Measuring Regional Pollution with Low- and Mid-Cost Sensors

Highly precise and accurate greenhouse gas measurements typically require costly instrumentation installations. However, measurements from lower-accuracy, lower-cost sensors can be processed to augment existing measurement techniques.

The accurate measurement of atmospheric greenhouse gases (GHGs) is vital to estimating pollution emitted by cities and metropolitan regions. These GHGs include carbon dioxide (CO₂), methane (CH₄), and many other pollutants that build up in the atmosphere and contribute to climate change. With accurate measurements, scientists can better understand environmental changes and target techniques to reduce emissions. Monitoring GHG concentrations in the atmosphere can be used to estimate emissions from cities, allowing for the evaluation of other emissions products or estimates that rely on various data sets such as fuel use, traffic counts, and other socioeconomic data. While this estimation approach based on socioeconomic activity is widely used, and it is the method employed by the U.S. Environmental Protection Agency (EPA) for the national inventory, in some cases, discrepancies have been found between this approach and methods based on concentration measurements (1). Active research is seeking to better understand the differing results, particularly at very local scales.

Instrumentation packages typically used to measure CO₂ and other GHGs with the accuracy needed to make policy decisions are expensive and can cost more than $100,000 for each measurement site (2). This high price tag limits measurement capabilities in a region, often leaving scientists to estimate emissions using sparse observations.

In regions where direct concentration measurements are available, the observed degree of variability is often quite small, so high-precision measurements are required to detect changes. For example, as of October 2023, the monthly average CO₂ concentration measured at the National Oceanic and Atmospheric Administration (NOAA) Mauna Loa observatory in Hawaii was 418.82 μmol of CO₂ per mol of dry air, or parts per million (ppm). This value represents only a small increase of about 3 ppm from the previous year (Figure 1) (3). Similarly, fluctuations of CO₂ in a city such as Washington, DC can be as small as a few ppm (4).

---

▲Figure 1. The monthly average CO₂ concentrations measured at the National Oceanic and Atmospheric Administration’s (NOAA’s) Mauna Loa Observatory in Hawaii from Jan. 2019 to Nov. 2023 show a steady increase in mean CO₂ concentrations with cyclical seasonal variations. The red dots and lines represent the monthly mean values, and the black bars and lines represent that same concentration after correcting for average seasonal cycling (3).

To distinguish direct concentration changes from small-scale variability, researchers estimate emissions by combining activity-based estimates with direct CO₂ concentration measurements in sophisticated models to estimate ground-level emissions (5). With this approach, activity-based estimates are further constrained by the CO₂ concentration measurements to better represent regional emissions and provide the accuracy needed to detect long-term trends and anomalies. For example, a combination of these estimation techniques allowed scientists to quantify how traffic reduced by COVID-19 lockdowns in the California Bay Area Region reduced regional emissions in 2020 (6).

The added value of combining activity-based estimates with direct CO₂ concentration measurements has encouraged the growth of urban emissions measurement networks across the U.S. (7). As cost is a limiting factor of such networks, scientists are developing approaches to use lower-cost and lower-precision equipment to make concentration measurements and increase the spatial density of measurements throughout a region. This article discusses how a new generation of low- to mid-cost sensors can be used to accurately measure CO₂ concentrations and increase measurement capabilities in an urban emissions measurement network...