hackAIR News & Updates

In June, the hackAIR team moderated 7 sensor-building workshops across Europe. More than 200 interested participants learned more about air quality and built their own sensor. Here is the first report from this amazing month, featuring the workshop in Brussels on June 13.

Air quality is a “hot” topic in Brussels and all across Belgium: in citizen activist groups, in stakeholder consultations, in policy making and in the media. There is a huge demand for more low-cost air quality sensors to be set up in Brussels. This event was co-organised by a broad range of partners: Influencair, Open Knowledge Belgium, Sandbox VRT, BRAL, Civic Lab Leuven, Luchtpijp, BeCentral, hackAIR. hackAIR was represented by VUB, Crevis and ON:SUBJECT.

It was the largest workshop so far for hackAIR: 100 citizens came – and many more had wanted to participate (no worries: there will be more workshops in Brussels in the next months!)

One of the participants shared about her motivation to build her own air quality sensor:

“I live in an apartment on a busy road in the middle of Brussels. I am very concerned about the impact of air pollution on my health. I want to measure the air quality where I live, with my own sensor. I want to make sense of the data that I will collect. Knowing more about the levels of pollution over time, will make it easier for me to decide whether I can stay in my current appartement. If the air is too bad here, I will move to a neighbourhood with healthier air.”

It was amazing to learn more about the diversity of the participants in conversations around the assembling tables. The age range was from 11 to 65+. There were concerned citizens, scientists in the field of air quality, medical doctors, policy makers, journalists, social entrepreneurs, innovators, open data experts, members of grassroots air quality initiatives (such as Filter Cafe). People came from Brussels and neighbouring municipalities, and also from across Belgium. A mix of French, Dutch and English could be heard in the lively conversations.

The sensors had been prepared by the Open Knowledge Belgium tech team. They did a great job in providing a “plug-and-play-sensor”: everything was already soldered. The participants just had to add four screws and connect a few wires. An easy and very fast experience of building an air quality sensor! Following the assembling phase, each participant was supported by a member of the tech team to connect his / her sensor to the luftdaten platform. There was only one last step to take at home: configuring the sensor with the home WiFi. With the step-by-step manual that was handed out, this was surely easy. Two weeks after the event, the vast majority of the sensors are up and running! 

 

During the workshop, participants were invited to take part in a survey conducted by the hackAIR partner VUB. This survey is part of the hackAIR overall project and explores the connections between engagement in activities about air quality and potential change of behaviour.

 

Our thanks go to all our partners who helped to make this workshop happen, to all who volunteered their time to support participants during the assembling – and of course to all who came and built their own sensor!

Stay tuned! We’ll spread the invitations for upcoming workshops on our events page and through social media. Also keep in mind: you can invite the hackAIR team to offer a sensor-building workshop in your neighbourhood or for your organisation. Or: you can organise your own workshop with our workshop toolkit! 

Green cities do better. They are healthy, with clean air, green spaces, safe drinking water, and offer a great quality of life to the people that live there. EU Green Week 2018 explored ways in which the EU is helping cities to become better places to live and work. Local authorities and citizens were invited to share their vision of a sustainable future.

We presented hackAIR at a full-day networking event during the EU Green Cities Summit in Brussels: Involving citizens in air quality monitoring through Citizen Science initiatives. Together with other initiatives such as iSCAPE and Ground Truth 2.0 we discussed how low-cost air quality monitoring can help provide real-time data and engage citizens in environmental monitoring.

The role of citizen engagement was highlighted throughout as central to delivering the promise of green cities. We heard stories of sustainable urban development, health and waste management in the panel session Towards greener cities: citizens as drivers for change  (link to video recording). Citizen science in particular is achieving environmental policy impact on diverse topics such as air quality and flooding (see the video recording of Making our cities green with Citizen Science for the full discussion).

One of the inspiring examples of the power of large-scale citizen science presented during Green Week is Curieuze Neuzen (Dutch for: Curious Noses). 20.000 citizens measured air quality in Flanders (Belgium) in May 2018. The project team found that citizen science is a powerful tool to raise awareness and get information across about environmental problems. In order to reach those who are not the “usual subjects”, simple and robust methods of data collection are needed. The Curieuze Neuzen campaign collaborated with print media, TV, radio and ads in public transport.

Some highlights of the EU Green Week about air quality and citizen science in quotes:

Did you know…

• that hackAIR calculates air pollution estimates for locations without measurements to create a continuous map in Germany and Norway?
• that it’s possible to estimate air quality using just a picture of the blue sky?
• that hackAIR uses machine learning algorithms to collect suitable images and can estimate the location where a sky photo was taken?

You can find more information about air quality estimation from sky images in an earlier blog post. Today, we want to introduce you to the following features:

1. Data fusion mapping
2.  Image collection and analysis

hackAIR contributes to the growing field of participatory air quality sensing, with the aim to improve air quality data in Europe. hackAIR is accessible – citizens can choose their own level of contribution, reaching from a one-off, simple engagement to high involvement and long-term measurement and data analysis. Users who register on the hackAIR platform can see more features and have a more personalized experience: they receive personal recommendations depending on preferences chosen in their user profile.

Data fusion mapping

In hackAIR we collect mostly point-based observations, through air quality sensors and sky photos sky. For hackAIR users, we want to offer personalized  information about air quality at every location and visualize these observations as a spatially continuous map.

The data fusion module has been running since August 2017. It currently provides hourly updates of the average air quality for the last 24 hours. When we compared the module index with data of official measuring stations, we found a good correlation (R = 0.72, 74% correctly classified). We provide estimates for each 5x5km square in the target countries.

The image to the right shows the average air quality in Germany between September 2017 and February 2018 – overall a very realistic pattern. Pollution hotspots in the Ruhr area and other metropolitan areas are visible, and larger forests and nature reserves can clearly be identified.

On the hackAIR platform, the fusion map is currently hidden. Switch it on by clicking on the little balloon icon and zoom out enough to see the spatial patterns in Norway and Germany.

As mapping air quality at urban scale requires a dense network of sensors with high-quality data, NILU is planning experiments for Oslo later in 2018. The hackAIR pilot partner NILU has been offering workshops to citizens in Oslo to build their own sensor. This will allow to create a network of hackAIR sensors in Oslo in order to map the air quality on street-level.

In the video below, NILU’s Philipp Schneider explains the science behind hackAIR’s continuous air quality map.

Image data collection and analysis

hackAIR integrates several sources of air quality monitoring: official data, open data and APIs. Thanks to a good collaboration, hackAIR also shows the measurement of Luftdaten-sensors across Europe.

To supplement user-generated photos submitted through the hackAIR app, hackAIR collects sky images from Flickr and webcams. As there are many images on Flickr without a tag for location, the hackAIR team has developed a machine learning algorithm to estimate the location through using the image metadata (such as title, tags, description).

Image not geotagged but clearly from Chania/Crete/Greece! Tags: Chania, architecture, building, sky

In a second step, the algorithm detects the portion of blue sky in the photo. As the lower parts of photos often show buildings or landscape, this part is discarded by the hackAIR system. From the recognized sky, the upper third is used for further analysis.

Sample sky detection using three different algorithms

In the video below, CERTH’s Eleftherios Syromitros-Xioufis explains the science behind hackAIR’s image data collection and analysis.

Want to try out hackAIR’s air quality detection from sky images? Download the app and upload a picture of the sky near you to get an estimate.

One common criticism towards low-cost air quality sensors such as the ones used by hackAIR is that they cannot achieve the accuracy of more expensive reference stations. We set out to test this claim. The hackAIR air quality test station is now active in Athens since October 2017. 

Using the setup pictured on the right, we wanted to make a single installation site for all hackAIR sensing devices, including all supported hardware and sensors. In particular, we were interested in understanding the impact of humidity and temperature on the measurements and whether the measurements would shift over time due to ageing. In addition, we included a commercial air quality sensor.

Sensors in the hackAIR air quality test station

• hackAIR home v2 (Wemos, sensors used: DfRobot SEN0177 and DHT11)
• hackAIR home v1 (Arduino Ethernet, sensor used: DfRobot SDS011)
• hackAIR home v1 (Arduino WiFi, sensor used: DfRobot SEN0177)
• hackAIR cardboard
• Dylos DC1100 (commercial reference sensor)

What did we find out?

We saw the measurements of the hackAIR low-cost sensors correlate sufficiently well with the reference measurements. hackAIR’s power saving protocol proved to be effective in protecting sensors from rapid ageing: we did not detect any shift after almost 6 months of continuous measurements. 

Humidity and temperature constitute two significant factors that influence the quality of fine particles and thus impact the corresponding sensor recordings. hackAIR is using a normalisation function on the sensor itself to adjust the air measurements in these situations. 

During the testing period, the devices showed acceptable network stability: no disconnections were recorded. We were able to fix a minor problem with the ethernet http connection. 

While this initial test site in Athens is still operational, we want to set up an additional installation next to an official measuring sites to be able to directly compare hackAIR measurements with official air quality information. 

This week’s blog post features four citizen initiatives that are taking action to reduce air pollution – in Belgium Kosovo, Bulgaria and Great Britain. Better air quality information enables us to fight for better air in where we live: Get inspired by these stories – and share with us what you are up to!

Help – I can’t breathe: Parents take action on air pollution around schools in Belgium 

On March 14, 2018 a report in Belgian national television informed the public about the dangerous levels of air pollution around schools. This documentary showed the results of a large scale study on air quality and used two schools as examples. 

The next day, the first protest actions happened at school streets. Filter Cafe Filtre was born.

The invitation is simple and powerful: “Close off the street where your school is to traffic every Friday morning before school starts. Have a cup of coffee together and let the children play in the car-free street. Invite local politicians to come and join you for a coffee.”

The initiative is growing: on April 20, 60 parent groups across Belgium took part. The mission is to raise awareness and change mindsets of parents and children, and to convey the urgency to policymakers.

Get data – Drive change: Youth-led air quality campaign in Kosovo

Kosovo is one of the most polluted regions in Europe. The project “Science for Change Kosovo” is a youth-led environmental movement, investigating air pollution in Kosovo and mobilizing people to take actions. The project was one of the partners in the Making Sense project. 

It works with local youth around environmental policy issues, how to collect data about air quality, and how to use this data to advocate for change. Their approach was documented in the Making Sense toolkit for citizen science initiatives – have a look!

Bulgaria: New collaboration between citizen activists and government to reduce air pollution

Air pollution is much worse in Eastern Europe. In Sofia, air pollution norms were exceeded on 70 days between October 2017 and March 2018. Government and activists are working hand-in-hand to improve the situation. The citizens’ group “AirBG.info” was founded in April 2017 and has rapidly expanded their network in the first year of their activities.

This initiative maps air quality with low-budget sensors, built by citizens. The open data from this citizen science project are collected at the luftdaten platform. Currently, 600 air quality sensors are operating in Bulgaria, nearly 300 of them in the capital. AirBG.info has been very successful in establishing a close collaboration with the Bulgarian government and with the environmental executive agency. This collaboration led to the decision that Bulgaria will integrate all existing air quality monitoring systems. The Environment Minister said: “We are trying to integrate all [air quality monitoring] systems: those of non-government organizations, of which there is a myriad but are less precise, and those of the Executive Agency for Environment, which are fewer but are more precise, in order to get a clear picture about the quality of air we breathe.”

Parents campaign for clean air in London

Mums for Lungs is a London based group of parents who are seriously concerned about the dangerous impact air pollution is having on children’s health. They have set up air quality campaigns in various contexts.

In March 2018, they campaigned for the extension of the Ultra Low Emission Zone in London. Within this zone, most vehicles will have to meet emissions standards or pay a daily charge. In London, around half of emissions of particulate matter (PM) come from transport. The Ultra Low Emission Zone will reduce the most harmful emissions generated by road traffic. Mums for Lungs, together with a number of local partners, has been calling to extend this zone so that more inhabitants of London can benefit from cleaner air.

In the ongoing local elections, Mums for Lungs has proposed a number of pledges to local parties – raising awareness about air quality. The pledges are also a way to hold policy makers accountable – after the elections.

Have you taken a sky photo with the hackAIR app? Are you curious how the app estimates the air quality, based on your photo? Do you want to know more about the science behind this feature?

Even with your bare eyes, you can see that the cloudless sky is not always the same blue. The color of the sky changes with the time of the year and time of the day, but it also changes when particles are present in the atmosphere. In very polluted cities the sky often does not appear very blue, while it can be deep blue when it has rained recently and it is very windy and it can also be deep blue in remote locations.

The feature

Once you have taken a sky photo with your hackAIR mobile app, the image is uploaded to the hackAIR server where we analyse the colour of the sky and estimate the current air quality. When the calculations have completed, the picture is shown on the hackAIR map with an estimate of the air quality.

Behind the scenes

• Step 1: A computer algorithm detects the portion of blue sky in the photo. As the lower parts of photos often show buildings or landscape, this part is discarded by the hackAIR system. From the recognised sky, the upper third is used for further analysis.
  
• Step 2: Using the average colour detected, hackAIR then calculates the ratio between red and green light bands (R/G ratio).
  
• Step 3: We then look up this ratio in a table that lists the corresponding air quality value in relation to the location and time of day at which the photo was taken.
  
• Step 4: The estimated rating of the air quality, together with the photo, is shown on the hackAIR platform.
  

The scientific background

The main idea is that the ratio of red and green band of the light (R/G) depends on the amount and type of aerosols in the atmosphere (R/G increases with increasing AOD).

Aerosols, which are tiny particles suspended in the atmosphere, are emitted by natural (e.g., volcanoes, desert dust, forest fires, sea salt) as well as human activities (e.g., biomass burning, combustion of fossil fuel, industrial activities). Aerosols affect the levels of surface solar radiation by scattering and absorbing the light coming from the sun reducing the visibility in the atmosphere.

By measuring the radiation that reaches the ground at specific wavelengths we can assess the degree to which aerosols prevent the transmission of light by absorption or scattering. This is expressed as aerosol optical depth (AOD). AOD is unitless and usually reported at a wavelength of 550 nm. A value of 0.01 corresponds to an extremely clean atmosphere (air quality = very good), and a value of 0.4 would correspond to a very hazy condition (air quality = bad).

To control for varying atmospheric conditions, we produced a set of look-up tables using a  radiative transfer model (RTM). RTMs calculate the intensity of the light transferred within the atmosphere under different user-input scenarios that include information about the position of sun (solar zenith angle) relative to Earth and various atmospheric parameters (e.g., clouds, aerosols, water vapour, ozone, surface albedo, etc.).

We then compare the measured light intensities with those expected conditions to retrieve an estimate of the current aerosol optical depth.

A similar approach has been used to examine the atmospheric effects of volcanic eruptions in historical paintings¹. Instead of the painter’s eyes, hackAIR uses mobile images and publicly available images from Flickr and webcams. We performed a number of tests to check the effect of the camera type on R/G ratios – as previous studies indicate, this is not a major issue.

Future potential

This analysis of sky photos makes it possible to monitor air quality in urban and rural environments with easily available tools: digital and mobile cameras. This estimation of air quality supports citizens in collecting more valuable information about the quality of the air: e.g. in their neighbourhood, around schools, in busy streets.

hackAIR uses the sky photo analysis not only for images provided by users through the mobile app. This image analysis is also used by the hackAIR tech team to analyse a very large number of photos from Flickr and webcams.

---

¹ Zerefos, C. S., Gerogiannis, V. T., Balis, D., Zerefos, S. C., and Kazantzidis, A.: Atmospheric effects of volcanic eruptions as seen by famous artists and depicted in their paintings, Atmos. Chem. Phys., 7, 4027-4042, https://doi.org/10.5194/acp-7-4027-2007, 2007.

Once you’ve set up your hackAIR sensors, the question is: what can you do with the data? Of course, you can check it day by day on the map, but wouldn’t it be cool to do more? Let’s see what’s already possible.

This blog post has been written for hackAIR users who are familiar with Excel and curious about data visualisation. Please check out what hackAIR communications lead Wiebke Herding has learned about visualising data! We’d also love to hear from you what you have been experimenting with.

From the sensor profile

Each sensor has its own profile page that you can find by going to Profile > Sensors. After a moment’s wait, you’ll get a list of the latest measurements translated as air quality ratings.

Using the ‘Export Measurements’ button, you can then download the exact measurements from your sensor in CSV format. This way, you can export 5000 measurements at a time – if you need more, just run multiple exports. Note: the timestamp in the export is in GMT – depending on your own time zone, you might need to adjust this (e.g. add an hour if you are based in Berlin).

Air pollution over time with Excel

After downloading the files, you can open them in Excel. I then added three columns to be able to access the values I was interested in:

• Date CET: =<Date>+”01:00″
  
• PM10: =IF(<Pollutant_Q_Name>=”PM10_AirPollutantValue”,<Pollutant_Q_Value>,””)
  
• PM2.5: =IF(<Pollutant_Q_Name>=”PM2.5_AirPollutantValue”,<Pollutant_Q_Value>,””)
  

Note: these formulas assume that you use English localisation settings. If you use Excel in a different language, you might need to adjust quotation marks and commas.

Using Insert >PivotChart, I created a chart with Date CET as the axis and PM10 and PM2.5 as the values. I set both values to show the average PM measurement and changed the chart type to a line graph. The result was a time series of daily averages:

Using the report filter buttons at the bottom right, I could then zoom into the hourly averages and finally into the individual measurements.

As I am currently running two sensors (120 is a hackAIR home v1 at the front of my house, 255 is a hackAIR home v2 in a more protected space at the back), I can also compare the two:

 

On average, the sensor at the front of my house picked up 53% more PM10 particles, and 23% more PM2.5 particles in the testing period.

Building graphs that update themselves

As downloading the files can get tedious over time, we could also use the hackAIR API to access our data. We can use a service like data.world for that. After setting up an an account and creating a new project, you can add your own sensor data using Add data > Add from URL.

Paste the following link: https://api.hackair.eu/hackair_data?access_key=1234 (replace 1234 with your own sensor’s access key). Add the extension .json to the file name, and you’re good. 

To enable automatic updates, go to project settings, and enable the Automatic Sync Options. You can then explore your data and build graphs like the one below.

Data.world alone will not give you graphs that you can permanently link to, but it’s provides good access to the API data. One option for building graphs is Google Data Studio. Log in and add a data source, adding data.world as a community connector. You’ll need the URL of your data.world project. To import all data from your sensor, add “SELECT * FROM <table_name>” as your SQL query. To be able to use the data, you’ll need to make a few adjustments:

• date_str: set the type to Date Hour (YYYMMDDHH)
  
• pollutant_q_value: set the aggregation to None
  

Now add two new fields and set both of them to aggregation = Average.

• PM10 with the formula “CASE WHEN pollutant_q_name=”PM10_AirPollutantValue” THEN pollutant_q_value ELSE 0 END”
  
• PM2.5 with the formula “CASE WHEN pollutant_q_name=”PM2.5_AirPollutantValue” THEN pollutant_q_value ELSE 0 END”
  

Now you can set up your report, for example using the time series chart or the data table. Play around – and when you’re done, you can share the link to your sensor data. Here’s the view of my recent measurements: https://datastudio.google.com/open/11iG_TgonCmPy0nFmUS0ObVWLudstQk4M

How about measurements from a specific geographic area?

You can download the latest measurements from a rectangular geographic area as follows:

1. Determine the coordinates of the top left corner of your area, e.g. by locating it on OpenStreetMap and selecting show address. This will give you a pair of two numbers, the latitude (e.g.  52.6315 for Berlin) and the longitude (e.g. 13.1259 for Berlin). We’ll call them lat1 and lon1.
   
2. Determine the coordinates of the bottom right corner of your area, e.g.  52.3153, 13.7569 for Berlin. Again, we’ll call them lat2 and lon2.
   
3. Go to the following location: https://api.hackair.eu/hackair_data?location=lon1,lat1|lon2,lat2 (replacing the lon and lat variables with your actual values)
   

Now you can either add this link to data.world (as explained above) or transform it to a csv file using a JSON to CSV service (like Konklone.io/json). After you import or connect this data to Google Data Studio, you’ll again need to make a few adjustments:

• date_str: set the type to Date Hour (YYYMMDDHH)
  
• pollutant_q_value: set the aggregation to None
  

Now add two new fields:

• with the formula “CASE WHEN pollutant_q_name=”PM10_AirPollutantValue” THEN pollutant_q_value ELSE 0 END”. Set the aggregation to average.
  
• coordinate with the formula “CONCAT(loc_coordinates_1,”,”,loc_coordinates_0)”
  

You can now add a map to your report. Here are, for example, some of the latest measurements in the Netherlands:

Over to you!

• What other ideas do you have to visualise and use the data you collect through hackAIR? Any cool tools we’ve missed?