and a residual contribution have been subtracted from the measured annual mean PM₁₀ concentration at background sites, to derive a “Corrected” measured annual mean PM₁₀ for calibration purposes. The uncalibrated modelled local area source contribution to ambient PM₁₀ concentrations was compared to these “Corrected” measurements from nine background PM₁₀ monitoring sites within the Scottish network to calibrate the model.

Figure 7 shows the calibration plot for the Scotland-specific PM₁₀ local area source model. The gradient of the line of best fit forced through the origin was 0.7945, and this value was used as the Scotland-specific PM₁₀ area source calibration coefficient. The regression line has been forced through the origin to provide a reasonable model output without imposing an unrealistically high residual on the area source contribution. For each grid square in Scotland the uncalibrated modelled local area source contribution was multiplied by the Scotland-specific calibration coefficient to calculate the calibrated area source contribution. The calibrated area source contribution was then added to the contributions from combustion in industry area sources, from secondary organic and inorganic particles, from point sources, from regional primary particles, from sea salt, from calcium and iron rich dusts and the residual, resulting in a map of total background annual mean PM₁₀ concentrations for Scotland for 2020 as shown in Figure 25. In 2020 the residual was set to zero in the total background annual mean PM₁₀ concentration. This value of the residual was found to provide the best fit to the monitoring data for PM₁₀ in 2020, consistent with that used in the 2020 UK mapping. A detailed description of the background modelling approach for PM₁₀ can be found in Section 6 of the 2020 UK PCM modelling report (Pugsley et al. 2022).

Figure 7: Calibration of the Scotland-specific background PM₁₀ model for the year 2020.

As noted above, background area source PM₁₀ model calibration coefficients applied in the Scotland-specific and UK PCM models are compared in Figure 8. This presents the calibration coefficients from 2008 to 2020 and highlights:

1. the year-to-year variation in the Scotland-specific background area source PM₁₀ model calibration coefficient;
2. the extent to which these values vary in both models.

From Figure 8 it can be seen that the 2020 Scotland-specific background area source PM₁₀ calibration coefficient was less than the equivalent value used in the UK PCM model in 2020 due to the different mix of sites included in the UK and Scotland-specific PCM model calibrations.

Figure 8: Background model calibration coefficients applied in the Scotland-specific and UK models for PM₁₀ for 2008 to 2020.

Since 2016, the Scotland-specific and UK calibration coefficients were lower than in earlier years and closer to 1. This is primarily a result of updates that improved the representation of two components of the model in 2016:

• Secondary Inorganic Aerosol (SIA): Monthly measurements are available for 28 rural monitoring sites within the UKEAP AGANet and NAMN networks and the measurments are interpolated to provide maps of SIA concentrations across the UK. The measurement method used within the AGANet was changed at the beginning of 2016. This revised method typically results in higher measured concentrations of sulphate and nitrate (Tang et al. (2015)).
• Domestic wood combustion: The methods used to estimate the spatial distribution of emissions from domestic wood combustion were revised for the NAEI 2015 to incorporate new information from a survey of domestic wood use (BEIS (2016)). The revised spatial distribution places a larger proportion of these emissions in large urban areas than in previous assessments.

3.3.2 Roadside PM₁₀ concentrations

As for NO_X, it is assumed that the annual mean PM₁₀ concentration at roadside locations is made up of two components: the background concentration as described above (PM_{10, background}) and a local roadside increment (PM_{10, roadside − increment}):

• PM_{10, roadside} = PM_{10, background} + PM_{10, roadside − increment}

The NAEI has provided estimates of PM₁₀ emissions for major road links in the UK for 2019 (Churchill et al. 2022). In 2020 the roadside increment was calculated using the PCM Roads Kernel Model (PCM-RKM) which was introduced into the UK PCM model in 2014. The PCM-RKM uses the ADMS-Roads 5.0 dispersion model to calculate the roadside increment to PM₁₀ concentrations on urban major roads. Individual model runs were carried out for 950 census points covering urban major roads in Scotland. For a short description of the application of the roadside increment modelling approach using PCM-RKM for Scotland-specific modelling see Section 3.2, and for a detailed description see Appendix 8 of the 2020 UK modelling report (Pugsley et al. 2022).

The PCM-RKM modelled roadside increment is calibrated by comparison with an empirically calculated roadside increment (measured roadside PM₁₀ concentration minus modelled background PM₁₀ concentration) at Scottish roadside or kerbside monitoring sites. Calibration of the PCM-RKM is shown in Figure 9. The background PM₁₀ component at these roadside monitoring sites is taken from the background PM₁₀ map described in the previous section. The calibration coefficient derived is then applied to the roadside increment calculated using the PCM-RKM for each road link. The background PM₁₀ is added to the calibrated roadside increment, resulting in a map of total roadside annual mean PM₁₀ concentrations for Scotland for 2020. It should be noted that the roadside concentrations included in this report correspond only to the urban major road links (A-roads and motorways) that are reported as part of the annual compliance assessment for the UK as a whole.

Of the sixty eight roadside PM₁₀ monitoring sites within the Scottish network, thirty five had sufficient data capture for PM₁₀ in 2020 and were located on modelled major road links close to census points at which the traffic flow along the road was measured. The roadside PM₁₀ measurements from thirteen of these sites were used to calibrate the model. Scottish PM₁₀ air quality monitoring data from twenty two roadside sites were used to verify the model output.

A total of eight roadside PM₁₀ monitoring sites were excluded on the basis of low annual data capture (annual percentage data capture <75%). A further nineteen roadside sites, with sufficient data capture, were excluded because they were not located on a modelled urban major road link. A total of six roadside sites were excluded because they were not located on a modelled urban major road link, and also had low data capture. The air quality monitoring sites outside of the AURN are typically installed at kerbside or roadside for AQMA purposes and may not be located on modelled major road links.

Figure 9: Calibration of the Scotland-specific roadside increment PM₁₀ model for the year 2020.

Figure 10 provides a comparison of the 2020 Scotland-specific and UK roadside PM₁₀ calibration coefficients. The Scotland-specific and UK roadside PM₁₀ calibration coefficients from 2014 are also presented. From Figure 10 it can be seen that the 2020 Scotland-specific roadside PM₁₀ calibration coefficient was less than the value used in the 2020 UK PCM model, which as noted in relation to the background model is due to the different mix of sites included in the UK and Scotland-specific PCM model calibrations.

Figure 10: Roadside model calibration coefficients applied in the Scotland-specific and UK models for PM₁₀ for 2014 to 2020.

3.4.1 Annual mean background PM_2.5 and contributions from local area sources

The dispersion kernels applied in the background NO_X modelling (as described in Section 3.2), have also been applied to calculate the contribution of local area source emissions to ambient PM_2.5 concentrations for the year 2020. This provided the uncalibrated modelled Scotland-specific PM_2.5 area source contribution.

The modelled PM_2.5 contributions from point sources, combustion in industry area sources, secondary inorganic aerosol (SIA), secondary organic aerosol (SOA), iron and calcium-rich dust, long range transport (LRT) of primary PM_2.5, sea salt and a residual contribution have been subtracted from the measured annual mean PM_2.5 concentration at background sites, to derive a “Corrected” measured annual mean PM_2.5 for calibration purposes. The uncalibrated modelled local area source contribution to ambient PM_2.5 concentrations was compared to these “Corrected” measurements from eight background PM_2.5 monitoring sites within the Scottish network to calibrate the model.

Figure 11 shows the calibration plot for the Scotland-specific PM_2.5 local area source model. The gradient of the line of best fit forced through the origin was 1.001, and this value was used as the Scotland-specific PM_2.5 area source calibration coefficient. The regression line has been forced through the origin to provide a reasonable model output without imposing an unrealistically high residual on the area source contribution. For each grid square in Scotland the uncalibrated modelled local area source contribution was multiplied by the Scotland-specific calibration coefficient to calculate the calibrated area source contribution. The calibrated area source contribution was then added to the contributions from combustion in industry area sources, from secondary organic and inorganic particles, from point sources, from regional primary particles, from sea salt, from calcium and iron rich dusts and the residual, resulting in a map of total background annual mean PM_2.5 concentrations for Scotland for 2020 as shown in Figure 27. In 2020 the residual was set to zero in the total background annual mean PM_2.5 concentration. This value of the residual was found to provide the best fit to the monitoring data for PM_2.5 in 2020, consistent with that used in the 2020 UK mapping. A detailed description of the background modelling approach for PM_2.5 can be found in Section 7 of the 2020 UK PCM modelling report (Pugsley et al. 2022).

Figure 11: Calibration of the Scotland-specific background PM_2.5 model for the year 2020.

This is the first year that Scotland-specific modelling of PM_2.5 has been undertaken. As such, a comparison of the 2020 calibration coefficient with previous years cannot be assessed. The Scotland-specific and UK background area source calibration coefficients for 2020 are compared in Table 3. The 2020 Scotland-specific background area source PM_2.5 calibration coefficient was greater than the equivalent value used in the UK PCM model in 2020 due to the different mix of sites included in the UK and Scotland-specific PCM model calibrations.

3.4.2 Roadside PM_2.5 concentrations

As for PM₁₀, it is assumed that the annual mean PM_2.5 concentration at roadside locations is made up of two components: the background concentration as described above (PM_{2.5, background}) and a local roadside increment (PM_{2.5, roadside − increment}):

• PM_{2.5, roadside} = PM_{2.5, background} + PM_{2.5, roadside − increment}

The NAEI has provided estimates of PM_2.5 emissions for major road links in the UK for 2019 (Churchill et al. 2022). In 2020 the roadside increment was calculated using the PCM Roads Kernel Model (PCM-RKM) which was introduced into the UK PCM model in 2014. The PCM-RKM uses the ADMS-Roads 5.0 dispersion model to calculate the roadside increment to PM_2.5 concentrations on urban major roads. Individual model runs were carried out for 950 census points covering urban major roads in Scotland. For a short description of the application of the roadside increment modelling approach using PCM-RKM for Scotland-specific modelling see Section 3.2, and for a detailed description see Appendix 8 of the 2020 UK modelling report (Pugsley et al. 2022).

The PCM-RKM modelled roadside increment is calibrated by comparison with an empirically calculated roadside increment (measured roadside PM_2.5 concentration minus modelled background PM_2.5 concentration) at Scottish roadside or kerbside monitoring sites. Calibration of the PCM-RKM is shown in Figure 12. The background PM_2.5 component at these roadside monitoring sites is taken from the background PM_2.5 map described in the previous section. The calibration coefficient derived is then applied to the roadside increment calculated using the PCM-RKM for each road link. The background PM_2.5 is added to the calibrated roadside increment, resulting in a map of total roadside annual mean PM_2.5 concentrations for Scotland for 2020. It should be noted that the roadside concentrations included in this report correspond only to the urban major road links (A-roads and motorways) that are reported as part of the annual compliance assessment for the UK as a whole.

Of the fifty eight roadside PM_2.5 monitoring sites within the Scottish network, twenty nine had sufficient data capture for PM_2.5 in 2020 and were located on modelled major road links close to census points at which the traffic flow along the road was measured. The roadside PM_2.5 measurements from eleven of these sites were used to calibrate the model. Scottish PM_2.5 air quality monitoring data from eighteen roadside sites were used to verify the model output.

A total of seven roadside PM_2.5 monitoring sites were excluded on the basis of low annual data capture (annual percentage data capture <75%). A further seventeen roadside sites, with sufficient data capture, were excluded because they were not located on a modelled urban major road link. A total of five roadside sites were excluded because they were not located on a modelled urban major road link, and also had low data capture. The air quality monitoring sites outside of the AURN are typically installed at kerbside or roadside for AQMA purposes and may not be located on modelled major road links.

Figure 12: Calibration of the Scotland-specific roadside increment PM_2.5 model for the year 2020.

This is the first year that Scotland-specific modelling of PM_2.5 has been undertaken. As such, a comparison of the 2020 calibration coefficient with previous years cannot be assessed. The Scotland-specific and UK roadside calibration coefficients for 2020 are compared in Table 4. The 2020 Scotland-specific roadside PM_2.5 calibration coefficient was less than the value used in the 2020 UK PCM model, which as noted in relation to the background model is due to the different mix of sites included in the UK and Scotland-specific PCM model calibrations.