Forest Hill Lockdown

From 2019 to 2020, Citizen Sense collaborated with participants in Forest Hill, London, to monitor air quality using the AirKit toolkit. This data story describes differences in pollution levels observed during the first lockdown due to Covid-19 from March to July 2020. As noted in the Brockley Rise Data Story, there were noticeable reductions in air pollution in London during the first lockdown, especially of nitrogen dioxide (NO₂). However, decreases in PM_2.5 were less pronounced during the first lockdown in comparison to pre-lockdown levels. Based on citizen data collected in collaboration with Clean Air SE23 and the Dalmain School, this data story investigates particulate matter trends during the lockdown in more detail to understand why particulate matter pollution levels remained relatively consistent across pre-lockdown and lockdown conditions. 

This data story details specific pollution patterns based on citizen data, weather data and local observations. Drawing on meetings, conversations, and workshops with local residents and community groups, the data story also suggests specific practices for addressing pollution in the area, from planning for better transport to reducing burning practices. 

This story summarises data that have been captured by a small citizen-led network of air-quality monitors in Forest Hill in South East London. Four monitors have been located in this area, including Dustboxes 2051, 2058, 2060 and 2062. The Dustbox 2.0 (or second-generation Dustbox) measures particulate matter 2.5 (PM_2.5), a pollutant that is especially hazardous to human health. Dustboxes 2052, 2060 and 2062 are near the South Circular Road, while Dustbox 2058 is in a residential area near Brockley Road (B218). 

On 23 March 2020, the UK government announced its first national lockdown to slow the spread of Covid-19. This had a profound effect on daily lives and routines. Prior to the lockdown, the government encouraged citizens to practice social distancing. In this story we look at data pre-lockdown (from 7th February to 22 March 2020) and during the lockdown (from 23 March to 15 July 2020). Four Dustboxes gathered data during this time period. Dustbox 2062 was deployed 27 January 2020. On 7 February, Dustboxes 2051 and 2060 were deployed. Dustbox 2058 joined the network 27 February 2020. An abbreviated timeline of lockdown events includes:

• 16 March 2020: The UK Goverment advised the public against ‘non-essential’ travel and social contact.
• 23 March 2020: The UK Goverment advised the public to stay at home apart from for “very limited purposes” including medical reasons, one exercise per day, travel to work and to purchase essential items.
• 10 May 2020: The UK Government advised those who could not work from home to return work but to avoid using public transport where possible. 
• 30 May 2020: People living alone were permitted to meet another person outside.
• 1 June 2020: Primary schools open for some students. 
• 8 June 2020: Those travelling to the UK from from international locations are required to quarantine for 14 days.  
• 23 June 2020: The social distancing rule of 2m is relaxed in some cases to 1m. 
• 4 July 2020: Pubs, restaurants and hairdressers open. 
• 10 July 2020: Quarantine lifted for travel to the UK from 59 countries.

There have been several media reports that suggest lockdown reduced air pollution across the UK. According to a report by King’s College London Environmental Research Group (ERG), there was a significant reduction in NO₂ during the lockdown. Busy roadsite sites in central London had reductions in NO₂ as great as 60 percent, and mean hourly NO₂ levels for all London roads were 21.5 percent lower during the lockdown.

However, in many ways the reduction in PM_2.5 concentrations due to the lockdown was less pronounced. When comparing pre-lockdown to lockdown levels of air pollution, we found a slight increase in PM_2.5 levels. This is likely due to the elevated PM_2.5 levels that typically occur in the spring in London due to regional agricultural emissions. The ERG report does show higher levels of nitrate (NO₃) and ammonium (NH₄) during pollution episodes in the lockdown period, which are agricultural emissions that can contribute to higher PM_2.5 levels.

While PM_2.5 levels in 2020 do not show a marked difference before and during lockdown, we also compared mean PM_2.5 levels from 23 March to 15 July 2019 to the same period in 2020 to see if there were noticeable differences from year to year. Using data from the London Air Quality Network (LAQN) in New Cross Gate (NXG) and Honor Oak Park (HOP) monitoring stations we found there are relatively similar trends in PM_2.5 levels in 2019 and 2020. As can be seen in Figure 1a, there is a period from late May to mid-June 2019 where the NXG monitor is offline. However, the HOP background monitor has complete datasets for 2019 and 2020. Data from Figures 1a and 1b show that for the lockdown period (23 March to 15 July) the mean PM_2.5 level was 2.64 µg/m³ lower in 2020 in comparison to 2019. This corresponds to a decrease of almost 22 percent

HOP is an ambient (or urban background) monitor whereas NXG is a roadside monitor. This means that pollution levels at NXG are normally higher as this monitor is exposed to traffic. In order to evaluate possible differences in pollution levels from 2019 to 2020 at this roadside NXG location while accounting for the missing data, we averaged particle concentrations from 23 March to 25 May for NXG and found that there was a decrease of 5.8 µg/m³ in 2020 in comparison to 2019. For the same period, the HOP monitor shows an average decrease of 4.7 µg/m³ in 2020 in comparison to 2019. Interestingly, this reflects a 28 percent mean reduction in PM_2.5 levels at both monitors. This period was the most restricted part of the lockdown, with travel limited to essential purposes. Figure 2 shows the difference in pollution levels between 2019 and 2020 for this period, with the yellow line indicating periods where noticeable decreases in pollution occurred.

Local sources of pollution
There are several possible emission sources in the area that relate to transport infrastructure. To the south of the monitoring site is the A205 South Circular, a major road that circumvents the south of central London. The route was not purpose built and is made up of existing roads. This results in many intersections that exacerbate levels of congestion. There are four train stations within close proximity of the monitoring area including Honor Oak Park, Forest Hill, Catford and Crofton Park. In addition to the possible sources related to transport infrastructure, there is a crematorium to the north of the site and three allotments, one to the north and two to the east, where residents report that burning occurs.

London-wide, regional and global sources of pollution
Particulate matter sources in London can be attributed to a broad range of emissions. Within London, PM_2.5 from transport (particularly diesel), industry, construction, cooking and heating all contribute significantly to London-wide levels. A significant amount of PM_2.5 emissions also comes from heavy industry and agriculture outside the UK, particularly France, Belgium, the Netherlands, Luxembourg, Germany and Poland. These emissions are thought to account for an urban background of approximately 10 µg/m³. The importance of these transboundary effects of PM_2.5 emissions from outside of the UK on the total London PM_2.5 can vary between 40 to 80 percent daily depending on weather conditions. When long-range pollution episodes do occur in London, they are generally carried on easterly winds. There are a number of global emissions events and practices that contribute particulate matter to London PM_2.5, including fuel production, industrial and domestic combustion, transportation, waste disposal, and agriculture, although these are harder to quantify. The lockdown restrictions meant that travel was only permitted for ‘essential’ purposes, which caused a significant reduction in people travelling by car and public transport. Transport for London report that up to 60 percent of underground services and around 80 percent of bus services were operational.

Observations
Overall levels of PM_2.5 have, somewhat counter-intuitively, in some cases increased during lockdown when compared with the pre-lockdown period. While travel has been restricted, overall levels of particle pollution were affected by prevailing easterly winds bringing pollution from Europe. These regional air pollution events from Europe were due to industry and agriculture, the latter of which can produce particulate matter pollution from the use of ammonium-based fertilisers. As noted above, these findings are supported by speciation undertaken by King’s College London ERG at Honor Oak Park, which shows high levels of nitrates and ammonium in particles during periods of elevated PM_2.5 from 5 to 9 April 2020.⁷

In addition to regional sources of pollution, some local pollution activity is evident. When meeting with members of Clean Air SE23 to discuss preliminary findings from the citizen data, they noted an increase in burning activities during the lockdown including outdoor fires, barbecues and burning at allotments. Lewisham is a smoke control area so there are some restrictions on what can be burned in chimneys. Bonfires are permitted as long as they do not contain plastics or expel dark smoke into the air. During the lockdown the council asked residents not to light fires due to their implications for respiratory health, as well as possible pressure added to the emergency services. These messages were shared on the Lewisham Council website and related social media. In wider London, there was an increase in burning activities. The London Fire Brigade (LFB) report a 7.4 percent increase in outdoor fires during the lockdown in April. The LFB suggest that this could have been caused by people burning waste due to the closure of waste and recycling plants.  

The Dustbox 2.0 device used to monitor PM_2.5 is an “indicative” monitor. This means that measurements can give an indication of pollutant concentrations, but cannot be directly compared with national and international guidelines and standards in an “official” or regulatory sense. Despite this, indicative monitors are a well-established method within atmospheric science for carrying out initial surveys of an area to establish whether a potential problem merits further investigation. Indicative monitors are also becoming increasingly available for citizen-based air-quality monitoring, similar to this study. 

Where possible, the Dustboxes were co-located at the start and the end of the study to account for drift and differences in the sensors during the monitoring period. The co-location of Dustboxes in this study with the NXG reference monitor indicates that there is a good similarity in measurements across the monitors, as well as with monitors in the extended community network, both at the start and end of the monitoring period. 

The World Health Organisation (WHO) has established a 24-hour mean guideline for PM_2.5 of 25 µg/m³ (although there is no safe level of exposure). The line graphs in Figures 3a and 3b show that the overall pollution concentrations for both Dustboxes remained below the WHO threshold prior to and during lockdown.

Figures 3a and 3b document that there have been episodes of increased PM_2.5 during the lockdown period between late March and mid-April. These elevated levels of pollution are caused by regional and London-wide pollution incidents. In Figure 3b we can see a peak in pollution levels around 25 to 27 March. This coincides with particles drifting to London from Europe. Another peak occurs between the 8 to 12 April, which is caused by agricultural emissions from the east. These emissions are observed in the King’s College London ERG report where the chemical composition was measured to be mostly nitrates, ammonium and organic mass. Similar peaks can be observed in Figure 1a which shows PM_2.5 concentrations for the equivalent period in 2019. Increases in PM_2.5 also occur at the end of March and in early April with levels of more than 90 µg/m³ detected at the NXG monitoring site. These peaks are noticeably lower for the 2020 period, which could be explained by differing weather and atmospheric conditions, reduced agricultural activity or a decrease in localised road pollution caused by the lockdown. 

When is the source most evident?
We are interested to identify possible local sources of pollution nearby Dustboxes 2051 and 2060. Here we consider time plots for Dustbox 2060 for the pre-lockdown and lockdown periods. Figures 4 and 5 group particulate matter concentrations across the monitoring period by hour, month and day of the week. These charts show some shifts in PM_2.5 levels throughout the days of the weeks that could relate to shifts in everyday routines and practices.  

Average levels are higher in the latter half of the week during lockdown. It should be noted that data collected pre-lockdown encompasses a shorter period (8 weeks) than during lockdown (16 weeks).  

In Figure 4, Dustbox 2060 levels were highest at approximately 19:00, which could reflect a late evening rush-hour peak. However, low levels of particulate matter are registered during the morning hours. During the lockdown period, the peak is much less pronounced as levels are more consistent at a similar concentration level. Figure 5 documents a gradual increase in PM_2.5 concentrations (2 to 4 µg/m³) during early morning hours and evening hours. This increase could be associated with traffic-related patterns at peak hours. It could also reflect the increased evening burning activities observed in the local area.  

Figure 6 documents time plot patterns for Dustbox 2058, which was located in a back garden close to the Brockley Rise. This data demonstrates that there is a rise in PM_2.5 concentration during the early morning hours until noon. These high concentration could be attributed to cooking as well as transport-related emissions from Brockley Rise. PM_2.5 concentrations are also higher during late evening which could reflect late evening rush hour or outdoor cooking and burning activities.  

Figure 7 documents how PM_2.5 concentrations remained lower throughout the daytime during the lockdown period. This could be because Dustbox 2058 was located farther from the roadside and also north on Brockley Rise away from the South Circular. This data documents that pollution concentration levels were higher on Wednesdays, Thursdays and Fridays. This is perhaps to be expected given that the monitoring period includes multiple London-wide pollution events that occurred in the latter part of the week. During March and April there is a clear increase in particulate matter concentrations due to these pollution episodes. As noted above, these pollution events typically occur in the spring in London and are related to agricultural emissions. 

Which direction is PM_2.5 coming from?
Particulate matter is carried by the wind from emissions sources to the monitoring area. The direction and speed of wind are therefore important ways to gauge the locations of emissions sources in relation to the Dustbox monitors. Note that the polar plots are averaged over a longer period (pre-lockdown 6 weeks and during lockdown 15 weeks) so mean pollution concentrations are lower than in line graphs. 

Pre-lockdown
Concentrations of PM_2.5 are low at Dustbox 2051, which is located within the sheltered outdoor play area of the school site (Figure 8a). Similar results are found at Dustbox 2060 (Figure 8b). The pre-lockdown plots show that pollution concentrations are slightly elevated at Dustboxes 2051 and 2060 when low-speed easterly winds are present. This pattern indicates that sources of pollution are present in the immediate area, which most likely includes Brockley Rise. Lower pollution concentrations are evident from the west across a range of wind speeds. A moderate level of pollution is evident to the north of the monitoring site, where strong winds from a more distant location are carrying pollution to the monitoring area. From the south there are lower pollution concentrations at a range of speeds.  

When investigating elevated particulate matter concentrations to the north of the monitoring locations, we found these were most evident on 27 February and 6 March 2020. The 6 March 2020 incident is also captured by the HOP monitoring location, as demonstrated in Figure 9 below. There is no record of a fire incident in the area from the London Fire Brigade website. Because the pollution was most pronounced on a Friday evening from approximately 19:00 to 21:00, when it quickly spiked to nearly 25 µg/m³, the elevated pollution could be due to burning at the allotment to the north, or due to standing traffic, but this would require further investigation. 

During Lockdown
Overall, concentrations of PM_2.5 were also low at Dustbox 2051. At Dustboxes 2051 and 2060, there continued to be slightly higher concentrations of PM_2.5 evident from the east, which were likely to be connected to traffic on Brockley Rise. However, moderate northerly and low westerly sources of particulate matter pollution are not evident in the data during lockdown, as they were prior to the lockdown as noted above.

Polar plots in Figures 10a, 10b, 10c and 10d suggest the presence of local sources of particulate matter as indicated by elevated PM_2.5 concentration at low wind speeds. The highest pollution concentrations were evident with north-easterly winds when windspeeds were low. For Dustboxes 2051 and 2060, Brockley Rise is likely to be an emissions source. For Dustbox 2062, the South Circular could be a likely source of pollution. With Dustboxes 2058 and 2062 located in back gardens, a potential source of emissions could be cooking or outdoor burning activities. It is also interesting to see that polar plots for all the Dustboxes show a potential emissions source to the southeast, which could relate to burning activities such as barbecues, although this would require further investigation and monitoring to establish the source or sources of emissions. 

Under which weather conditions are PM_2.5 levels most evident?
Different sources of pollution will act in distinct ways according to the weather. For example, windblown dust will primarily occur during dry, windy conditions. Sometimes, you can learn about a source by characterizing this weather-related behaviour. 

Pre-lockdown
Figures 11a and 11b show that the highest concentrations of PM_2.5 are evident in the period from 7 February to 22 March 2020 when the wind direction was to the east. This source is likely to be Brockley Rise, which runs in a north-south direction to the east of the site. Pollution sources are also circulating to the site from the west and northwest.  

During lockdown
Higher levels of pollution are evident during the lockdown at Dustbox locations 2051 and 2060. Overall pollution levels are higher due pollution events coming from European sources to the east, where particulate matter concentrations reached nearly 60 µg/m³. From Figures 12a and 12b, it can be observed that there were elevated pollution levels at low wind speeds with majority of emissions coming from north-east direction.

Using the tools provided through the Citizen Sense Airsift Data Analysis Toolkit, we have analyzed the Dustbox 2.0 PM_2.5 data for the pre-lockdown and lockdown periods in Forest Hill, South East London. We have also drawn on citizen observations, datasets and tools from nearby LAQN reference monitors, along with research from King’s College London ERG. Our observations are as follows: 

• During the lockdown period, while NO₂ concentrations decreased there was not a similar decrease in PM_2.5 concentrations. On the contrary, average PM_2.5 levels were often higher during the lockdown period in comparison to pollution concentrations pre-lockdown in 2020. This is most likely due to seasonal emissions from agricultural activities.
• By comparing PM_2.5 levels at LAQN monitoring sites across 2019 and 2020, it is possible to establish that there was a reduction of between 22 to 28 percent in pollution year on year. This could be attributed to lower levels of traffic or decreases in agricultural activity caused by the lockdown. 
• While regional sources of pollution were detected, based on the analysis of line graphs there was clear evidence of additional local source or sources in some areas, most likely related to road traffic, and potentially linked to burning activities. 
• Throughout the monitoring period, pollution levels are generally below the WHO 24-hour mean threshold for PM_2.5. However, patterns indicate elevated pollution levels that are clearly evident during times of peak traffic activity. 
• During the pre-lockdown and lockdown period, PM_2.5 concentrations are highest when a low-speed easterly wind is present. This indicates that sources of pollution are likely nearby, with the most likely emissions source being traffic on Brockley Rise. 
• London Fire Brigade note a 7.4 percent increase in outdoor fires across London in April. There is some evidence of increased burning activity in the area through the observations of residents including barbecues and outdoor fires. Localised sources not linked to traffic could be related to outdoor burning activities. 
• Lewisham Council’s burning policy currently restricts what can be burnt in chimneys, as well as bonfires that produce dark smoke. The council have requested that residents avoid starting outdoor fires during lockdown. Nuisance burning activities can be reported to the ‘Crime Enforcement and Regulation Service’ by email or phone listed on the council webpage.  

In relation to the evidence and findings from the Dustbox citizen monitoring study, preliminary actions are proposed here that consider the neighbourhood context and existing community organisations and initiatives. The key areas for addressing air pollution include transport, construction, green infrastructure, burning policies and practices, and additional monitoring. These actions have been developed in consultation with monitoring participants, community groups and local area residents.  

Traffic and transport

• Developing and testing alternative transport modes could help to reduce PM_2.5 levels. These measures could include developing transport alternatives so there are fewer cars on roads, using fuel efficient vehicles, and minimising idling activities.  
• Additional sustainable transport measures could include supporting the extension to existing cycle lanes as planned for Forest Hill and areas throughout the Lewisham Borough. 

Air quality monitoring

• Provide resources for community organisations and residents to continue to monitor air quality over time in order to assess improvements from preventative and mitigating actions.  
• Deploy additional air quality monitors near the South Circular (Stanstead Road) to understand pollution patterns related to traffic and the introduction of ultra-low emissions zone plans. 
• Develop monitoring schemes around allotment sites to better understand the impact of organic waste burning on the local area.
• Host air pollution monitoring and awareness events in green spaces to raise awareness about the importance of urban design and planning in relation to mitigating and preventing air pollution.

Policy

• Ensure there are clear policies and stricter controls on burning of organic matter and waste.
• Improve communication to residents and workers on the impact of waste burning on air quality.
• Investigate alternatives to the burning of (garden) waste, including not charging for green or brown bins and improving access to waste disposal sites.

The Citizen Sense and AirKit projects are led by Professor Jennifer Gabrys. These AirKit data stories were developed working in collaboration with Dr Sachit Mahajan and Dr Joanne Armitage. Thanks are due to our collaborators including:

Common Knowledge contributed to the development of the final Airsift platform. Lau Thiam Kok and Tassos Noulas contributed to the development of the data architecture that informed the first-version Airsift platform. Some plots and graphs in this data story use and adapt openair software developed by Dr David Carslaw.

The Citizen Sense Dustbox 2.0 included collaborative contributions to the materials design and 3D printing by Andrea Rinaldi, who built on an earlier Citizen Sense Dustbox 1.0 designs developed in collaboration with Francesca Perona and Helen Pritchard.

Special thanks are due to the participants and residents in the Forest Hill area of London who contributed to the development and testing of the Dustbox monitoring kit, as well as to the collection and analysis of data and communication of results to wider publics and regulators. For more information on project contributors, see Citizen Sense People.

These data stories are generated using the first version of the Airsift data platform, which was developed to allow for citizen-led interpretation of datasets. The core data available for interpretation is the Dustbox PM_2.5 sensor data. An earlier version of the Airsift toolkit was used to bring in air quality data from select sites in the London Air Quality Network (LAQN) for comparison with the citizen data. Please note the Terms of Use for Citizen Sense data.

At the start of the monitoring period, the Dustboxes were co-located with a TSI AM 520 at the University of Cambridge. The site for the co-location study was strategically chosen so that the devices could be evaluated under varying environmental conditions with variable pollution levels. For hourly averaged PM2.5 concentration, the average Pearson correlation (R) was 0.98. Because the Dustboxes demonstrated a high degree of accuracy during the co-location, no scaling factor has been applied to calibrate the devices. Because the sensors were co-located and calibrated during a time of low to moderate pollution, the scaling factor could slightly amplify higher readings in relation to the LAQN readings. However, this would require further testing to demonstrate, since when comparing Dustbox levels with nearby LAQN levels (where available), readings are often comparable. For more information, see “Dustbox 2.0 Calibration and Field Studies.”

This data story is prepared under the assumption that all pollutant, cartographic and meteorological measurements are valid and not sufficiently biased to cause misrepresentation of results. Please refer to the Citizen Sense Terms of Use for further information.

The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. 313347, “Citizen Sensing and Environmental Practice: Assessing Participatory Engagements with Environments through Sensor Technologies,” and from the European Research Council under the European Union’s Horizon 2020 research and innovation programme (ERC Grant Agreement n. 779921), “AirKit: Citizen Sense Air Monitoring Kit.” The University of Cambridge provided additional support through the ESRC Impact Acceleration Account (2020) for enabling impact.

AirKit is available to use under a CC BY-NC-SA 4.0 license. You are free to use, adapt, and share this toolkit for non-commercial purposes under the same CC BY-NC-SA license, and with attribution to Citizen Sense.

To cite this data story, please use the reference: Citizen Sense (2020) “Forest Hill Lockdown,” Covid Data Stories, 31 December 2020. Available at: https://datastories-covid.citizensense.net/forest-hill-lockdown/.