Cost-Effective Sensor Networks for Modeled, Highly Accurate Air Quality Monitoring

Advances in Sensor Technologies in Environmental Monitoring
Oral Presentation

Prepared by D. Later1, P. Gaillardon2, K. Kelly3, C. Luft1, T. Becnel1, N. Page1, S. Charas1
1 - Tellus Networked Sensor Solutions, 2319 S. Foothill Drive, Suite 140, Salt Lake City, UT, 84109, United States
2 - University of Utah, 201 Presidents' Circle, Department of Electrical and Computer Engineering, Salt Lake City, UT, 84112, United States
3 - University of Utah, 201 Presidents' Circle, Department of Chemical Engineering, Salt Lake City, UT, 84112, United States

Contact Information: [email protected]; 18015573557


Air quality (AQ) is critical to human health and the environment. Poor AQ is responsible for 6.5 million premature deaths annually, and particulate matter (PM) pollution is a key driver of adverse health effects including asthma, lung cancer, heart attacks, and premature death. Cost-effective sensor networks can generate highly accurate, near real-time estimates of PM2.5 levels for Air Quality Monitoring (AQM) purposes. An example of such a sensor network is the Air Quality & You (AQ&U) deployed in Salt Lake City, UT. It leverages measurements from three established PM2.5 sensor networks: (1) the Utah Division of Air Quality (UDAQ) higher cost FRM/FEM monitoring stations and the cost-effective sensors in the (2) University of Utah’s/TELLUS’ AirU® network and (3) PurpleAir’s (PA) network. Collectively these 3 sensor networks consist of over 300 sensors. The infrastructure for the AQ&U sensor network was originally developed under a National Science Foundation (NSF) grant, and the back-end services, automated quality assurance, and visualizations via the AirView™ map were developed in a partnership with Drs. Kelly, Gaillardon, and Whitaker, along with their collaborator Dr. Meyer, as well as scientist and engineers at TELLUS Networked Sensor Solutions, Inc. (formerly TETRAD Sensor Network Solutions, LLC). This integrated solution includes cost-effective, air quality sensing nodes, regulatory measurements, robust quality assurance strategies, data fusion/statistical modeling, and engaging visualizations to provide communities with neighborhood-scale, real-time estimates of particle pollution levels.

Sensing nodes. The PM measurements include three types of sensors including the AirU and PA-II cost-effective sensors, as well as UDAQ FEM/FRMs. Each AirU node contains sensors for temperature and relative humidity, oxidizing and reducing gas species, GPS location and altitude, and PM2.5 concentration. The AirU platform has onboard WiFi capabilities to upload data to the cloud, as well as capabilities of storing over 10 years of air quality measurements to an onboard micro-SD card. Each AirU sensor pushes its measurements to a Google Cloud database at 2-minute intervals. The PA sensors also measure PM concentration. The AirU and PA PM2.5 sensor technology is based on light-scattering, which is a common method for measuring PM concentration in both regulatory and reference measurements.

Quality assurance strategies. All AirU sensors are calibrated to ensure measurement precision prior to deployment. Light-scattering based PM sensors, like those used in the AirU or PA, require a correction factor to convert light scattering to particle mass concentrations, which are the basis for air-quality metrics. The AQ&U network develops and applies seasonal and event-specific correction factors. These factors are developed by co-location with FRMs. There are three PA-II sensors and three AirU sensors collocated at a UDAQ monitoring site and we have demonstrated that, after applying seasonal correction factors, the AirU and PA sensors meet EPA’s performance targets.

Data fusion, statistical modeling, and visualizations. The AQ&U sensor network provides rigorous automated data screening to remove outliers and sensors that drift. The screened and corrected PM2.5 sensor measurements are incorporated into a statistical model, which includes covariates for time, space, and elevation. The AirView software displays a color-coded heat map for all populated areas of Salt Lake County. In addition, each sensor node is displayed as color-coded dot based on detected particulate matter (PM2.5) concentration at that location. The visualizations of PM concentration use the EPA AQI color scheme, because of its familiarity with the public. The AQ&U network has documented the performance of the cost-effective PM2.5 sensors, the infrastructure, and the accuracy of the PM2.5 estimate maps in several peer-reviewed publications.

The University of Utah and TELLUS teams have spent years developing the cost-effective sensor strategies to ensure that the PM measurements, estimates, and visualizations are accurate by screening the measurements for outliers, by developing local and seasonal correction factors for the AirU and PA sensor networks, and by comparing AQ&U results to reference measurements from the Utah DAQ FRM/FEM site. These strategies help people make sense of the often-imperfect data coming from crowd-sourced, cost-effective sensor measurements. Through AQ&U these technologies are now available to Salt Lake City and County communities and the data can help individuals and policymakers make informed decisions about air quality and their personal exposure. In addition, other similar sensor networks are in use in Kansas City, MO, Chattanooga, TN, Springfield, MA, and Cleveland, OH. The demand for air quality monitoring is increasing, with more communities across the United States eager to understand their exposure to air pollution as climate change events occur. TELLUS stands ready to respond to this growing interest by expanding cost-effective air quality sensor networks to serve more communities, providing valuable information and insights on local air quality.