Filling gaps in Africa’s air quality sensor network through a successful public-private partnership
By Dr Amelia Hilgart, SAEON uLwazi Node
By Dr Amelia Hilgart, SAEON uLwazi Node
Fine particulate matter (PM2.5) can cause a range of human health and environmental impacts at high concentrations, including various respiratory symptoms, aggravated asthma, irregular heartbeats, nonfatal heart attacks and, potentially, death in people with heart or lung disease. Children and the elderly, as well as people with heart or lung disease, are most likely to be impacted (EPA, 2020).
PM2.5 is the measurement of fine inhalable particles, with diameters that are generally 2.5 micrometres and smaller. Domestic fuel burning is the major source of PM2.5 in Africa, followed by traffic and natural sources such as dust. In other regions of the world, traffic and industry are the major sources. PM2.5 is a common measure of air quality, but it normally requires ground-based sensors to measure.
In the environment, particulate matter can be responsible for depleting soil nutrients and changing the nutrient balance in coastal waters and large river basins as well as making lakes and streams acidic and contributing to acid rain. This negatively impacts ecosystems by affecting biodiversity and damaging sensitive forests and farm crops (EPA, 2020).
There is a global effort, led in part by the Paris Climate accords and the Sustainable Development Goals Program (SDG) through indicator 11.6.2: Annual mean levels of fine particulate matter (such as PM2.5 and PM10) in cities (population weighted) to monitor and reduce PM2.5 output. Through the South African Air Quality Information System (SAAQIS), PM2.5 and other aerosols are monitored.
However, there are many gaps in the South African monitoring network which leave data for the country rather sparse. In addition, many of the monitoring stations are not reporting (Figure 1).
Figure 1. A comparison of SAAQIS sensor network locations (A) and PM2.5 model locations across South African population centres (B). In Figure A, the SAAQIS monitoring stations show air quality (yellow and green) and station inactivity (grey). In Figure B, the focus of the modelling collaboration was the impact of PM2.5 on the human population; the colour of the circles represents the air quality, and the size of the circle represents the population of the population centre.
Last year an opportunity arose to fill some of these data gaps in South Africa and across the greater African continent when Zindi launched a data science competition through their platform to detect and predict PM2.5 using a combination of remote sensing and weather data. Once the competition was completed and a winning solution submitted, the data science teams from uLwazi and Zindi discussed ways to distribute the resulting data and generate useful data products based on the model outputs.
Through the gap analysis in the South African Risk and Vulnerability Atlas (SARVA), it was resolved that higher resolution air quality data was needed to better support the South African District model and SDG Indicator 11.6.2. The winning model was then used to predict air quality for 73 cities as well as 496 population centres with population densities greater than 1 000 people/km2 over more than 2 km2.
Figure 2. A comparison of sensors across the African continent from the World Air Quality Index project (A) and PM2.5 model outputs from 1 466 population centres across the African continent (B).
As seen in Figure 1, this fills gaps in the sensor network particularly in the northwestern and southwestern parts of the country. This analysis was also conducted across the African continent for a total of 1 466 population centres, filling in gaps across the continent but especially in Sub-Saharan Africa and along the Mediterranean coastline (Figure 2).
These data were then added to an online dashboard (Figure 3) so that the results could be easily viewed and as a touch point for the raw data. The data and model details were published to SAEON’s Open Data Platform (ODP) as well as to the larger African spatial data community through ESRI’s Africa Geoportal.
For additional details about the collaboration, see Zindi’s article Zindi and SAEON fill in the air quality data gaps for African cities.
Figure 3. Air quality dashboard generated to allow users to interact more easily with the data.
The World Air Quality Index project. (2020). World’s Air Pollution: Real-time Air Quality Index. Waqi.Info. https://waqi.info/
US EPA, O. (2016, April 26). Health and Environmental Effects of Particulate Matter (PM) [Overviews and Factsheets]. US EPA. https://www.epa.gov/pm-pollution/health-and-environmental-effects-particulate-matter-pm
Department of Environmental Affairs. (2020). South African Air Quality Information System. http://saaqis.environment.gov.za/
Whitaker, J., Ayami, Y., Bray, A., Mishra, N. K., Cossentini, K., Haddar, A., Ksouri, A. & Klai, M. H. (2020). Daily air quality estimates for urban centers in Africa [Data set]. Zindi. https://doi.org/10.15493/SARVA.301020-2
Zindi. (2020a, April 22). Zindi solutions: A useful open-source model of urban air quality for Africa | by Zindi | Medium. Medium. https://zindi.medium.com/zindi-solutions-a-useful-open-source-model-of-urban-air-quality-for-africa-709a5b15f107
Zindi. (2020b, December 10). Zindi and SAEON fill in the air quality data gaps for African cities. Medium. https://zindi.medium.com/zindi-and-saeon-fill-in-the-air-quality-data-gaps-for-african-cities-566662e0ddea