COMPAIR recently completed its mapping of 100 Citizen Science (CS) projects in Bulgaria, Germany, Belgium and Greece. In a series of upcoming articles, we will present the main findings of this research, starting with a general overview and conclusions, before moving on to individual country samples, their unique features and characteristics. The mapping data gathered in the process will be made available as country blogs are published. We will conclude the series with a novel typology of CS regimes. The new framework advances the state of the art of CS research by proposing new ways of categorising CS landscapes at a national level.
COMPAIR is a Citizen Science (CS) project that helps cities make their air cleaner for everyone in the context of green and digital ‘twin’ transition. To achieve that, COMPAIR pilots a multi-pronged methodology that includes extreme CS, low-cost air sensors and advanced digital tools in four geographically varied locations across Europe: Sofia and Plovdiv in Bulgaria, Berlin in Germany, the region of Flanders in Belgium, and Athens in Greece.
Established in these cities will be local CS labs, each with its unique focus:
Bulgarian CS lab: will create awareness of how different commuting patterns affect air quality; improve air quality around schools through new mobility measures; champion sustainable behaviour among young people
Berlin CS lab: will decrease the city’s impact on climate by introducing more car-free streets; use CS data on traffic counts and air quality to model an optimal policy scenario
Flemish CS lab: will help city inhabitants avoid pollution hotspots by providing personalised recommendations based on air quality data from wearable sensors; measure traffic around schools using Telraam sensors to support the implementation of a school street scheme
Athens CS lab: will create a greener city by measuring and understanding the environmental impact of everyday habits e.g. wood burning, smoking, recycling, heating etc.
The review sets the context for COMPAIR CS labs by providing a snapshot of the national CS landscapes in the four countries.
Research design
We started our mapping by conducting searches on two popular CS databases (EU-Citizen.Science, CitSci-X) and Google, and also asked local teams to fill in gaps by recommending any additional projects they know that would add value to the pilot sample.
Included in these samples were all kinds of initiatives. Those in which citizens act as sensors (crowdsourcing), those that use citizens as basic interpreters (distributed intelligence), those that engage citizens in problem definition and data collection (participatory science), and those that go one step beyond by allowing citizens to also take part in the analysis stage (extreme citizen science). Because our goal was to capture CS diversity that extends beyond levels of participation, we chose to include projects that play more of a supporting role in the national CS landscape. Such projects don’t necessarily conduct any CS experiments themselves, but thanks to their presence the national CS ecosystem is able to grow and develop, with results disseminated far beyond the country’s boundaries. Examples include national or regional CS aggregation portals, capacity building platforms and technical initiatives that do some or all of the following: provide curated collections of CS projects, train and advise those who want or already run a CS project, develop CS kits for use by volunteers during CS projects.
Due to time constraints, we capped the total effort at 100 initiatives. In order to have some balance in the overall composition, we split the country samples as follows:
BG sample 20 initiatives
DE sample 29 initiatives
BE sample 31 initiatives (predominantly Flanders and Brussels)
GR sample 20 initiatives
More projects were reviewed for the German and Belgian samples because CS landscapes in these two countries are considerably bigger than those in Bulgaria and Greece.
The actual mapping followed a predefined schema. In addition to general description, we collected information on three areas of particular interest to COMPAIR. These are stakeholder engagement, data collection and analysis, and impact.
Finally, to make our analysis more focused, we split each country sample into two clusters. The primary cluster includes air and traffic related projects, the secondary one all other projects e.g. those dealing with biodiversity, water pollution, humanities and arts. The ensuing critical analysis was applied more rigorously to the primary cluster as it is more relevant to COMPAIR.
Bulgaria
Field | Reviewed BG projects | Total |
Air quality monitoring | METERAC, AirBG, HEAL Sofia, Dustcounters, IQAir Sofia | 5 |
Biodiversity monitoring | Alien CSI, The Quest for the Storks, ANEMONE, RECONNECT, Gecko monitoring, PECBMS, Let's count the sparrow | 7 |
Monitoring of water quality, vegetation, environment, waste and odour pollution | DNOSES, EdnoDarvo, GLOBE BG, Shared Compost, Watermap of Bulgaria | 5 |
Social sciences, humanities and arts | CitizenHeritage, REFRESH, Citizens’ App | 3 |
The Bulgarian sample provides an interesting snapshot of the CS landscape in the country. Our mapping shows that the landscape is bigger and diverse despite the limited amount of information available for Bulgaria on popular CS aggregation platforms. While many initiatives arrived in the country as an extension of an international project, there are some domestic projects initiated by local stakeholders that attest to the internal capacity to start and manage CS projects with some degree of success.
Projects that measured air pollution did so using low cost sensors that varied depending on pollutant type e.g. diffusion tubes (NO2), NodeMCU based IoT sensors (PM). People who used them to make measurements did receive some prior training and guidance but it’s not clear to what extent they were involved in stages preceding data collection (e.g. problem formulation, location selection) and stages that followed it e.g. reflection, analysis, lobbying for change. Not surprisingly, we couldn’t find any references in projects’ documentation to ECSA’s 10 principles that promote deep and meaningful engagement beyond passive data collection, however it’s possible that some principles were followed in practice if not in name.
Other aspects of stakeholder engagement that were equally poorly visible in the primary cluster are i) participation of hard-to-reach groups (the only exception might be HEAL as it worked with school children) and ii) involvement of all members of the quadruple helix community (the only project that clearly acknowledged the need for this model is METER.AC).
Impact in the primary cluster is confined to the field of technology. Here, it is METER.AC’s approach to data sharing and Dustcounters’ Arduino-based sensor that caught our attention. Impact on individuals and policy was limited to a few statements that appear more as project objectives than facts established through research.
This was also a problem in the secondary cluster, where impacts are presented more as ambitions than survey findings or concrete steps taken by, say, policy actors as a result of the project. Even if we found no attempts to measure impact, it’s interesting to see what projects thought their potential benefits might be e.g. ability to counter the influence of fake news (ANEMONE), better mental health and strengthened social bonds (RECONNECT), reduced usage of plastic bottles (The Watermap of Bulgaria).
Germany
Field | Reviewed DE projects | Total |
Air quality and traffic monitoring | hackAIR, Measuring the Berlin Air, SenseBox, HEAL, BerlinAIR NO2 Atlas, Sensor.Community, PolDiv, Envirocar | 8 |
Biodiversity monitoring | Fledermausforscher in Berlin, Stadtwild Tiere Berlin, InsktenMobil, Muckenatlas, NaturGucker, My Ocean Sampling Day, Ornitho, Tauchen für Naturschut, ArtenFinder | 9 |
Water quality monitoring | Plastik Pirates, Citclops, FLOW, BeachExplorer | 4 |
Environmental and atmospheric monitoring | GLOBE, Tator Gewasser, Netatmo CWS, PV2Go | 4 |
Monitoring of odour pollution and soil quality | DNOSES, Open Soil Atlas | 2 |
Other | Burger Schaffen Wissen, SimRa | 2 |
Initiatives included in the German review cover around a fifth of the German CS landscape. But even with such small sample we were able to uncover a great variety of initiatives: those that arrived in Germany as an extension of another (international or EU-funded) project, those that started in Germany but have since spread to other countries, those that are initiated, funded and managed by domestic stakeholders, and those that support the growth of CS community without running any CS activities themselves. The discovery of ‘supporters’, ‘inside-out’ initiatives and a high volume of domestic projects is especially encouraging as their presence can be an indicator of the country’s advanced CS maturity level.
Projects that monitored air pollution and traffic did offer some training and guidance to participants, but the effort was mainly oriented toward technical skills needed to successfully operate a DIY sensor. Some projects experimented with different outreach tactics, and it appears that retention rates are better when people take part in live demos and workshops, as opposed to just being targeted online. We couldn’t find any evidence of a more comprehensive engagement strategy being used, except in the secondary cluster, where one of the projects (Citiclops) developed a five-step process to turn volunteers into co-managers of a CS lab.
In both primary and secondary clusters we found projects that worked with schools to monitor air and water quality (HEAL, SenseBox, GLOBE, FLOW). Several also claim to have achieved quadruple helix in their engagement (hackAIR, Dnoses, Open Soil Atlas, Citiclops, Tator Gewasser). Even though projects do not acknowledge following ECSA’s 10 principles, many did so in practice. In several air related projects, participants were involved in coding and training (hackAIR), in preparing and evaluating passive samplers in the laboratory (BerlinAIR NO2 Atlas). This shows that projects may be following at least some of the principles even if there is no reference to them in project documentation.
Devices used for measuring air pollution varied depending on pollutant type. NO2 was measured the traditional way (i.e. diffusion tubes) in the HEAL project. Particulate matter was measured using DIY sensors provided by Sensor.Community and senseBox. An exception was HEAL that used a commercial, albeit also low-cost, device with a light-scattering property, for PM measurements.
The project whose technical impact impressed us the most is Sensor.Community. Its DIY sensor sensor kit has been used by thousands of individuals and projects worldwide. For about 50 euros, people can build a compact device to measure air pollution around them, share data on a map (if they want), and become part of a global CS network. Other projects that caught our attention are those with a strong focus on open standards (enviroCar, Citiclops).
Where attempts were made to measure behavioural change (hackAIR), the most commonly reported impact was increased knowledge and interest in air pollution. The project’s impact assessment did not find any evidence that CS activities triggered some preventive behaviours among participants, such as avoiding going outside, keeping one’s windows closed, and wearing a mask. But some survey respondents did mention that they tried to avoid busy roads more often (protective behaviour) as a result of increased awareness about risks posed by air pollution.
Projects in both clusters managed to attract a considerable number of volunteers, though air related projects often share the number of sensors installed rather than the number of volunteers engaged e.g. Sensor.Community (14.388 sensors), senseBox (9551 sensors). Clear leaders by volume in the secondary cluster are NaturGucker (with 111.395 volunteers), Muchekatlas (29.000 volunteers) and Plastik Pirates (15.000 volunteers). The latter started as a German project but has eventually spread to Portugal and Slovenia thanks to the endorsement of the Trio-Presidency of the Council of the EU, demonstrating that buy-in from senior policy makers can scale the project considerably within and across borders.
Finally, besides Plastik Pirates, some political interest in CS results was observed in relation to senseBox (the Ministry of Education used them to coordinate learning opportunities for students), Tauchen für Naturschut (the idea of nature-conservation diving was introduced in the Mecklenburg Lake District) and Muchenatlas (authorities have benefited from more accurate risk assessments of mosquito-borne diseases which allowed them to implement better public and animal health policies in Germany). Other projects either didn’t target policy makers as a priority group or their policy impact appeared more as an ambition than something established through some form of impact assessment.
Belgium
Field | Reviewed BE projects | Total |
Air quality and traffic monitoring | AIRbezen, iSCAPE, Curieuze Neuzen 2016, Curieuze Neuzen 2018, Leuvenair, HASSELair, Meet Mee Mechelen, hackAIR, CurieuzenAir, Luchtpijp, InfluencAir, ExpAIR, WeCount, BikeSTEM for Schools | 14 |
Biodiversity monitoring | Snapp nature, Bugs 2 the Rescue, TrIAS, Animals Under Wheels, My Gardenlab | 5 |
Monitoring of atmospheric conditions, water and soil quality, odour and noise pollution | Butterfly, Stiemerlab, Curieuzeneuzen in de Tuin, Omniscientis, NoiseTube | 5 |
Social sciences, humanities and arts | SOS Antwerpen, MamaMito, Citizen's talk, CitizenHeritage | 4 |
Other | Scivil, Iedereen Wetenschapper, AstroSounds | 3 |
Belgium has north of 180 CS initiatives. By reviewing just 31 of them we barely scratched the surface of the country’s CS landscape. Even so, with our small sample we hope that we managed to capture a good mix of domestic initiatives (local and regional), EU projects with pilots in Belgium, and CS ecosystem ‘supporters’ i.e. domestic platforms that help others to start, manage and promote CS projects. The latter type was also found in Germany and we think it can be an indicator of the landscape’s maturity level.
The Belgian sample has the largest primary cluster comprising 12 air related and two traffic related initiatives. Here, most projects avoided a hands-off approach, preferring a deeper engagement instead. Where this happened, participants were involved in pre- and post-data collection activities that ranged from brainstorming, workshops and training, to analysis, reflection and planning.
At least six projects in the primary cluster worked with schools to monitor air quality. In projects where schools are the main stakeholder, the number of participants can exceed dozens and even hundreds of institutions (181 schools in the case of AIRbezen, for example). Inclusion of schools in CS is a welcome sign that shows increased awareness of threats that air pollution poses to young people, and also the contribution that CS can make to school curriculum, in particular STEM education.
Through schools, projects have a channel by which to reach out to vulnerable and excluded groups, though very few projects actually succeeded in engaging people with a lower socioeconomic status. WeCount tried but ultimately failed. CurieuzenAir is the only success story that managed to achieve inclusivity by including special community locations in its measurement network and by working with local NGOs that deal with issues like health and poverty.
Two projects that set a record for engaging the largest ever number of citizens in air pollution monitoring (in Europe at least) are both from Belgium. We say two but actually it’s the same project (Curieuze Neuzen) that ran two campaigns, a local one in 2016 in Antwerp (2000 volunteers) and a regional one two years later across Flanders (20.000 volunteers). Other projects that followed can also boast some impressive numbers e.g. CurieuzenAir (3000), BikeSTEM for Schools (1200).
We noticed little innovation in the way pollutants were measured. The SDS011 sensor popularised by Sensor.Community is clearly the preferred choice for PM measurements. For nitrogen dioxide it is still passive samplers (diffusion tubes). Only in iSCAPE, NO2 was measured through a computerised sensor, an Arduino-compatible Smart Citizen Kit.
Although sensor selection was quite standard, what really impressed us was the way in which some project results were presented. While online maps is by far the most common way of presenting results, displaying these maps on a newspaper’s website is a practice we only observed in the Belgian sample. De Standaard, a Flemish daily, provided this kind of dissemination support to two projects, Curieuze Neuzen in de Tuin and more recently to CurieuzenAir. Providing map based visualisations on websites of popular media outlets can give a significant boost to any project-level dissemination effort, allowing more people to learn about CS results.
Impact on individuals and policy is not something that can be easily identified based on project documentation. Oftentimes impacts are formulated as project objectives, with no clear evidence of how they were achieved. But sometimes projects do evaluate their work. Based on evaluation results of several projects (hackAIR, iSCAPE, Ground Truth 2.0), the most common benefits to CS participants are greater awareness of and improved motivation to get involved in air issues, as well as higher propensity to engage in soft-mobility behaviour. Policy impacts usually vary from project to project. WeCount, for example, reported a clear intention by the city of Leuven to make Telraam a permanent resource that can be drawn upon to implement, monitor and evaluate new mobility measures. iSCAPE mentioned the interest of the Hasselt authority regarding the use of CS sensor kits in higher education projects that could be undertaken in the coming years to help solve mobility and air related problems.
Potential to deliver impact is not limited to air and traffic related projects. Those in the secondary cluster also motivate people to carry on with CS activities (Curieuze Neuzen in de Tuin) and support policy making in different areas and ways e.g. by providing a platform for STEM lessons on gardens (My Gardelnab), by broadening environmental sensing in cities (NoiseTube), by providing information on the impact the road network is having on nature (Animals Under Wheels), by providing a hands-on manual for towns and cities on when and how to use CS (Scivil).
Greece
Field | Reviewed GR projects | Total |
Air quality monitoring | hackAIR, Cos4Cloud, URwatair | 3 |
Biodiversity monitoring | Sharks and Rays in Greece and Cyprus, Hellenic Fauna CS Project, eBird, "Is it Alien to you...Share it!", Alientoma, Biodiversity GR, CrabWatch, iNaturalist | 8 |
Monitoring of soil quality, environmental conditions, urban waste, water and odour pollution, and seismic activity | Scent, DNOSES, Waste4Think, GROW, GLOBE, Hackquake | 6 |
Humanities and art | CitizenHeritage, Wreck History | 2 |
Other | INCENTIVE | 1 |
The Greek review is based on the mapping and analysis of 20 initiatives. Despite its modest size, the sample conveys some interesting insights on the state of play of CS in Greece. We noticed a large presence of EU projects, more than in any other pilot country, active in all fields covered by the review, from air monitoring to humanities. Also noticeable are renowned international projects with a large volunteering force in Greece reaching thousands (eBird GR) and sometimes tens of thousands of people (iNaturalist GR). Despite considerable external influence, the landscape is not without domestic projects, with some interesting examples found at local and national level, in fields ranging from air to biodiversity monitoring. We even found one project that has potential to evolve the country’s CS landscape into a more mature state, by establishing a CS hub at one of the country’s leading universities.
Engagement within the primary cluster involved some close-knit cooperation between CS projects and stakeholders. The latter benefited from training, sometimes in the form of an online course, designed to improve their capacity not just to build sensors but also develop CS scenarios for integration into institutional structures (e.g. education curriculum) to achieve a more lasting impact. Projects were therefore clearly thinking of pre- and post-data collection activities, often taking active steps to support their implementation. This corresponds to several ECSA’s principles e.g. 1, 2, 3, 4. But like in other country samples, no explicit references to these guidelines were found in projects’ documentation, which underscores yet again that projects usually follow some principles, if not in name.
While projects in the primary cluster worked with just several schools, the number was considerably higher in the secondary cluster, where Hackquake alone set up seismographs in 100 schools across the country. Between them, the two clusters managed to engage a significant number of young people. However we don’t know how many of them were from vulnerable groups, and how exactly CS benefited students apart from improving awareness and interaction on issues like air pollution and earthquakes. Only hackAIR provides a glimpse of behavioural change stimulated by CS, as one participating school reported increased motivation among students to use CS experience as a springboard for new activities e.g. entrepreneurship.
We noticed that participation of industry was more often than not missing in the reviewed initiatives, with triple helix being the most common achievement. Where this was the case, projects often just referred to different stakeholder types to stress the result, and it’s only Scent that went so far as to provide a detailed breakdown per group e.g. 62.6% private, 16.4% public, 16.4% students. COMPAIR and other projects would be well-advised to copy this practice as it’s far more convincing than simply referring to the triple/quadruple model or its different constituent parts.
Data collection tools used for monitoring air pollution are pretty standard: a DIY sensor popularised by Sensor.Community and a couple of low-cost, commercial solutions all measuring PM2.5 and PM10. In the secondary cluster, projects that monitor biodiversity are increasingly using apps to capture data. Many domestic initiatives use international platforms as a dissemination tool. It appears that iNaturalist is now to biodiversity projects (at least in Greece) what Sensor.Community is to air related ones. Biodiversity is one area where we noticed many projects with gamification elements, such as leader boards, that aim to promote competitive spirit among participants.
Looking at technical innovations in the secondary cluster, we were fond of projects that adapted international standards to enable real-time monitoring of odour nuisance by any citizen, anytime, anywhere (DNOSES), that made time-series datasets on biodiversity available for use by citizens, researchers, media and policy makers (Biodiversity GR), and that extended GEOSS and Copernicus services to improve the monitoring of land-use changes locally (Scent). A project with tangible personal impact is GROW, whose CS activities helped participants learn and implement regenerative food growing techniques. There aren’t many projects with a demonstrable impact on policy, but many are clearly thinking about policy areas they want to affect (emergency planning, marine policy, education, policy, conservation policy etc.), judging by project objectives.
Finally, Greece’s landscape may not be at the same level of maturity as that of Belgium and Germany, but the emergence of enabling/supporting projects like INCENTIVE leaves one sanguine about the prospect of this transformation happening sooner rather than later. INCENTIVE aims to transform universities, starting with AUTh, into scientific hubs that would drive CS agenda by promoting science-society interfaces. If successful, INCENTIVE may help evolve Greece’s CS landscape into a buoyant ecosystem with hundreds of projects and opportunities for experienced CS professionals as well as those new to the field.
Conclusion
By critically examining other projects’ results and achievements, we tried to identify learning points that could be used to improve pilot deployment and technical development in COMPAIR. So what have we learned from the review?
Participant engagement and retention: Projects that experimented with different engagement tactics found that participants who attended several co-creation and training workshops tend to stay longer (and may even become ambassadors for the project at a later stage) than those whose engagement can be described as ‘shallow’ e.g. they learn about the project from social media, they have no contact with the organisers, they only follow online instructions on how to assemble a sensor.
Levels of participation: Projects are generally aware of different levels of participation in which citizens can act as sensors (level one), interpreters (level two), data collectors (level three), data collectors and co-creators (level four), with many offering engagement opportunities at level three or above. Very few projects try to extend this standard model by offering engagement at even higher levels, such as when the ownership of a project is transferred to local communities in whole or in part (citizens scientists as co-managers).
Focus on youth: In many projects, schools are a priority stakeholder and sometimes the only stakeholder-cum-participant in CS activity. This shows that health risks posed by air pollution to young children are increasingly becoming more of a concern to schools, parents and policy makers, for whom CS is a means to understand the scale of the problem, raise awareness about the issue, and co-develop strategies to address it, for example through school streets.
Targeting vulnerable groups: Through schools CS projects have potential to engage participants from lower socioeconomic (LSE) backgrounds, but this depends on the type of school (private or public). In general, few projects make it an explicit priority to target hard-to-reach groups, and fewer still manage to demonstrate success in this area. One engagement tactics that worked well in a recently finished project in Brussels is to use local charities that focus on health and poverty issues to identify LSE groups in areas that suffer disproportionately from air pollution.
Results presentation: The use of online maps is by far the most common way of presenting air quality data. Sometimes projects build their own platform to display results, sometimes they use platforms of international initiatives, with Sensor.Community being the most popular example. Sometimes data is accessible via API, sometimes as raw datasets on Zenodo. On a few occasions, projects in Flanders used a media outlet to disseminate findings, with results published as map-based visualisation on a daily’s website.
With this, we conclude our first blog on the general findings of the CS landscape review. The next four articles will present pilot specific results in more detail, starting with Bulgarian initiatives.
Have you participated in any CS projects in these four countries? If so, do share your experience in the comments below.