High Value Air Quality Data From Inexpensive Sensors

The quality of air we breathe affects the quality of the lives we live. Clean air is essential for good health, and bad air quality can lead to hazards ranging from poor outdoor visibility to cardiovascular illnesses. To help researchers and decision-makers assess and protect air quality, NASA routinely gathers data from an assortment of space-based, airborne, and ground-based instruments. 

Recently, NASA’s Earthdata began offering a new collection of measurements that combines data from small, inexpensive, ground-based air quality sensors around the world. The collection, called the “Low-Cost Air Quality Sensor Harmonization Database,” can complement measurements made by satellites and provide unique, hard-to-acquire ground-level data for pollution studies.

"Low-cost" is part of the name because the data come from single or multi-sensor devices costing $1,000 or less. The devices can measure a range of gases and pollutants in the air, from carbon monoxide and nitrogen dioxide to particulate matter. The sensors are used by a variety of people and groups, including college researchers for local studies and student training, city governments for pollution monitoring, and curious citizen-scientists installing sensors in their own backyards. The networks may span a neighborhood or an entire nation.

“The organizations and projects running these networks do a lot of quality assurance work to make sure data from these low-cost sensors are scientifically reliable,” said Kristen Okorn, an air quality scientist at NASA’s Ames Research Center in California, who is leading development of the database. “We wanted to unify the data and place them in the same archive that so many people visit for Earth science satellite and modeling data.”

The result is a database of lesser known but high-quality, interoperable, ground-level air quality data that can be found in the same archive as NASA’s marquee atmospheric datasets.

If You Build It, They Will Come

The database originated with a low-cost system Okorn deployed herself, called Inexpensive Network Sensor Technology for Exploring Pollution (INSTEP). That network is made up of 30 lunchbox-sized air quality sensor packages placed in urban areas around California. The goal of the project was to test the ability of the inexpensive sensor package to gather scientifically useful data.

“After I built the network, other scientists were increasingly reaching out to me and asking if I had a sensor that they could use for data comparison or for validating satellite measurements,” Okorn said.

The frequent requests told Okorn there was an appetite for this type of data, so in the summer of 2024 she started developing a database of networks that are similar to the one she operates. 

Okorn found funding for the database through the NASA-sponsored Global Learning and Observations to benefit the Environment (GLOBE) program and NASA's Early Career Research (ECR) program as part of the Early Career Investigator Program (ECIP).

"ECR supports outstanding next-generation researchers across every element of the Earth Science Division,” said Yaítza Luna-Cruz, program manager for ECR at NASA Headquarters. "The future of Earth science is being written by the people we empower today, such as Dr. Okorn." 

Okorn initially reached out to colleagues to see if they were interested in sharing their data and to ask how to best do it. One key was harmonizing the data; that is, taking observations from different sources and reformatting them in a standardized way so that they can be used together. The researchers decided to format the data using the International Consortium for Atmospheric Research on Transport and Transformation (ICARTT) standard. The majority of other NASA atmospheric datasets use the ICARTT standard, so following suit with the low-cost air quality data would make them more compatible.

Okorn has been assisted in developing the database by a team of undergraduate and graduate school interns. They have studied scientific papers and talked to researchers who combined data in similar ways for other projects. They have worked to find recommendations for metadata content, the flagging of data for quality control, calibration, and other guidance. The aim has been to make it as easy as possible for researchers from other networks to contribute to their database.

“Some of these networks may not have a strong database or API,” said Okorn. “I ask them to just send us their data and fill out the Google metadata information form. We put the DOI in their name and handle the rest, which has really helped the participation in the dataset grow.”

One example is the Ribbit Network, which is a citizen-science organization building the world’s largest network of CO2 monitoring sensors. 

“Working with Kristen and her team has been a genuinely seamless experience,” said Keenan Johnson, Ribbitt Network’s founder and executive director. “They handled the heavy lifting of reformatting our data into the common framework, embedding the necessary metadata, and more. It made participation easy for our smaller-community network to contribute meaningfully to large-scale scientific efforts.”

Okorn has made her offer to researchers around the world, and so far more than 20 networks operating in 80 countries have contributed to the database.

Data Details

The Low-Cost Air Quality Sensor Harmonization Database can be found either by using NASA’s Earthdata Search or by visiting a clickable ArcGIS map. The datasets extend back to 2012, and Okorn hopes including them in Earthdata will help researchers find them while browsing the archive for satellite and related sensor measurements. 

“This is a way they may discover useful ground-based data they didn’t know were there,” said Okorn. “We also hope that the format is straightforward enough that other community users will find it useful as well.”

Okorn's team rates each dataset with a quality level, evaluating how well the data is calibrated. Users can go to the network map and select data they are interested in based on its quality level, as well as its location, date range, and pollutant of interest. The metadata also include detailed explanations for how measurements were calibrated.

Putting the Data to Use

Okorn suggests the data could be useful for validating measurements made by atmospheric chemistry sensors mounted on satellites. 

For instance, the hourly measurements that the Tropospheric Emissions: Monitoring of Pollution (TEMPO) instrument makes from orbit could be compared with air quality samples from the ground to check the sensor’s accuracy and provide additional, local information. To test that idea, Okorn's team is developing a project to see how low-cost sensors from different networks in the same area can be used for modeling and validating TEMPO observations of California.

Another benefit of the database is simply helping researchers discover the smaller or lesser-known networks in the world and promoting collaborations for more pinpointed studies. Having datapoints from these low-cost networks also helps fill in the picture of what air quality and pollution are like beyond the field of view or sensitivity of satellites or major ground-based sensors. 

“Here in the Bay Area, regulatory monitors are denser than in a lot of other U.S. regions, but there are still 20-mile long stretches without EPA monitors," said Okorn. "Datapoints from smaller networks in between could help us understand the gradients between measurements from those larger sensors.” 

As the database grow, the team keeps working to increase its value to users. Soon they plan to offer visualization tools, Jupyter notebooks for data analysis, data processing automation, and more international networks from places including Ghana and Pakistan. 

For more information on the database and its ever-growing list of networks, visit the Low-Cost Sensor AQ Data project page. Likewise, see the Early Career Research Program webpage for more information on its opportunities.