A decade of research with low-cost air quality sensors: air pollution insights, key learnings, and the road ahead

Over the past decade, I have had the privilege of collaborating with multiple teams in the USA, Europe, Africa, and Asia on projects to develop, characterise, and use low-cost sensors for air quality studies. These experiences and lessons for the future will be summarised in this talk.

In 2015, we partnered with SenSevere (now part of Sensit) to develop the Real-time Affordable Multi-Pollutant (RAMP) monitor, showing that startups can quickly enter this field with innovative ideas. However, extensive field testing by an academic group was required to make the devices usable, following which a network of 50 RAMPs was deployed across Pittsburgh, Pennsylvania. The high spatial density, high time resolution monitoring enabled by the sensor network revealed the impact of a large point source outside the city, while also highlighting hyperlocal pollution and street canyon effects. The extended deployment also showed that electrochemical sensors for nitrogen dioxide have a relatively high detection limit (~15 ppb) at odds with laboratory data and need to be replaced annually.

Hurricane Maria critically impaired traditional monitoring in San Juan, Puerto Rico, but the portability and low power requirements of sensor-based devices enabled rapid, solar-powered deployments of RAMPs that found significant sulfur dioxide pollution from generator usage across the city.

Sensing with RAMPs and Modulair-PM nodes and on-site observations identified hyperlocal sources of pollution at stadiums in Qatar and justified mitigation actions to ensure that football fans breathed cleaner air. Multi-year sensor-based monitoring in Kigali, Rwanda; Nairobi, Kenya; Abidjan, Côte d'Ivoire; and Accra, Ghana (coupled with CHIMERE air quality modelling over East Africa and 3D satellite data over West Africa) identified key sources of local air pollution, the influence of regional transport including Saharan dust in West Africa, and the need to develop local emission inventories.

The West African and Middle-Eastern experiences also showed the inability of low-cost PM sensors to detect supermicron dust. Evaluations at the first India Sensor Evaluation and Training (Indi-SET) facility in Bengaluru, India over 2024 showed that more expensive OPCs can better detect supermicron construction dust, but using similar internal sensors does not guarantee similar performance across device integrators.

Field collocation with reference monitors is ideal to improve sensor data quality, but this is not always possible especially in the Global South. Global reanalysis data sets can help reduce known artifacts in PM sensors. Establishing more facilities like Afri-SET (Accra, Ghana) and Indi-SET can also help improve sensor data quality.

This work would not be possible without international collaborations and networks like AfriqAir, CAMS-Net, ASIC, IGAC/Allin Wayra, and WMO/GAFIS. Collaborators, especially early-career researchers, will be acknowledged on the respective slides.

I will conclude with key learnings and prospects for the use of low-cost air quality sensors to address the grand challenge of clean air for all.

(Note: I have no financial interest in any sensor company.)