Air quality study: assessing variations in roadside air quality with sampling height

1 Introduction

1.1 Purpose of the Study

A wide range of air quality monitoring is undertaken in Scotland, primarily by the Scottish Government to fulfil the requirements of EU Directive 2008/50/ EC on ambient air quality and by local authorities under the Local Air Quality Management regime ( LAQM) as set out in the Environment Act 1995 and associated regulations. Other monitoring is undertaken by SEPA, Transport Scotland and various academic, research and commercial organisations.

Although both the Directive and LAQM are focused towards protecting human health, current monitoring strategies fail to consider variations in air quality with height above the ground. Heights of monitoring station sampling points vary depending on local conditions and with the type of equipment installed.

One of the main sources of air pollution in Scotland is road traffic, and most vehicle emissions tend to originate at heights of less than 1 m above the ground. Consequently, in urban environments close to roads, human exposure to air pollution may be influenced by the age and height of the exposed population, with children and pushchairs located closer to vehicle emissions. Current fixed-site monitoring approaches may not adequately reflect any vertical variations in air quality. The application of a mobile monitoring platform simultaneously measuring a number of air pollutants at adult breathing height (1.68 m) and child/pushchair breathing height (0.80m) within an urban environment thus provides the opportunity to guide and inform any future detailed work on the potential impact of air quality on children's health.

The study has two key aims:

• Determine the relationship between height from pavement and air quality under a range of conditions; and
• Investigate the relationship between concurrent air quality sampling obtained from mobile and fixed sampling stations.

The focus of the study is on the following pollutants:

• Particulate matter with a mean aerodynamic diameter of 2.5 m (PM _2.5).
• Particulate matter with a mean aerodynamic diameter of 10 m (PM ₁₀);
• Ultrafine particles ( UFP) between the sizes of 10 nm and 300 nm.
• Black carbon ( BC).

With the following additional pollutants also measured:

• Nitrogen dioxide (NO ₂).
• Sulphur dioxide (SO ₂).
• Carbon Monoxide (CO).
• Ozone (O ₂).
• Carbon Dioxide (CO ₂).
• Benzene (C ₆H ₆).
• Particulate matter with a mean aerodynamic diameter of 0.5 m (PM _0.5);
• Particulate matter with a mean aerodynamic diameter of 1.0 m (PM _1.0);
• Particulate matter with a mean aerodynamic diameter of 5.0 m (PM _5.0);
• Total particulate matter ( TPM).

1.2 Policy Background

The air quality objectives applicable to Local Air Quality Management ( LAQM) in Scotland are set out in the Air Quality (Scotland) Regulations 2000 (Scottish SI 2000 No 97), the Air Quality (Scotland) (Amendment) Regulations 2002 (Scottish SI 2002 No 297), and are shown in Table 1.1. This table shows the objectives in units of micrograms per cubic metre µg m ^-3 (milligrams per cubic metre, mg m ^-3 for carbon monoxide) with the number of exceedances in each year that are permitted (where applicable). For these pollutants, Local Authorities have an obligation to work towards achieving these objectives.

Table 1.1 Air Quality Objectives included in Regulations for the purpose of LAQM in Scotland

Pollutant	Air Quality Objective
	Concentration	Measured as
Benzene	16.25 µg m -3	Running annual mean
	3.25 µg m -3	Running annual mean
1,3-Butadiene	2.25 µg m -3	Running annual mean
Carbon Monoxide	10 mg m -3	Running 8-hour mean
Lead	0.50 µg m -3	Annual mean
	0.25 µg m -3	Annual mean
Nitrogen Dioxide	200 µg m -3, not to be exceeded more than 18 times a year	1-hour mean
	40 µg m -3	Annual mean
Particulate Matter (PM 10) (Gravimetric)	50 µg m -3, not to be exceeded more than 7 times a year	24-hour mean
	18 µg m -3	Annual mean
Sulphur dioxide	350 µg m -3, not to be exceeded more than 24 times a year	1-hour mean
	125 µg m -3, not to be exceeded more than 3 times a year	24-hour mean
	266 µg m -3, not to be exceeded more than 35 times a year	15-minute mean

New air quality objectives for PM _2.5 were adopted in May 2008 under European Directive 2008/50/ EC. The Directive introduced additional PM _2.5 objectives targeting the exposure of the population to fine particles. These objectives are set at the national level and are based on the average exposure indicator ( AEI). The AEI is determined as a 3-year running annual mean PM _2.5 concentration averaged over the selected monitoring stations in agglomerations and larger urban areas, set in urban background locations to best assess the PM _2.5 exposure to the general population.

In line with the stricter PM ₁₀ objectives, the Scottish Government has adopted a draft national air quality objective of 12 g m ^-3 as an annual mean and is currently considering adopting the more stringent WHO's Guideline Value of 10 g m ^-3 as an annual mean. The PM _2.5 objective has not yet been incorporated into LAQM Regulations and therefore Local Authorities are not currently required to monitor concentrations of PM _2.5. PM _2.5 is currently monitored at 6 Automatic Urban and Rural Network monitoring sites throughout Scotland located at:

• Aberdeen
• Auchencorth Moss
• Edinburgh St Leonards
• Glasgow Kerbside
• Grangemouth
• Inverness

A Consultation ^[1] on the existing LAQM regime within Scotland was carried out by the Scottish Government during 2014. The role of Local Authorities in monitoring PM _2.5 and the current monitoring programme for PM _2.5 may therefore change following the publication of the conclusions of the review. Black carbon and ultra-fine particles ( UFP) are not covered under the current EU or UK air quality legislation but are currently the focus of significant international research regarding their potential impacts on health and the environment.

Scottish, UK and European policy on assessing exposure to air pollution is currently geared towards fixed location monitoring "in the breathing zone", typically accepted to be up to about 4 m above the ground. In practical terms, it is usually extremely challenging to measure at heights much below 1.5 m, because of analyser infrastructure requirements and the risk of vandalism. For multi-pollutant sites, the analysers are generally housed in walk-in enclosures, which further hampers the ability to measure air quality at reduced heights.

A key policy driver for LAQM and EC Directive compliance is to ensure that measurements are made using methods that fulfil specified Data Quality Objectives ( DQO). This ensures that measurements across regions and member states are comparable and of sufficient quality to enable robust assessment of compliance with standards and objectives. Portable analysers have not yet demonstrated compliance or equivalence to these requirements, so are not currently used to determine or inform policy in this way. The weakness in the current approach to fulfilling statutory obligations for air quality monitoring and assessment is that it may not provide a representative measure of personal exposure and therefore enable accurate assessment of the likely health impact.

This study provides a cutting edge assessment of a situation that falls outside of current statutory monitoring policy; measurement of vertical concentration gradients in a moving and variable environment is not currently mandated in EC Air Quality Directives or LAQM. However, the European AQ monitoring community is currently extremely interested in micro- scale monitoring through initiatives like AirMonTech ( http://www.airmontech.eu/) which are part of the ongoing EC Directive Review. This study contributes valuable insight into personal exposure and health impact through the use of portable sampling devices and profiling spatial AQ concentrations both horizontally and vertically.

1.3 Conclusions Drawn from Literature Review

As part of this project a literature review was undertaken. For the review, the online scientific database ScienceDirect was utilised together with wider internet and literature searches. Three main search criteria were used, which provided a targeted examination in the first instance. Further refinement of the search criteria was made by specifying other sub-criteria. For example, it was found while searching for air quality versus height studies that street canyon studies were also informative to the study. As a result, a map of relevant search criteria was developed and this is summarised briefly below:

• Air quality versus height.
  + Urban air quality.
  + Street canyons.
  + Pollutants measured.
  + Samplers used.
  + Data processing/visualisation.

• Mobile monitoring studies.
  + Urban air quality.
  + Street canyons.
  + Pollutants measured.
  + Samplers used.
  + Data processing/visualisation.

• Personal exposure studies.
  + Pollutants measured.
  + Samplers used.
  + Data processing/visualisation

1.3.1 Air Quality versus Height Studies

One of the principal objectives of the study was to determine the nature of the relationship between pollutant concentrations and height; in this case, 2 exposure heights, average child and adult breathing heights. Of the published studies that have investigated vertical profiles of pollutants, the majority of these have focussed on heights of greater than 2 m and have reported a range of conclusions. Trompetter et al (2013) and Ferrero et al (2011) explored the variation in concentrations of BC at heights of up to 100 m and 500 m, respectively. By measuring the vertical profile of BC the studies found that concentrations decreased with height and were mostly confined to heights of less than 50 m.

Imhof et al (2005) measured aerosol particle number, surface area (particles with a diameter of between 30 nm to 10 m) and NOx upwind and downwind of a motorway at heights of 5 m to 50 m, together with wind speed and direction. In this case it was found that particle numbers, surface area and NOx concentrations decreased with height at downwind locations at times before noon, thus supporting the general findings of Trompetter et al, 2013. However after noon, it was found that the maximum concentration of particles was measured at 10 m, indicating an increase in particle concentrations between 5 m and 10 m. Vertical profiles of the pollutants at upwind locations showed concentrations were constant throughout the sampling height range. This research suggests that meteorology and sampling location relative to emission sources will have a direct influence on the vertical profile of pollutants. The study undertaken by Imhof et al (2005) was carried out using fixed monitoring points and therefore the pollutant concentrations measured do not necessarily reflect personal exposure.

Further studies looking at particle concentrations at heights closer to ground level indicate that pollutant concentrations increase with height for the first few meters from ground level. Meilu et al (2011) investigated horizontal and vertical dispersion of particles emitted from freeway vehicles. Three monitoring sites located on flat terrain 15, 50 and 100 m from the road measured particle numbers at 9 sampling heights between 0 - 10 m. It was found that at low wind speeds <1 m s ^-1, the particle number concentrations were unchanged up to approximately 7.7 m above ground level. For wind speeds of >1 m s ^-1, particle number concentrations increased to a maximum at approximately 3.4 m above ground level. Again it was found that wind speed and direction relative to the emission source had a direct influence on the vertical profile of pollutant concentrations. However, this monitoring study was also carried out using fixed monitoring points and therefore the pollutant concentrations measured may not necessarily reflect personal exposure.

Local topography can also play an important role in pollutant dispersion and in street canyons are a common feature in many urban environments. Micallef and Colls (1998) investigated vertical profiles of suspended particulate matter (PM) within a street canyon at six heights between 0.35 and 2.88 m ground level. It was found that daily average concentrations of PM ₁₀ and PM _2.5 were 35% and 12% greater at 0.81 m than at 2.88 m, respectively. This monitoring study was carried out at a fixed monitoring location and therefore does not necessarily reflect personal exposure. However, the study again indicates that a concentration gradient does exist within the first 3 m from ground level.

In summary, it has been shown that research has been carried out investigating pollutant concentrations at a variety of heights. However, this review has identified no research that has been undertaken to investigate both pollutant concentrations at heights of less than 2 m utilising a mobile monitoring platform. From the findings of the literature review, it was concluded that Glasgow mobile monitoring study provided the opportunity to help inform future air quality monitoring techniques and also assess the potential differential impact of urban air pollution on children and adults respectively.

1.3.2 Mobile Monitoring Studies

Mobile air quality monitoring provides data on the spatial and temporal variability of air pollutants. The Glasgow study investigated mobile air quality monitoring versus the current system which uses fixed monitoring sites to assess population exposure. In this section we look at the general approach employed for mobile monitoring studies carried out to date.

Through the review it was found that mobile monitoring campaigns take the form of monitoring whilst on the move around a predetermined route(s). Of the mobile monitoring studies reviewed for this report, it was noted that the authors used either portable monitoring equipment carried by a person (Poppel et al, 2013; Leonard et al, 2012; Pirjola et al, 2012; Cambridge Urban Mobile Sensing (CambMobSens), 2010) or a mobile laboratory to carry out the measurements with monitoring equipment installed within a vehicle (Martinez et al, 2012; Hu et al, 2012; Westerdhal et al, 2005). The obvious necessity for all studies was that GPS was used in combination with the mobile monitoring.

Poppel et al (2013) proposed a methodology for the setup and data processing for a mobile monitoring campaign within an urban environment. The study focussed on UFP, BC and PM _2.5 measurements using a TSI P-Trak at 1 s resolution, Magee Scientific AE51 Micro-Aethelometer at 1 s resolution, and a GRIMM dust monitor at 6 s resolution, respectively. The sampling equipment together with GPS was installed on a modified bicycle, The Aeroflex (Berghmans et al, 2009), and a fixed sampling route was repeated 20 times in 10 days (all weekdays). For analysis, the route was split into six zones based on traffic characteristics e.g. major road, green space, city centre road. This reinforces the approach that was developed for the Glasgow study; using a predetermined route, carrying out repeat runs over a number of separate days and categorise areas/roads within the sample route.

When data processing was considered, it was identified in the Poppel et al (2013) study that the BC data collected using AE51 micro-aethelometer suffered from 'data noise' especially at low concentrations. In the study, an optimized noise-reduction averaging ( ONA) algorithm was used (Hagler et al, 2011) in post-processing. The algorithm reduces noise by increasing the time-averaging window at low measured BC concentrations and decreasing the time-averaging window at high measured BC concentrations. The data resolution used for the AE51 BC monitor in the Glasgow study was 1 minute and consequently 'data noise' did not pose the same problems.

In terms of mobile laboratory-based monitoring, a wider range of pollutants can be measured. For example, Hu et al (2012) investigated the spatial distribution of UFP, PM _2.5, BC, particle-bound polycyclic aromatic hydrocarbons ( PAH), carbon monoxide (CO), CO ₂, NOx combined with fixed meteorological measurements and traffic conditions. All pollutants were sampled at a single height. Again, a predetermined route was sampled twice daily during 11 weekdays further validating the approach used for the Glasgow study.

Pirjola et al (2012) utilised a mobile laboratory to investigate UFP, BC, PM _2.5 and NOx in combination with rooftop meteorological measurements. In the study, three street canyons of differing widths, lengths and orientations were selected to carry out the mobile sampling. Both mobile sampling and fixed sampling were carried out to investigate pollutant concentrations at a number of distances from the road and on upwind and downwind sides of the street canyons. The monitoring data from the mobile laboratory were also compared to that from a fixed roadside air quality monitoring site located on the same road as the three canyons. Again, the sampling runs were carried out two to three times daily during a two week period. It was found that pollutant concentrations were greater on the upwind sides of the canyons than the downwind side. In addition, higher concentrations of UFP were measured during the mobile monitoring runs than at the fixed monitoring site.

In summary, the general methodology for carrying out mobile monitoring is consistent, whether using portable equipment or mobile laboratories. All studies that have been reviewed used at least two sample runs daily around a predetermined route. It has also been shown that fixed meteorological measurements are generally recorded. It was therefore deemed appropriate to collect similar fixed meteorological data as part of the Glasgow Study, in addition to the mobile measurements that were taken.

This review indicated that all of the reviewed mobile monitoring studies undertaken to date have only sampled at one height and there are no studies that have combined mobile monitoring with sampling at two or more heights.

1.3.3 Personal Exposure Studies

Personal exposure can be defined as exposure of individuals as experienced during their real-life, day-to-day activities and as such take into account indoor and outdoor air quality. Where personal exposure studies and the Glasgow study converge is with the monitoring equipment used, the mobile monitoring aspect and the pollutants measured. The Glasgow study provides data mirroring the personal exposure of individuals (children and adults) to UFP, BC and PM _2.5 in Glasgow City Centre. Measurements of these pollutants were carried out using the portable Philips NanoTracer ( UFP), Magee Scientific AE51 micro-aethelometer ( BC), the Harvard-PEMs (gravimetric PM _2.5) and the Lighthouse IAQ 3016 (PM).

A number of recent personal exposure studies have been carried out using the AE51 microaethelometer and the Philips NanoTracer (Buonanno et al, 2013 and 2012; Dons et al, 2012 and 2011) combined with GPS. In these studies, individuals carried the samplers throughout a typical day of activity. Buonanno et al (2013 and 2012) investigated personal exposure of adults and children to UFP and BC. The studies classified a number of microenvironments as school, indoor, outdoor and transport. It was found that UFP and BC exposure varied depending on the activity and microenvironment the individual was in with the highest exposure measured indoors.

A studies by Dons et al (2012 and 2011) these findings were confirmed, however, a further emphasis was carried out in the Dons et al (2012) study where the focus was on personal exposure to BC in transport microenvironments e.g. car passenger, car driver, on foot, bus and train. Again, it was found that the BC concentrations individuals were exposed to were dependent on the transport microenvironment; with the highest concentrations measured for the car driver.

Although these studies are not directly comparable to the Glasgow study, they demonstrate that the Philips NanoTracer and AE51 Micro-aethelometer have been successfully deployed in a mobile monitoring regime. It is also reasonable to hypothesise, using these studies and the research discussed in Sections 1.3.2 and 1.3.1 that the pollutant concentrations measured during the Glasgow study may be dependent on the microenvironments that exist within Glasgow City Centre e.g. street canyons of varying orientations, pedestrian areas and open street environments. Again, the review of personal exposure of air quality research indicates that no studies have been carried out that combine both personal exposure and sampling at multiple heights.

1.3.4 'Other' Relevant Findings

A further useful criteria investigated during the literature review was the processing and visualisation of mobile air quality data. It was found that data processing is dependent upon the type of sampler used, for example the AE51 micro-aethelometer may suffer from 'data noise', as discussed in Section 1.3.2 (Poppel et al, 2013). Hudda et al (2013) also mentions the AE51 sampler, suggesting the micro-aethelometer also suffers from interference from vibrations and mechanical shocks in the form of data spikes. However, it was concluded that these spurious data are easily distinguishable from 'normal' data and can be removed from the dataset during processing; in the case of the Hudda et al (2013) study, < 3% of data were rejected. Such studies provided useful information and tools for the processing AE51 micro-aethelometer data.

The Glasgow study used time-stamped video recording alongside pollutant measurements. This provided information on the possible causes of episodes of high pollutant concentrations that are likely to be encountered during each sample run. Kaur et al, 2006 used video recording in conjunction with measured UFP mobile monitoring time-series data to investigate personal exposure in a variety of microenvironments.

In terms of data analyses, there are a number of approaches that have been identified. Box plots have been used in a number of studies (e.g. Buonanno et al, 2013 and 2012; Dons et al, 2012 and 2011; Hu et al, 2012). These plots provide information of the spread of pollutant concentrations measured during a specified time period. Due to the large amount of data collected during the Glasgow study, box plots are used to compare mobile and fixed monitoring results.

For direct comparison of different pollutants e.g. UFP and BC, regression analysis will be a useful. Westerdhal et al (2005) used a scatter plot matrix to visualise the correlation between 7 pollutants. Similar analysis has been carried out in this study with orthogonal regression analysis being used to investigate the relationship between pollutant concentrations measured at child and adult height.

GPS combined with pollutant measurements have provided the opportunity to carry out GIS-based analyses of data. Martinez et al (2012) and Pirjola et al (2012) used a GIS based visualisation to show UFP number concentrations change throughout a mobile sampling route. In terms of this study, this helped with the identification/classification of different microenvironments. As discussed in Section 3.2.1, vertical profiles of pollutants are influenced by a number of factors, including topography.