top of page

COMPAIR Air Quality Sensor Strategy for Citizen Science

Updated: Jun 6, 2023

Having a well-planned sensor strategy is critical to the success of a citizen science project. Using appropriate sensors and monitoring techniques can help ensure consistent data collection and data quality, ultimately making citizen science data suitable for use in policy making and scientific research. All these considerations are firmly embedded in COMPAIR, where the goal is to collect hyperlocal data on air quality and traffic in different neighbourhoods around Europe.


In a new blog mini series, we will set out our sensor strategy, explaining what will be measured, how, and for which purposes. In this first article, the reader will learn about different air pollution types and how we are planning to measure them in COMPAIR.

Hand touching an air quality sensor device

In recent years, citizens are getting more and more aware about the adverse effects of poor air quality and are willing to take actions to improve it. The first step in this direction is knowing about air quality and atmospheric pollution around them. Advancements in sensor technologies and development of the Internet of Things (IoT) has made it possible to significantly reduce the cost of portable sensing devices, which can measure anything from particulate matter (PM) and nitrogen dioxide (NO2), to temperature and humidity, black carbon (BC), and even ozone (O3). All these will be discussed in turn along with their relevance and possible measurement in COMPAIR.


Particulate matter

PM can be defined as the air-suspended mixture of both solid and liquid particles. Generally, they are classified as coarse, fine and ultrafine particles. Particles having a diameter of between 10 µm and 2.5 µm (micrometres) are classified as coarse particles. Coarse particles settle relatively quickly whereas fine (1 to 2.5 µm in diameter) and ultrafine (<1 µm in diameter) particles stay suspended in air for longer. To get a gauge of the sizes of particulate matter, one can consider the size of a human hair which has an approximte diameter of 50-70 µm and PM10 is referring to particles smaller than 10 µm. These particles include dust, pollen and mould spores. PM2.5 refers to particles smaller than 2.5 µm and comprises mostly combustion particles, organic compounds and metals.


The main source of PM is both human and natural resources. Though it is important to study natural sources like forest fires, pollen, mould etc., it is essential to understand the human generated sources and its impact on the environment as well as human health.


When it comes to measuring PM air pollution in urban environments, it is important to focus on PM2.5 as they have been found to be most detrimental to human health. In recent studies, particles of diameter less than 2.5 μm have been observed in the brain and leading to neurodegenerative diseases (Maher et al, 2016).


As part of COMPAIR it is an essential requirement to measure particulate matter as part of the air quality monitoring. Fortunately, low cost sensor technology has given rise to several off-the-shelf PM sensors that can be used to create affordable but effective PM monitoring devices. However these sensors need to be assessed before they can be used for the development of an air quality monitoring device as they often show degradation depending on the environmental conditions.


There are several studies available that have evaluated different sensors in the market and summarised their findings. Based on the requirements set by the COMPAIR pilots for the air quality monitoring device (portable yet accurate device), it can be seen that a good PM sensor has to qualify on 4 key criteria that are as follows:

  • The sensor should have low power consumption during the data acquisition phase so that it can be used in an IoT environment

  • The sensor should be able to operate reliably under different ambient conditions

  • The sensor should have low integration complexity with data acquisition hardware and should not require special conditioning components

  • Finally, a good sensor would be able to report information other than PM values such as measurement errors, resolution etc.


Based on the criterias set above, two sensors that are found to meet all the specs are the SPS30 by Sensirion and PM2012 by Cubic sensors ltd. The basis of this selection was the research published by the Italian National Agency for New Technologies that has compared around 50 low-cost PM sensors and evaluated them on several criterias such as accuracy as well as usability of sensors for development of IoT devices (Alfano B. et al, 2020). In order to be sure about our sensor choice for PM, we looked at a study by RIVM (Dutch National Institute for Public Health and the Environment). Over the years, RIVM has tested several sensors such as Shinyei PPD42, SPS 30 and Nova Fitness SDS011. RIVM recommends SPS30 to be used in the future as it is very good at measuring PM2.5 values (RIVM, 2021). Based on these studies SPS30 seems to be one of the best sensors to build a portable device that can be used for measuring air quality. Another advantage of SPS30 is that it has been found to be quite reliable when it comes to measuring PM values even when the environment is continuously changing (device containing the sensor is moving) due to its short sampling period of 1s.


Nitrogen dioxide

Nitrogen dioxide (NO2) is one of the six widespread air pollutants that are included in the international air quality standards. It is formed during combustion processes, such as fuel combustion from vehicles or industrial combustion from power plants and boilers (Chauhan et al, 1998). Indoors, NO2 can also be emitted by kerosene or gas space heaters and gas stoves in substantial amounts (Hesterberg et al, 2009). Exposure to high levels of NO2 can lead to a number of adverse effects including inflammation of the airways, reduced lung function and asthma, and is also linked to cardiovascular harm with significantly increased harm in susceptible populations (Kawamoto et al, 1993). Urban areas with high vehicle density are of particular concern for elevated NO2 levels. Such local pollutants may not be captured at the source by the sparsely distributed reference grade measurement stations deployed by the government and health institutes.


Low-cost sensor technologies are becoming widespread for measuring NO2 pollution in a fine-grained manner. However, the sensors suffer from susceptibility to environmental conditions (temperature, humidity) and cross-reactivity to other pollutants, leading to sensor drift and decreased sensitivity over time. This decreases the reliability of measured values and thus makes off-the-shelf low-cost sensors less suitable for a citizen science deployment.

Several studies are available in literature that aim to evaluate various low-cost NO2 sensors. However, comparison of sensor performance using a multitude of studies or reports is often challenging since sensor performance can be greatly influenced by the electronic architecture of the sensor unit, design of sensor enclosure or the deployment conditions e.g. exposed concentration, presence of interfering pollutants, length of study. For the selection of NO2 sensor to be integrated in COMPAIR, we focus on the LIFE VAQUUMS project led by the Flemish Environmental Agency which is a comprehensive study that covers selection of sensor components, their lab and field evaluation with a long field testing period exceeding a year.


During the project, five low-cost NO2 sensors were selected and tested out of 10 candidates based on price, availability and expert advice. The sensors were tested both under laboratory and field conditions, by placing the low-cost sensors next to a reference-grade chemiluminescence analyser (42i ThermoFisher Scientific) operated according to standard EN14211. During the 400-day field test, sensors were tested according to data correlation with reference grade sensor, data availability, between-sensor uncertainty and expanded uncertainty, prior to and after calibration. When all of the quantified performance indicators are considered, the Alphasense NO2-sensor B43F demonstrated acceptable performance.


The EU ambient air quality directive 2008/50/EG, which does not have a framework for evaluation of low-cost measurement methods, states a maximum relative expanded uncertainty for indicative NO2 measurements at 25%. Few of the sensors, when evaluated at the hourly limit value, meet this measurement uncertainty target. The CEN/TC264/WG42 standardisation working group prepared a Technical Specification (TS 17660) that describes the test protocol, Data Quality Objectives and performance requirements. The classification includes three classes, of which the lowest class (not related to regulatory measurements) can be used for citizen science projects and requires an expanded uncertainty of less than 200% at limit value (Certification of Sensors System for Air Quality Monitoring | Ineris Services, n.d.).


Alphasense sensor NO2-B43F was also evaluated by the National Institute for Public Health and the Environment of the Netherlands (RIVM) in a 2017 study, which showed similar performance and stated that despite the existing limitations, the sensors can provide meaningful information (Wesselink, n.d.). Therefore we select this sensor to integrate into a newly to be developed air quality device. Combined with the optimum sensor unit electronics, design and calibration, the sensor can provide valuable insights regarding traffic and gas heating related pollutants in the citizen science deployments.


Temperature and humidity

An air quality measurement device is incomplete without integrating it with ambient temperature and humidity sensors. These values are not only required for user experience but are also valuable data points for researchers who want to run advanced algorithms to gain valuable insights and patterns. There are several types of temperature and humidity sensors available in the market which are suitable for the current use case. For selection of the best suitable sensor we focused on only low-cost sensors that can be reliably integrated with IoT devices without adding much hardware complexity and their ability to be used in power savings mode when they are not measuring data.


Based on these criteria and our experience with using temperature sensors in past projects, we shortlisted BME280 and SHT31 as the most suitable sensors. As performance and ease of usage of both sensors are similar, the final choice between BME280 and SHT31 will depend on their global availability as currently there is a global shortage of temperature sensors after the COVID pandemic. A quick scan of the market showed that BME280 is currently having very long lead times for procurement which makes SHT a better choice. Both sensors can have very short sampling intervals making them suitable for measuring temperature and humidity even while moving.


Black carbon

BC (also known as soot) is formed by an incomplete combustion of fuel, such as emissions from diesel engines, cook stoves and wood burning heaters. BC particles are good at absorbing sunlight, hence the black colour. BC is one of the major contributors to climate change, possibly second only to CO2.


Some COMPAIR pilots are keen to measure BC in their citizen science campaigns. One sensor under consideration is the BC meter developed by the Environmental Action Germany. The device measures BC concentration by sampling airborne particles through a filter and then monitoring the attenuation of light in the collection area. The first evaluation of BC sensor data showed that the trial dataset was too limited to draw generalisable conclusions on whether the sensor is fit for purpose. In particular, it was hard to distinguish BC emissions from diesel vehicles and wood burning. Another challenge is that unlike the other sensors considered by the project (e.g. PM, NO2, traffic), the BC sensor may need frequent maintenance, such as changing filters and checking attenuation, while the integration of sensor data into COMPAIR platforms (e.g. policy monitoring dashboard) may prove to be challenging. The decision on whether to include BC sensors in COMPAIR citizen science campaigns is still pending.


Ozone

Ground-level ozone (O3) is a harmful air pollutant and is the main ingredient of smog. It is created by chemical reactions between nitrogen oxides (NOx) and volatile organic compounds, accelerated by the presence of solar radiation and high temperatures. Exposure to elevated concentrations of ozone is associated with a wide range of respiratory diseases such as pneumonia, asthma and allergic rhinitis (Ebi, K. L., & McGregor, G. 2008). Five types of ozone sensors from various suppliers were also evaluated and field tested in the LIFE VAQUUMs project mentioned in the previous section. Considering all the evaluated parameters (reference correlation, between-sensor uncertainty, data availability and expanded uncertainty), the Alphasense OX-B431 shows promising results in the field tests. It is noted that the Alphasense sensor measures the sum of NO2 and O3 gases, therefore in order to derive the ozone concentration NO2 measurements also have to be included in the monitoring plan.

Since ozone can react with compounds emitted from vehicles (NOx) and get removed from the air, ozone pollution is more prominent in rural areas than in cities. Therefore a critical evaluation needs to be performed to determine whether adding ozone measurements to the urban microenvironment monitoring in the COMPAIR project will provide significant insights on air pollution.


References

Alfano B., Barretta L., (2020) A Review of Low-Cost Particulate Matter Sensors from the Developers’ Perspectives https://doi.org/10.3390/s20236819


Chauhan, A. J., Krishna, M. T., Frew, A. J., & Holgate, S. T. (1998). Exposure to nitrogen dioxide (NO2 ) and respiratory disease risk. Reviews on Environmental Health, 13(1–2), 73–90


Certification of sensors system for air quality monitoring | Ineris services. (n.d.). Prestations.Ineris.Fr. https://prestations.ineris.fr/en/certification/certification-sensors-system-air-quality-monitoring


Ebi, K. L., & McGregor, G. (2008). Climate change, tropospheric ozone and particulate matter, and health impacts. Environmental Health Perspectives, 116(11), 1449–1455. https://doi.org/10.1289/ehp.11463


Hesterberg, T. W., Bunn, W. B., McClellan, R. O., Hamade, A. K., Long, C. M., & Valberg, P. A. (2009). Critical review of the human data on short-term nitrogen dioxide (NO2 ) exposures: Evidence for NO2 no-effect levels. Critical Reviews in Toxicology, 39(9), 743–781. https://doi.org/10.3109/10408440903294945


Kawamoto, T., Matsuno, K., Arashidani, K., Yoshikawa, M., Kayama, F., & Kodama, Y. (1993). Personal exposure to nitrogen dioxide from indoor heaters and cooking stoves. Archives of Environmental Contamination and Toxicology, 25(4), 534–538. https://doi.org/10.1007/BF00214345


Maher B., Imad A. M., Karloukovski V., (2016) Magnetite pollution nanoparticles in the human brain https://doi.org/10.1073/pnas.1605941113


RIVM (2021) Sensoren voor luchtkwaliteit. https://www.rivm.nl/burgerwetenschap/samen-meten-aan-luchtkwaliteit


Wesselink, J. (n.d.). Meten van NO2 Met Goedkope Sensoren: https://www.samenmetenaanluchtkwaliteit.nl/document/rivm-notitie-meten-van-no2-met-goedkope-sensoren

41 views

Comments


bottom of page