The UK particulate matter air pollution episode of March–April 2014: more than Saharan dust

An air pollution event of elevated surface concentrations of particulate matter (PM) occurred in the UK between 26 March and 8 April, 2014. Observations from UK air quality monitoring networks showed markedly elevated PM surface concentrations across the majority of the country during this time. The UK Automatic Urban and Rural monitoring Network (AURN) recorded hourly values of PM₁₀ of up to 100 μg m⁻³, well above the hourly concentrations of a few tens μg m⁻³ normally recorded; for example across 45 urban background AURN sites the median, inter-quartile range and 95th percentile of hourly PM₁₀ concentrations in 2014 were, respectively, 38, 36–44 and 53 μg m⁻³ (http://uk-air.defra.gov.uk).

The spring 2014 PM episode was widely perceived as being associated with an enhanced surface concentration of mineral dust from the Sahara (e.g. as reported by the BBC⁶ , the Guardian⁷ , and the Daily Mail⁸ ). An implication is that this was, at least in part, a 'natural' event. However, it is known that the UK's geographic proximity to industrialised countries in Western Europe (e.g. Germany, France and the Netherlands) make it vulnerable to the import of anthropogenic primary and secondary gaseous pollutants and PM . Previous analyses have highlighted this issue for ozone (Vieno et al 2010, Francis et al 2011) and for secondary inorganic components of PM (Abdalmogith and Harrison 2005, Jones and Harrison 2006, Yin and Harrison 2008, Charron et al 2013, Vieno et al 2014). For effective policy action to mitigate the impacts of PM on human health it is essential to understand the composition of the atmosphere and the causes of high levels of PM (e.g. Heal et al 2012), of which this episode was an example. Of the studies investigating the origins of elevated PM cited above, all except Vieno et al (2014) were based on analysis of measurements of PM chemical components at Harwell or Birmingham in south-central UK. The key feature of the current work was the combination of speciated PM measurements, which are only sparsely undertaken in the UK (as elsewhere), and a regional 3D Eulerian atmospheric chemistry transport model (ACTM) that simulates to 5 km horizontal resolution over the region of study, to assess the validity of attributing this PM event to Saharan dust over the entire UK domain.

2.1. Atmospheric chemistry transport modelling

A Eulerian ACTM, the EMEP4UK version rv4.3 (Vieno et al 2010, Simpson et al 2012, Vieno et al 2014, Vieno et al 2016), was used to study the spring 2014 UK air pollution event. The meteorological driver used for EMEP4UK for this purpose was the Weather Research and Forecasting (WRF) model version 3.6. The WRF model included data assimilation (Newtonian nudging) of the numerical weather prediction model meteorological reanalysis from the US National Center for Environmental Prediction/National Center for Atmospheric Research Global Forecast System at 1° resolution, every 6 hours (Saha et al 2011). The model domain covers the whole of Europe at a horizontal resolution of 50 km × 50 km with a higher horizontal resolution nested inner domain over the British Isles at 5 km × 5 km, as shown in figure 1.

Zoom In Zoom Out Reset image size

A full description of the EMEP4UK model can be found in Vieno et al (2014) and Simpson et al (2012). In addition to gaseous components, the EMEP4UK model simulates the concentrations (at hourly time resolution) of both PM₁₀ (particles with aerodynamic diameter <10 μm) and PM_2.5 (particles with aerodynamic diameter <2.5 μm), the two PM size fractions forming part of EU air quality legislation (Heal et al 2012). The chemical scheme used for the present study is the EMEP-EmChem09 (Simpson et al 2012, www.emep.int) and the gas/aerosol partitioning was the model for aerosols reacting system (MARS) (Binkowski and Shankar 1995). In the model version used here, PM_2.5 is derived as the sum of the fine fraction of: ammonium (${{{\rm{NH}}}_{4}}^{+}),$ sulphate (${{{\rm{SO}}}_{4}}^{2-}),$ nitrate (${{{\rm{NO}}}_{3}}^{-}),$ elemental carbon (EC), organic matter (OM), sea salt (SS), mineral dust, and 27% of the coarse nitrate. PM₁₀ is derived as the sum of PM_2.5 plus the coarse fraction of EC, OM, ${{{\rm{NO}}}_{3}}^{-},$ SS, and dust. In the current version of the EMEP4UK model, the split between PM_2.5 and PM_2.5–10 for ${{{\rm{NO}}}_{3}}^{-}$ is somewhat uncertain as discussed in Aas et al (2012). Fine ${{{\rm{NO}}}_{3}}^{-}$ production is modelled using the MARS thermodynamic model and the coarse ${{{\rm{NO}}}_{3}}^{-}$ formation in the current model is a function of humidity and HNO₃ using a parametrization derived from Riemer et al (2003), as described in Simpson et al (2012), but it is not explicitly linked to the surface area of the existing coarse aerosol. Both nitrate generation mechanisms compete for the same HNO₃ therefore constraining the total amount of nitrate produced.

Anthropogenic emissions of NO_x, NH₃, SO₂, primary PM_2.5, primary coarse PM, CO and non-methane VOC for the UK are derived from the National Atmospheric Emission Inventory (NAEI, http://naei.defra.gov.uk) aggregated to 5 km × 5 km resolution. The EMEP 50 km × 50 km resolution emission estimates provided by the Centre for Emission Inventories and Projections (CEIP, www.ceip.at/) are used for the European domain. Shipping emissions estimates, for the inner domain, are derived from ENTEC (2010).

The boundary conditions at the edge of the European domain (figure 1) are prescribed concentrations adjusted for each simulated year as described in Simpson et al (2012). The EMEP4UK model version used here uses the default climatological Saharan dust boundary conditions, based on monthly average dust concentrations for the year 2000 from the global CTM2 at the University of Oslo. This treatment may lead to an inaccurate representation of the source strength of Saharan dust for individual events. The atmospheric flow and deposition processes transporting the dust and controlling its distribution over Europe and the UK during the simulated period are fully represented within the model. For ozone, a 3D field from climatological ozone-sonde data-sets, modified monthly against clean-air surface observations was used as initial and boundary condition for the European domain.

Two EMEP4UK model simulations were carried out to quantify the contributions of UK and regional emissions to the episode of elevated surface PM: (1) A base-case run, called 'BASE' hereafter, and (2) a no-UK anthropogenic emissions run, called 'NoUK' hereafter. The difference between the BASE and the NoUK simulations therefore represents a quantitative measure of the enhancement of the UK PM concentration that is caused by domestic emissions over the European background. The percentage fraction of PM generated by UK emissions (f_UK) was calculated as:

$$\begin{eqnarray}&&{f}_{{\rm{UK}}}=100\times ({\rm{BASE}}-{\rm{NoUK}})/{\rm{BASE}}.\end{eqnarray} \tag{ 1 }$$

2.2. Measurements

The dominant meteorological feature for the duration of the PM pollution event was an area of high pressure located over Europe, with relatively low surface temperatures of 8 °C–14 °C. This limited the evaporation of ammonium nitrate into ammonia and nitric acid, enhancing the lifetime of particle nitrate and the potential influence of its long-range transport.

Figures 2 and 3 compare the modelled and measured daily mean surface concentrations of PM_2.5 and PM₁₀, respectively, between 20th March and 13th April 2014 at the four selected AURN sites. All four sites experienced elevated PM during this period when compared with their 2014 annual averages, which for PM_2.5 and PM₁₀ were: 9 and 14 μg m⁻³ (Harwell), 15 and 19 μg m⁻³ (London Bloomsbury), 10 and 15 μg m⁻³ (Aberdeen) and 7 and 8 μg m⁻³ (Auchencorth Moss) (uk-air.defra.gov.uk). The correlation coefficients between model and measurements over the simulation period are greater than 0.77 for all four sites included in this work for both PM_2.5 and PM₁₀ (figures 2 and 3).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Two distinct episodes are apparent in the PM_2.5 and PM₁₀ concentrations: one centred on the 30th March (referred to subsequently as episode 1: 'EP1') with an approximate duration of 5 days, and the second centred on the 3rd April ('EP2') with an approximate duration of 4 days. The spatial distributions across the UK and the rest of Western Europe of the daily mean surface PM during these episodes are shown in figures 4 and 5 for PM_2.5 and PM₁₀, respectively.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figures 2 and 3 show that during the full period presented, PM_2.5 and PM₁₀ were dominated by secondary inorganic PM components, especially for PM_2.5. The estimated proportions of daily-mean PM_2.5 and PM₁₀ derived from UK emissions are shown in figures 6(a) and 7(a), respectively. The proportions were typically less than 40% from which it can be inferred that EP1 and EP2 both have a large component of PM derived from long-range transport of secondary inorganic PM and secondary PM precursors. In contrast, on the 1st April, a date between the two episodes, up to 70% of PM_2.5 and PM₁₀ were estimated to have been generated from UK emissions. The contribution of transboundary transport to PM pollution events in the UK has been reported before; for example, Vieno et al (2014) showed that up to 80% of particle ${{{\rm{NO}}}_{3}}^{-}$ in episodes may be imported into the UK from continental Europe.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To further test the model, figure 8 shows the hourly modelled inorganic composition of PM₁₀ at Auchencorth Moss and at Harwell over the full time period compared against the MARGA measurements (uk-air.defra.gov.uk). Both the speciated MARGA measurements and the model follow the TEOM-FDMS PM₁₀ temporal variation shown in figure 3 relatively well. Gaps in the measurements indicate when the MARGA instrument was offline. As for total daily PM_2.5 and PM₁₀, the hourly observations of speciated PM₁₀ components during EP1 and EP2 are well captured by the model, both in timing and magnitude. The notable enhancement of ${{{\rm{NO}}}_{3}}^{-}$ concentrations during this event is clearly evident in both the MARGA observations and in the model simulations as shown in figure 8. The fact that the ${{{\rm{NO}}}_{3}}^{-}$ enhancement was accompanied by an equivalent enhancement in ${{{\rm{NH}}}_{4}}^{+}$ supports the conclusion that observed PM_2.5 enhancements are minimally impacted by dust or by ${{{\rm{NO}}}_{3}}^{-}$ formation on dust surfaces.

Zoom In Zoom Out Reset image size

With respect to the Saharan dust component, the modelling indicates that the PM_2.5 and PM₁₀ composition differed between EP1 and EP2. Figures 6(b) and 7(b) show that Saharan dust constituted a greater component of both surface PM_2.5 and PM₁₀ during EP2 than during EP1. However, the dust was present at substantial surface concentrations only in the southern UK, as illustrated by the data presented here for the representative southern UK monitoring sites at Harwell and London Bloomsbury. Overall, the percentage of modelled Saharan dust in PM_2.5 and PM₁₀ is less than 20% for most of the UK for both EP1 and EP2. The geographic extent of the Saharan dust incursion to the surface (relative dust contribution to PM > 20%) for EP2 is shown in figures 6(b) and 7(b) for PM_2.5 and PM₁₀, respectively. During EP1 the high concentrations of PM in the plume were derived from anthropogenic emissions in continental Europe and advected to the UK as shown in figures 4 and 5. It was only during EP2 that the Saharan dust plume was also present at the surface (for parts of the UK) augmenting the secondary PM from continental Europe. The Saharan dust plume can be observed having evolved and moving north across Europe from the 30th March, arriving at the UK surface level on the 3rd of April (figure 2). Saharan dust in the atmosphere was also predicted by the model during EP1, but at higher altitudes with little impact at the surface. To visualise this altitudinal variation in PM composition between the two episodes, a 3D iso-surface of ∼5 μg m⁻³ at the surface for ${{{\rm{NO}}}_{3}}^{-}$ (green) and Saharan dust (yellow) is shown for 12:00 on 30th March during EP1 (figure 9(b)) and for 3rd April during EP2 (figure 10(b)). Figures 9(a) and 10(a) illustrate the geographic extent of when Saharan dust and ${{{\rm{NO}}}_{3}}^{-}$ concentrations at the surface were greater than 1 μg m⁻³ for these two occasions. Figure 9 shows that during EP1 there was a low-altitude ${{{\rm{NO}}}_{3}}^{-}$ plume advected in from continental Europe and a separate high altitude plume of Saharan dust. The model simulation shows that during EP1 the high altitude Saharan dust did not reach the UK surface apart from a small area in Wales as shown in figure 9(a). In contrast, figure 10 shows that for EP2 the Saharan dust was present at lower altitudes and had reached the surface in the south of the UK on the 3rd of April (the dust plume had also reached the surface in the UK on 2nd April).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

From the analysis of these two PM events, which although temporally adjacent have distinct characteristics due to their air mass history, it has been shown that speciation measurements and modelling can allow the clear, quantitative diagnosis of the contributions to PM mass. In this case, the MARGA measurements characterised the inorganic composition and the modelling demonstrated how that composition developed. These measurements do not provide explicit information about the external/internal mixing properties of the PM, e.g. interactions of the secondary PM on the surfaces of mineral dust, which is likely to be occurring as noted in previous studies (Fairlie et al 2010). However, since in the EMEP4UK model the production of NH₄NO₃ is not linked to the additional particle surface area provided by the dust, the model's ability to reproduce measured concentration of these compounds suggests that the importance of the Saharan dust promoting secondary aerosol formation is not a key process for this event.

The modelled low levels of Saharan dust in EP1 (<20% of PM mass) coupled with the PM measurements indicate that the UK did not experience the consequences of a Saharan dust episode at that time. Rather, this was mainly an anthropogenic pollution event, exacerbated by the transport of secondary inorganic aerosol precursors emitted outside of the UK, which, in principle, could be mitigated against in future with appropriate international policies to reduce primary air pollutant emissions (especially NO_x and NH₃). The importance of long-range transport of ammonium nitrate demonstrated here for this UK PM_2.5 episode corroborates previous work (Vieno et al 2014) and confirms similar findings on a European scale presented by Kiesewetter et al (2015). The policy implications derived from accurate identification of emission sources contributing to PM concentrations are discussed further in the next section.

The findings presented in this paper illustrate the capability of advanced ACTMs to reliably identify the underlying causes of complex air pollution episodes. The validation of the model results against chemically-speciated measurements of PM components further highlights the need for the development and implementation of better integrated model and measurement systems to routinely apportion sources and target the most relevant contributing activities for the design of effective pollution control policies (Reis et al 2015). It is also clear that it is feasible to provide robust information on the state of atmospheric composition as an evidence base to policy makers in a more operational manner.

The substantial contribution of long-range transport to ambient concentrations of fine PM in the UK, and the accurate identification of the relevant sources, has far-reaching policy implications. The pollution episode in spring 2014 was widely attributed to Saharan dust in the UK media, thus placing a (false) emphasis on a natural phenomenon, which cannot be addressed by policy action. The focus on an 'exciting headline' may be attributable to the complexity of communicating atmospheric processes and the public observations of dust-fall on surfaces. This suggests that while scientific analyses to determine the real causes of particular PM events through measurements and modelling are well established, the translation into communicating outside the atmospheric science arena needs improvement. Therefore, there is a danger that important messages to high-level policy-makers may be obscured by attention-catching headlines.

The initial mischaracterisation of the PM event may also represent a missed opportunity to inform and educate the general public about the role of anthropogenic emissions, specifically agricultural emissions of ammonia, as a key contributor to many high PM pollution events in recent years. As a consequence, the lack of public awareness may result in reduced public interest in, and a weaker policy mandate for, any emissions reduction targets for ammonia emissions and their contribution to transboundary fluxes of secondary aerosol components. The importance of public support has been illustrated by Reis et al (2012) specifically for the implementation of air pollution control policies through a robust science-policy interface. The extent to which explicit value judgements are a further underlying cause for the—to date—only moderate ambition levels in reducing ammonia emissions on a European scale (Voinov et al 2014) is difficult to quantify.

These implications are not only relevant for Europe. The PM pollution episodes in China and other fast growing economies (e.g. Oh et al 2015, Yan et al 2015, Zhang et al 2015, Zhou et al 2015, Tan et al 2016) are more severe examples of those in the UK, Paris, and other parts of Europe (Gualtieri et al 2015). While a major part of urban PM_2.5 in these other regions arises from primary PM_2.5 emissions from local transport and industrial activities, as has also been demonstrated for the UK (Vieno et al 2016), agricultural ammonia emissions have also been identified as key contributors. In these regions also, as for Europe, the need for close collaboration in policy development across provincial or national borders is evident and needs robust modelling and monitoring activities to provide underpinning scientific evidence. Such transboundary collaboration is emerging, e.g. in the case of tackling air quality and ozone pollution in Hong Kong and the Pearl River Delta (Zhong et al 2013). The combination of local measures and regional policies to reduce emissions within and far away from urban areas will likely be more effective in reducing the public health impacts of severe episodes of PM.

The EMEP4UK regional ACTM has been shown to successfully reproduce elevated surface PM concentrations during the spring of 2014 in the UK. The combined analysis of model results and speciated PM measurements indicates that these elevated PM concentrations were mainly driven by ammonium nitrate, much of which was derived from European NO_x and NH₃ emissions outside the UK. The contribution of Saharan dust, which was widely reported in the UK media, has been shown to be restricted (at surface level) to the latter part of the elevated PM episode and regionally to the southern UK. Both types of PM and PM precursors arrived in the UK with the same southerly air flow, but originated from quite independent sources. The analyses presented in this paper illustrate the complementary use of high resolution observations (temporal and chemical speciation) and an advanced ACTM to identify the underlying causes of complex PM air pollution episodes. Specifically, the findings suggest that more emphasis on reducing ammonia emissions in Europe would have marked benefits in reducing episodic PM_2.5 concentrations and population exposure in the UK. The need for close collaboration in policy development across provincial or national borders applies worldwide.

The authors acknowledge the UK Department for Environment, Food and Rural Affairs (Defra) and the Devolved Administrations, through the projects: development of the EMEP4UK model (AQ0727), the 'UK Eutrophying and Acidifying Atmospheric Pollutants (UKEAP) project' (AQ0647) which funds the MARGA observations, as well as access to the AURN data, which were obtained from uk-air.defra.gov.uk and are subject to Crown 2014 copyright, Defra, licenced under the Open Government Licence (OGL). Partial support for the EMEP4UK modelling from the European Commission FP7 ECLAIRE project is gratefully acknowledged.