2025 Theses Doctoral

Measuring Everything, Everywhere, All At Once: Developing Low-Cost Sensors, Tools and Algorithms for Measuring Pollution

Raheja, Garima

Air pollution is a major threat to human and environmental health around the world, ranking as the second leading contributor to the global disease burden. Traditional monitoring of air pollution has relied heavily on reference grade monitors costing more than USD$100,000 for purchasing, transportation, and climate control. This cost creates a high barrier for monitoring in low-resource areas, many of which are most severely impacted by air pollution.

Recently, there has been a rise in “low-cost sensors” with initial cost of hardware priced in the USD$250-25000 range. However, data from these low-cost sensors often needs to be corrected, and the development of this correction has traditionally relied on a colocation with a local reference-grade monitor, which then returns the user to the high cost problem. In this dissertation, I seek to fill a number of gaps in the existing literature and study of environmental pollution, by

-exploring the different technologies being used to study air pollution,
-proposing calibration methods for bettering these technologies,
-explaining what these measurements show us about the changing nature of cities and the broader state of the planet,
-exploring the applications of low-cost sensors for soil pollution monitoring, and
-suggesting scientific and policy instruments towards fostering a livable future

In Chapter 2, I collaborate with researchers across Ghana to present the first intercomparison of low-cost sensors in Africa finding that the network daily average PM2.5 concentration in Accra was 23.4 μg/m3, which was 1.6 times the 2021 World Health Organization Daily PM2.5 guideline of 15 μg/m.

In Chapter 3, by collaborating with scientists in Togo to deploy a network of five low-cost PurpleAir PM2.5 monitors over 2 years (2019–2021), I present the first multiyear ambient air pollution monitoring data results from Lomé, Togo, showing that the strong regional influence of the dry and dusty Harmattan wind increases the local average PM2.5 concentration by up to 58% during December through February.

In Chapter 4, I work with a large coalition of scientists, activists and policymakers to compare measurements from a citizen-led participatory low-cost, high-density air pollution sensor network in Belmont County, Ohio to regulatory-grade monitors; this network analysis combined with complementary models of emission plumes reveal the inadequacy of the sparse regulatory air pollution monitoring network in the area, and open many avenues for public health officials to further verify people's experiences and act in the interest of residents' health with enforcement and informed permitting practices.

In Chapter 5, I develop a universal correction model for optical low-cost PM2.5 sensors by creating a training dataset using 101 colocations from different regions, atmospheric conditions, and sensor types around the world using the Gaussian Mixture Regression machine learning model.

In Chapter 6, I pivot from low-cost air pollution monitoring to low-cost soil pollution testing, showing results from the MINESCAPES Summer School in June 2024 in the Harz mountains in Germany, seeking to highlight the pedagogical benefits of combining history, physical sciences and biological sciences in a field study of mining landscapes, as well as the capacities of a low-cost soil lead testing kit for training and interdisciplinary exchange.

In Chapter 7, I list a number of the collaborative projects, such as policy papers I led or scientific papers that I co-authored during my PhD, discussing how they fit into the broader effort to democratize air pollution monitoring and enhance the quality of data from various technologies in different regions. In Chapter 8, I discuss the outlook of the field of air pollution monitoring, and suggest areas of future research and development.