The systematic literature search was conducted in the principal database, Google Scholar and other databases of PubMed/PMC, ScienceDirect, ACS publications, MDPI journals and Springer nature Link up to 2026. Geographic terms (“ECOWAS”, “West Africa” and the fifteen West African countries) were included in the search along with the monitoring and innovation terms “air quality monitoring”, “low-cost sensor”, “PM_2.5”, “sensor calibration”, “satellite remote sensing”, “machine learning”, and “citizen science”. Reference lists of retrieved articles of review articles were hand-searched (snowball/reference-chaining approach), following previous regional air quality management strategy scoping reviews conducted throughout Africa [4], to identify any missed eligible studies.

Studies were included if they described the development of air quality monitoring infrastructure, technology, or methodology in or covering one or more member countries of ECOWAS; if they were peer-reviewed journal articles or peer-reviewed conference proceedings or credible institutional/grey literature (such as WHO, Climate and Clean Air Coalition); or if they could be evaluated based on the nature of the innovation being described. Only studies that contributed methodological innovations that were directly relevant to the comparative discussion of methodological innovations for ECOWAS countries were included in the analysis if they did so, and were otherwise excluded from the analysis unless relevant to East Africa for direct evidence of methodological innovation (e.g., citizen science case studies) were offered.

The titles and abstracts from the search were screened for relevance and shortlisted abstracts were full-text evaluated. The information from a total of 34 sources was considered suitable and was retained for a narrative synthesis including peer-reviewed articles in journals including Environment International, Environmental Science & Technology, ACS ES&T Air, ACS Earth and Space Chemistry, Atmosphere, Aerosol and Air Quality Research, Environments, Discover Atmosphere, Discover Sensors, PLOS Climate and the International Journal of Environmental Research and Public Health, in addition to suitable institutional sources. Data gathered at each study country/city included information on information collected, monitoring technology, calibration method (if any), performance measures reported, and barriers. As a common approach for scoping-style systematic reviews of environmental monitoring interventions in low-resource settings, a narrative and tabular synthesis was employed due to methodological diversity of studies (social science/quasi-experimental designs, diverse outcome concepts and geographic scope).

All published sources agree that Senegal is the country in the region with the weakest institutional monitoring of air quality. Senegal is the only ECOWAS country that was identified in several reviews as having a continuous government-operated air quality monitoring system, namely with the Centre de Gestion de la Qualité de l’Air (CGQA) that is monitoring at six fixed sites throughout the Guédiaway, Médina, Yoff, Bel-Air, HLM and Cathedral districts [3][5]. Ghana has also made more modest but significant strides with the long-running U.S. Embassy reference-grade monitor in Accra, and recently, an Afri-SET reference site for the support of calibration research was established [6][7]. In the past, monitoring capacity in Nigeria has been limited, with the United States Embassy operating the two reference-grade stations in the capital city of Abuja and in the most populated city of Lagos, increasing coverage of stations with the deployment of low-cost sensors in the early 2020’s [3][8]. The literature reviewed also did not show that the other ECOWAS countries (Cote d’Ivoire, Togo, Burkina Faso, Niger, Guinea, Guinea-Bissau, Mali, Benin, Sierra Leone, Liberia and The Gambia) have any sustained government-based monitoring network, nor the West African cities scoping review [3].

The most common innovation reported in ECOWAS countries and sold under multiple different names was low-cost optical particle sensors, mainly using the Plantower PMS5003/PMS5003T sensor element [6]. There was strong correlation for all the low-cost monitors with a reference-grade Teledyne monitor for PM_2.5; however, they were all biased high, with the lowest mean absolute error being obtained by the QuantAQ (3.04), followed by PurpleAir (4.54) and Clarity Node-S (13.68) [6]. An AirGradient OpenAir PM monitor was also deployed at an Afri-SET reference site in Accra, which showed a similar temporal trend to the reference instrument, and a correlation between the patterns of PM_2.5 observed in a neighbourhood in Accra and charcoal and fuelwood burning, and traffic [7].

The number of cities that have conducted low-cost sensor studies has increased in Nigeria [8][9]. Pollution measurements were done for particulate matter and gaseous pollutants in a semi-urban community of Ile-Ife, over a two-month period, and the pollution sources were allocated to residential or industrial activities, based on conditional bivariate probability functions [8]. Low-cost optical sensors quantified PM_2.5 in Akure, in Ondo State, in the Harmattan dry season, at two urban sites, from approximately 14 to more than 590 micrograms per cubic metre, and found clear evidence of local emission enhancement [10]. It was noted that Atmotube PRO and PurpleAir sensors had good relationships at both household and school level with a median ratio of 0.93 between indoor and outdoor PM_2.5, and there was good consistency of the data over time, although there was no regulatory-grade sensor in Nigeria to validate the data from the low-cost sensors [11]. As part of a project funded by the US Department of State, 30 low-cost sensors and a reference-grade sensor for local calibration were installed in Lagos, representing one of the widest and most available sources of air quality information in the city [12].

The second, five-site PurpleAir network in the region, in Lomé, Togo, resulted in the first multi-year ambient PM_2.5 dataset for the city, with a network-wide mean seasonal increase between December and February of up to 58 percent due to the Harmattan, and a mean daily concentration of 23.5 micrograms per cubic metre [2]. The first systematic spatial and temporal characterisation of PM_2.5 in Burkina Faso was conducted using a low cost sensor network (19 sites) and a reference grade co-located sensor for field calibration [13] and a follow-on study on the implementation of low-cost sensor networks in three countries, including install and maintenance of sensors for field calibration, field, semi-urban and rural deployments of factory-calibrated sensors, and remote area deployments in Guinea [14].

The feasibility of the long-term monitoring of the low-cost PM_2.5 monitors was tested with 15 PurpleAir-II-SD monitors set in eleven cities in the ECOWAS countries in Sub-Saharan Africa, and the result shows that all the cities have annual mean concentrations above the WHO guideline [15]. Low-cost monitoring platforms such as AirQo established at Makerere University in Uganda have been replicated in Lagos and Accra as part of a network of over 200 sensors across the various cities in Africa [16].

The low-cost sensors are sensitive to humidity and temperature, as well as the composition of the particles, which will be an important research priority. However, Gaussian Mixture Regression was applied to calibrate a PurpleAir sensor in Accra, which was co-located with a reference PurpleAir sensor (BAM-1020) for a full year, and achieved a moderate correlation (r² = 0.53) and high bias [6] with the uncorrected data. At the regional level, AirQo developed and tested a set of machine-learning calibration models (k-nearest neighbours, support vector regression, ridge and lasso regression, XGBoost, random forest, gradient boosting and others) and packaged the best-performing model into an open-source tool, AirQalibrate [17][18], specifically designed to support low-cost sensor calibration in data-scarce Sub-Saharan Africa without users needing to maintain their own reference-grade instruments. The non-linear calibration approach with AI has been used in the past for low-cost PM_2.5 and PM10 calibration in other LMICs, and has been shown to be more effective than simple linear regression in cases where the aerosol mixture is more complex, such as in cities in West Africa [18].

With the lack of ground-based reference monitors, a number of studies have used aerosol optical depth (AOD) derived from satellite data with reanalysis meteorological data to estimate PM_2.5 levels in regions without direct measurements. A multi-model study developed several machine learning algorithms such as Extreme Gradient Boosting (XGBoost), trained them with available surface PM_2.5 observations to generate daily estimates of PM_2.5 at 1 km resolution across West Africa with near-universal exceedance of the WHO air quality guidelines over the entire region [19]. The multiple linear regression (MLR) and the artificial neural network (ANN) models were used to estimate the AOD at the national scale in Ghana using MODIS satellite data over the period 2003-2019 [20]. The first 48-h air quality forecasting system for the megacities of Dakar, Senegal, was developed using four different machine learning ensemble models (Random Forest, Extra Trees Regression, XGBoost, and CatBoost) with reference to satellite Earth observations, as described in the paper [21]. As air quality research moves towards the natural sciences, and satellite data become more widely used for African air quality, yet another cross-continental air quality research telecoupling review revealed that the capacity and authorship of satellite data applications and integration into air quality research in Africa continue to be largely driven by the interests and expertise of natural scientists and scientists outside of Africa [22].

However, the use of citizen science applications is not as well developed in the ECOWAS region as in East Africa, where working similar models have been taken up and can be adapted to the region. The air quality study was a citizen-science effort that involved installing the optical particle sensors and a community engagement workshop to build capacity for the community to interpret and respond to the air quality data they collected, with the aim of creating a national air quality stakeholder network [23]. The SEI-led project in the Mukuru settlement, Nairobi, trained community champions to use personal monitors and to share the results with community and policymakers [24]. Further examinations of citizen-collaborative air quality platforms also highlight the increasing potential for distributed citizen monitoring, thanks to the proliferation of low-cost, Wi-Fi-enabled sensors, and the fact that hybrid platforms that incorporate multiple data sources and leverage AI and IoT technologies are best suited to citizen monitoring and public forecasting [12][25]. Community-oriented deployments have so far been integrated into low-cost sensor projects funded by donors in ECOWAS (e.g., in Lagos and Accra), and there is potential for more of this type of innovation.

The status of government reference monitoring and some of the innovative monitoring programmes identified in the literature for the ECOWAS member states or group of states are summarised in Table 1 and Table 2.