Please answer our online survey!
More information is here
Citizen Science Air Quality Project: Air pollutants and healthier travel choices
Survey now live – available here!
Air quality is a national and global issue with significant costs upon the UK – estimated by the Government at £8.5-20bn per annum health impacts (Defra, 2010: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/69340/pb13378-air-pollution.pdf). This is twice the costs of physical inactivity and comparable to the cost of alcohol misuse to society (Environmental Audit Committee, 9th Report – Air Quality: a follow-up – Volume I, 14 November 2011). There is increasing national interest and recently Client Earth won a case against UK Government (http://www.clientearth.org/major-victory-health-uk-high-court-government-inaction-air-pollution/).
The whole of the city of Liverpool is declared an Air Quality Management Area for Nitrogen Dioxide (NO2) by Liverpool City Council (see: http://uk-air.defra.gov.uk/aqma/details?aqma_id=229). Air pollution is a contributory factor towards respiratory illness and related conditions, and directly affects health. An estimation of 239 deaths in Liverpool in 2010 were attributed to illnesses associated with air pollution (Public Health England 2014 ‘Estimating local mortality burdens associated with particulate air pollution’).
Liverpool Friends of the Earth are launching a Citizen Science project to investigate air quality awareness and perceptions within Liverpool and to investigate the question:
Can transparency of air quality data through citizen science increase awareness and empower people to make healthier travel decisions through reducing their exposure to air pollutants?
The project has been kindly funded by the Liverpool Clinical Commissioning Group and is run by Michael King and Stella Shackel. The funding has enabled 4 portable sensors to be purchased. These sensors can measure the air pollutants: particulate matter, Carbon Monoxide and Nitrogen dioxide. Data from these sensors can be overlaid on Liverpool maps to show the pollution levels for the routes along which the sensors are carried. More information is at: http://liverpoolfoe.org.uk/air-quality-in-liverpool/.
Volunteers will be engaged in the data gathering and feedback processes to help understand how Citizen Science approaches can help individuals and community groups. The data generated could also be used in the future for other purposes, such as comparing air quality data with the presence of green infrastructure. Friends of the Earth also has a national air quality campaign: https://www.foe.co.uk/page/air-pollution-campaign-clean-air
This project is about investigating low-cost methods to empower citizens with information to help them understand air quality levels and decide for themselves how they might act. The air quality data collected will be made available on the Liverpool FoE website.
Please answer our online survey to assess air quality awareness and help to promote it by re-tweeting our tweets from our twitter @LiverpoolFoE. Please answer the survey if you can and contribute to our evidence base! Thank you!
We are looking at air quality within Liverpool as part of our local efforts.
This is a small-scale project with some simple objectives:
No doubt there will be other learning along the way. For example, we are interested in how this sort of community-level engagement and “hyperlocal” data gathering can be used to supplement and complement city-level and national air quality monitoring and management.
This post gives some further examples of the data gathered from our volunteers. The previous post showed some NO2 data captured by the Air Quality Egg sensors.
This post will share some example PM 2.5 data captured using the AirBeam sensors – more info on these sensors is available in previous posts.
The map below shows pm 2.5 readings for 26 Feb 2017 pm. Key roads are highlighted. The readings are generally green and good.
Below is the time-series equivalent of this data showing the data was captured between 16:28 and 17:09. The rise in the levels at the end of the chart may or may not be the result of higher traffic levels as peak hour begins. As always with this sort of data a lot of caution should be exercised before drawing conclusions or insights too quickly.
The map below is also for 26 February but this time during the morning. This time from 11:52 to 13:30. Again the readings are generally green.
There are a batch of readings coloured yellow which appear to be actually located in a large retail store. This can reflect internal local effects, such as a cafe or restaurant with gas burners for cooking. Again, interpretation of this data on its own should always be treated with caution.
The final mapping shared here is from 27 February 2017. All of these readings (more than 5,000) are green and show low levels of PM2.5.
This project is about trying to understand how this sort of data and technology can help. We will post further here on the lessons learnt from discussion with our friends at Breathe Easy and elsewhere.
The technical testing of the sensors has completed and we are grateful to various technical volunteers who have carried the sensors and given us feedback. This project has moved into engaging with the final user group who were always the focus. The aim is to engage with people who might benefit most from this technology and we were very lucky to find the Breathe Easy Liverpool North group, who are a great bunch of people and have been very welcoming to our efforts. We are very grateful.
We first presented to this group on 18 November 2016 to explain the objectives of the project. The group doesn’t meet during the middle of Winter and so we arranged to come back to them in February when their meetings started up again. By February we would have finished our technical testing as per the plan and we would have sensors for them to use if they were willing.
One Friday 3 February 2017 we presented to the Breathe Easy North Liverpool group again and we were grateful that there were two volunteers willing to take 2 sensors: one AirBeam to measure particulate matter and one Air Quality Egg to measure NO2.
We had some technical problems with the AirBeam which meant we didn’t get any data in the first 2 weeks, but early tests on the data captured by the Air Quality Egg looked positive. The mapping of two trips are shown below to illustrate what this sort of technology can do.
The map below shows the data from the Air Quality Egg (Tweedledee) for 20 February 2017. We won’t name the volunteer who captured this data for data protection and privacy reasons, however we are grateful to their efforts.
This data can also be looked at as a time-series and this is shown below. The readings start at approximately 14:00 and finish approximately 15:00 on 20 February 2017.
The goal of this project is to share these findings with the group to get their feedback and see if this tallies with personal experiences and local knowledge. Do the sensor readings agree with personal views on areas of poor air quality? etc.
The second mapping is shown below for 22 February 2017 with the same sensor. This shows similar routes but also some new areas of the city explored. This mapping does show more red readings than the previous mapping. Also, Breck Road is green and has good air quality readings for both trips. Chavasse Park is green, as might be expected for a pedestrianised area – giving some more confidence in the sensor readings. Again, this will be explored further with the group to see if this matches with personal experiences.
As above this data can also be shown as a time-series as below. This shows the readings taken between approximately 13:30 and finish approximately 18:00.
We will analyse more data and share that shortly. More to come.
Note: all the data here should be treated with caution. These are spot measures with low-cost sensors and so they should not be used as a basis for assessing air quality across a longer duration or a wider area.
In a previous post some early tests were performed on one of our two Air Quality Eggs. This post will continue that process but show the two of them together.
Here’s a picture of the two Liverpool FoE Air Quality Eggs (AQEs), which I’ve named Tweededee and Tweedledum!
In this mode they are static AQEs, connected to power, monitoring local air quality and reporting NO2, CO, temp and relative humidity to the opensensors data store every 4 seconds. Below is a picture of the pair of Eggs reporting their data taken 17 November 2016 at 15:32 GMT.
This picture shows one of the challenges of devices such as this. The two Eggs are next to each other but Tweedledee is reporting an NO2 figure of 44 ppb, whereas Tweedledum is reporting 63 ppb.
Investigating this further, the photo below includes another AQE that I have had for a year or so. Again it’s in the same location and picture was taken within minutes of the photo above.
Tweedledee now reads 52 ppb, Tweedledum 74 ppb (a similar delta), whereas the other AQE on the right gives a reading of 22 ppb. The temperature, Carbon Monoxide and Relative Humidity figures are all identical or very close. It’s the NO2 numbers that have some variation.
There are various technical answers to this variation in the data. The Eggs were calibrated before they were shipped, but it’s possible that they need re-calibration, which is not a simple process at all. The key lesson for this project is that individual points from these sort of sensors should be treated with caution.
Before moving on this can be investigated a bit further. Below is a chart that shows the temperature readings for Tweedledee and Tweedledum from 12 November 2016 to 14 November 2016. As above, the sensors are indoors and next to each other. The blue line is Tweedledee and the orange line Tweedledum.
The chart shows the temperature readings slightly differing, but they definitely move in step with each other. The correlation of these data sets is 97% for those who like those sort of statistics!
The final chart in this post is for the same 2 Eggs and the same time period, but this time the data is the NO2 readings from each Egg.
The top line is Tweedledum and the lower one Tweedledee. The chart shows that Tweedledum consistently reports higher than Tweedledee. The movement of the two data sets is harder to see, but they do appear to move in step. The correlation between the two data-sets is 69%. Detailed technical transformations to analyse this data further will not be performed beyond this. The key point from these tests is that the data is potentially valuable, but individual data points should be treated with caution.
A previous post has reported that we have recently received the 2 Air Quality Eggs that are part of the Liverpool Friends of the Earth Citizen Science Air Quality Project. I have used these Eggs before, but there are some new features with these devices, which are important for this project – specifically they offer:
After the unpacking the first stage in the project process is to test them and make sure they are going to be fit for purpose for our volunteers.
The initial set up connects to my WiFi and data is reported to openSensors. Without too many hiccups the Egg is reporting data. The validation of this is not pretty, but is a format like the screengrab below, which shows the Egg reporting Carbon Monoxide, Nitrogen Dioxide, Temperature and Humidity (just off-screen):
Once data is flowing it’s then useful to do some sense-checking on the data and also check the new features i.e. GPS and the onboard storage. So I took the Egg on a trip into Liverpool. During the journey, the Egg is powered by a battery and stores data on the SD card. This data was then later downloaded for analysis.
The picture below shows this data imported into QGIS – an open source GIS package – for analysis.
The GPS appears to be working fine. This was indeed the route taken, using a mix of bus and walking. You can see the larger spacing of the readings when the bus is moving at speed.
The colour coding used is just a spread of the data. Work to be done later is to determine the appropriate colour coding according to national guidance from health agencies and other expert bodies. For now, the colour coding simply shows the spread of data – with dark red being the highest values and white the lowest. The legend in the top right shows the schema and the associated numbers in parts per billion (ppb) of NO2. These figures should not be used in isolation. There’s always the possibility that an individual sensor is faulty or there is some other issue. Confidence comes with increased numbers of sensors and consistent readings over time.
One of the challenges that the test revealed is more about the physical side of the egg than the electronic and information side. We’re going to have to think about how someone can carry the egg, because it’s more bulky than the AirBeam PM sensor. We also need to make it a bit more weather proof and probably buy 2 portable batteries for our 2 sensors when they go mobile.
First tests are looking good though. Next is to get the other Egg up and running, do similar tests with it and then also do some comparison tests to see if these two eggs report similar data. I can compare these two to my other Air Quality Eggs that I’ve got from previous work over the last few years. More on this later.
For this project we wanted to test two different sensors.
We’ve already received the AirBeams that measure Particulate Matter. These two sensors are undergoing testing before we can give them to volunteers. Unfortunately we had a problem with one of these PM sensors and so we are waiting on a replacement to come from the USA.
The other sensor we decided to use is the Air Quality Egg. Importantly, this sensor measures nitrogen dioxide and so provides us with data on this critical air pollutant to support our view of PM from the AirBeams. The AQE sensors are a new model which has GPS and local storage. I’ve used Air Quality Eggs in my own house for several years and so I’m comfortable with how they work, however there will be some uncertainty and learning with this new kit.
The Eggs were caught up in customs from the USA, which meant we had to pay some additional customs charges before they were released.
We paid the customs charges and finally, on 27 October 2016 we received the Eggs. Here’s a photo from the first opening:
Those who are technical minded might be interested in the photo below. It shows the latest Air Quality Egg (on the left) that Liverpool Friends of the Earth has procured for this project. On the right is one of my personal Air Quality Eggs. A lot more technology has been packed into that same case over the last few years. Great work by the guys at wicked device.
Now the Eggs are here I can start testing them. First things first though: stick a label on it!
More updates to follow as testing progresses
As discussed in previous posts we have received our 2 PM sensors – AirBeams – and the first steps in our process are about setting them up and then testing the sensors as preparation before we engage with potential volunteers.
Unfortunately we’ve hit a problem. One of our sensors (and we’ve only got two!) has a problem. This didn’t appear straight away. Both sensors were generating readings in line with expectations and all seemed fine. However one of them (unit #1) seemed to have a problem with its power. It would work fine when connected to the mains supply, but stopped working on battery.
At this stage the AirBeam simply wasn’t powering up at all without a power connection. Our supplier – HabitatMap – have been great and offered excellent support. Together we began investigations to figure out what was going wrong and how to fix it.
Here’s a photo of the opened up AirBeam. The temperature and humidity sensor is the upright white “thing”, the particulate sensor is on the other side of the board to the left.
The final stage of stripping the AirBeam is shown below. This allowed us to get access to the battery. We swapped the existing battery (not shown here) with another battery we had in and sure enough the unit powered up. You can see the red light showing the unit running off battery power.
So we knew there was some fault with power, but was it a faulty battery or a problem with charging?
The photo below shows us charging up the existing AirBeam battery. If the battery wouldn’t take a charge for some reason, then we could focus in here and get a replacement battery swapped in.
When the battery was fully charged we put it back into the AirBeam, reassembled everything and did further tests. These tests led us to the conclusion that the battery is fine, but there is a problem with the charging mechanism in the AirBeam. Therefore we are sending this unit back for replacement or repair. HabitatMap have been great and will reimburse our postal costs.
So we are left with only one PM sensor for now. Our guess is 2-3 weeks for a new one to arrive, but let’s wait and see.
Although it’s frustrating, this was the point of the testing phase. Things do go wrong with technology and it’s better to pick these things up now, than impact the experience with volunteers. This technology is still pretty new!
Following on from previous posts describing the PM sensors that we are using, the next steps are about testing them before giving them out to volunteers.
This post will describe some of these tests and also give some insight into what data is generated and how it can be used.
Some simple trips with the sensors allow me to capture data and make sure they are working as expected. The data that is captured is stored on the phone (android smartphone) and can be later uploaded to the aircasting website or to any other source. The aircasting website is a community mapping development by HabitatMap and provides a form of crowd-sourced data. An example of this is shown below and can be found here.
Beyond this pooling of data in a community map like the aircasting website, it’s also possible to look at individual data streams. This was part of the testing process for our AirBeam sensors. The data from one trip with one of the sensors (unit 1 for reference) is shown in the diagram below:
This shows a single trip in south Liverpool near Allerton Road. The individual data points are visible, with a simple colour coding according to the spread of the data i.e. green are lower values and red are higher values: this does not necessarily mean that red is unhealthy however, it’s just a higher reading.
The diagram below shows the same trip and data but zoomed it for one part. The reading for one data point is also shown. This reading was taken on 13 October 2016 and the PM reading was 13.12 micrograms per cubic meter.
At this stage I will not go further into how this data can be used. The purpose of this post is simply to share some insight into the testing process. A goal of the project is to find out how this data can be useful to individuals and community groups – at this stage we just want to make sure the data is being generated and that it looks in line with expectations. The journey above supports the view that this sensor is working as expected.
More to follow on testing.