About us

Started in 2018, the open-seneca initiative has deployed pilot citizen science networks of low-cost air quality sensors in cities in the UK (Cambridge), Argentina (Buenos Aires and Mendoza), Kenya (Nairobi), and Brazil (Belo Horizonte). The team has in-depth experience in the design of open-source, mobile, low-cost air quality monitors and capacity building in partner cities, the education and engagement of citizens, and calibration, analysis, and visualisation of the data collected with the low-cost sensors to aid with decision making.

open-seneca personal air quality sensor
The open-seneca team 2020

Motivation

Cities around the world are suffering from the negative effects of outdoor and indoor air pollution. According to the WHO, around 7 million premature deaths globally are caused by air pollution every year, 80% of which are from heart disease and strokes, and 20% are from respiratory illnesses and cancers related to exposure to fine particulate matter (PM2.5)1. Recently, a link between the mortality of COVID-19 and long-term exposure to PM2.5 has been made2. It was found that an increase of only 1 µg/m3 in PM2.5 results in a 15% higher death rate, highlighting the urgent need to reduce air pollution levels to protect human health.

open-seneca personal air quality sensor nairobi
Nairobi, Kenya

Air quality (AQ) is usually monitored using large and expensive reference stations, which require regular calibration and costly maintenance. Moreover, reference stations only provide AQ data from one fixed location, which generally is neither representative for an entire city nor for the experience of individuals moving around in the city. In low- and middle-income countries (LMICs) where reference stations are often unaffordable, there is a general lack of appropriate policies to control AQ and a lack of awareness about the negative health effects associated with the often-alarming levels of air pollution experienced by ordinary citizens. According to the 2018 WHO Global Air Quality Database1, 97% of cities in LMICs with more than 100,000 inhabitants do not meet WHO AQ guidelines.

Low-cost sensing

Numerous low-cost AQ sensors have become available in the market in the last decade. If their limitations regarding data accuracy are understood and proper calibration is performed, these low-cost devices become a powerful tool for AQ monitoring in urban environments. Large networks of these sensors can be deployed to supplement existing official monitoring stations, filling spatial and temporal gaps, and their data can be uploaded to global platforms such as the UNEP AirVisual platform and OpenAQ. Moreover, they have shown the potential to obtain air pollution data with a spatial and temporal resolution un-achievable by reference stations. The data from such approaches provide a rich depth of information, which could be used to drive urban planning and policy towards creating a healthier environment.

Reference station Citizen science
+ High accuracy + Enables identification of pollution hotspots
- Expensive + Low cost
- Stationary + High spatial/temporal resolution
+ Engaging and educational

Combining the use of low-cost sensors with a citizen science approach makes them even more powerful by enabling the monitoring of personal exposure and the identification of hotspots of air pollution. Engaging citizens allows raising their awareness about the importance of taking action to reduce air pollution and to empower them as active stakeholders in the search for solutions. A society that is aware can have a huge impact in reducing air pollution: by reducing their own individual emissions, by minimising their personal exposure, and by complying to measures to reduce emissions implemented in cities by local governments.

Our sensors and methodology

The EPA Air sensors guideline3 classifies the performance of AQ sensors in five types (Tiers I-V). The suggested performance goals are summarised in the table below

Tier Application area Uncertainty
I Education and information ±50%
II Hotspot identification and characterisation ±30%
III Supplemental monitoring ±20%
IV Personal exposure ±30%
V Regulatory monitoring ±10%

Open-seneca monitoring devices use the Sensirion SPS30 Particulate Matter Sensor. This sensor has received the MCERTS4 Performance Standards for Indicative Ambient Particulate Monitors for PM2.5 in the range of 0 - 75µg/m3. The uncertainty of this sensor is ±8.9%, complying with the data quality objective for indicative measurements.

According to this classification, the SPS30 sensor used in the open-seneca AQ sensor would be suitable for any of the applications of Tier I-V. The aim of the open-seneca initiative is to empower citizens with data about their personal exposure to particulate pollution (Tier IV), to raise awareness and drive behaviour change (Tier I), and to provide high spatial and temporal resolution pollution maps that highlight hotspots of particulate pollution (Tier II and III) to inform policy. However, in a citizen science setting where the sensors are mobile, the uncertainty required for Tier V might be compromised, and the initiative does not aim to ensure regulatory compliance of cities to current AQ regulations.

Key steps

In addition to providing sensors, we aim to educate. We host workshops prior to each deployment. In each workshop we cover the relevant topics related to air pollution and how it affects our health and our environment. We also cover how one can monitor AQ and how each sensing method works. This then follows an interactive session where we show how one can build a sensor, using one of our open-source designs.

open-seneca personal air quality sensor lecture buenos aires workshop
Introductory lecture given by Matias Acosta in Buenos Aires.

In order to ensure the quality of the data collected with open-seneca AQ sensors, a two-stage sensor calibration is performed. The first stage is done by the manufacturer (laboratory calibration under controlled conditions) and the second one is performed by open-seneca before and after a pilot period (co-localisation to ensure each sensor output is cross-comparable, recommended duration of 2 weeks). By integrating the data collected with the low-cost monitors with the existing regulatory stations, further calibration can be performed in a dynamic way during data collection periods.

open-seneca personal air quality sensor colocalisation pm2.5
Co-localisation (20/04/2020) of a few sensors in Buenos Aires one year after the initial pilot scheme. Data on PM2.5 in µg/m3.

After initial calibration, the mobile AQ sensors are handed over to volunteers to carry for the period of the pilot, to collect geo-tagged air quality data. We recommend a minimum of 1 AQ sensor for every 4 km2 desired covered per month of the pilot. Each volunteer, also called citizen scientists, are asked to follow a strict protocol to ensure data is collected in a consistent manner. The data they collect is protected under GDPR. However, they are welcome to share their routes publically should they choose to, much like on popular sports activity apps (e.g. Strava). Each collection session is viewable on our online interactive platform, and details information about their personal exposure to pollutants, e.g. PM2.5.

open-seneca personal air quality sensor colocalisation pm2.5
Our online platform, showing one publically shared route collected in Buenos Aires in May 2019.

At the end of the pilot, the sensors should be colocated again. From both colocation sessions, each sensor has a final calibration curve applied and any anomalous sensor data removed. The data is anonymised and aggregated to produce maps of the regions covered by the citizen scientists. The aggregated maps enable the identification of hotspots in the covered region. Anonymised data products are made public and passed onto policy makers.

open-seneca personal air quality sensor city map buenos aires pm2.5
Aggregated map, after processing, highlighting hotspots in the city of Buenos Aires from May 2019.

Summary of the key steps:

Educational workshops

  • Lecture on air pollution and health
  • Introduction to sensing method and sensor
  • Interactive session, building sensors in groups

Data collection

  • Sensor co-location and calibration
  • Sensors carried by volunteers for period of project
  • Data uploaded by volunteers onto our online interactive platform

Evidence based policy

  • Aggregated data processed
  • Results passed onto local policy makers

Mobile air quality sensors

We currently use off-the-shelf components interfaced with a 'plug & play' style PCB. They aim to be simple to build and versatile enough to be applied to any setting, and to be modular and accesible to everyone. Our designs are open-source and available from our public GitHub.

open-seneca personal air quality sensor pcb pm2.5
Current open-seneca plug & play PCB.

Below details the raw breakdown cost of an open-seneca sensor, rounded to the nearest £0.50 (prices subject to change, as dependent on suppliers/exchange rates, doesn't include labour):

Item Model Raw cost / £
Core components
Plug & Play PCB v2, with integrated
- BLE (HM11)
- Temperature/humidity sensor (SHT31)
*25.00
MCU STM32F103 Black Pill 2.50
PM sensor Sensirion SPS30 + cables 25 - 40.00

Optional modules
VOC sensor Sensirion SGP30 10.00
GPS Generic via UART 5 - 20.00
MicroSD reader Generic 0.50
MicroSD card Generic 2.50
GSM SIM868 10.00
RTC DS3231
1.00
OLED display 0.96" SSD1306 2.00
Powerbank Generic > 10.00
Case 3D printed, or junction box > 5.00

*cost depends on volume, but hovers around this price for 20 units.
similarly depends on volume, check here for latest prices.

The device requests data every second from each respective sensor, and stores it on the microSD card. The requested data is similarly transmitted via BLE or GSM. Additional sensors and other common IoT communication protocols (e.g. WiFi, LoRa, Sigfox, NBIoT) can be added upon request. We currently have a basic Android app to interface with our sensors via BLE, which can be found here.

Hope you enjoyed reading about us, please feel free to contact us for further information.


1https://www.who.int/airpollution/data/cities/en/
2https://projects.iq.harvard.edu/covid-pm/
3https://www.epa.gov/air-sensor-toolbox/how-use-air-sensors-air-sensor-guidebook#pane-1
4https://www.csagroupuk.org/wp-content/uploads/2020/01/MC20035000.pdf