You are here:
CAIRSENSE Study: Real-world evaluation of low cost sensors in Denver, Colorado
Citation:
Feinberg, S., R. Williams, J. Rickard, R. Brown, D. Garver, G. Harshfield, P. Stauffer, E. Mattson, R. Judge, AND S. Garvey. CAIRSENSE Study: Real-world evaluation of low cost sensors in Denver, Colorado. Presented at National Ambient Air Monitoring Conference, St. Louis, Missouri, August 08 - 11, 2016.
Impact/Purpose:
This will be a presentation on field tests of low cost sensors at the National Ambient Air Monitoring Conference in St. Louis. The audience will be primarily technical staff members from EPA, states, and local agencies.
Description:
Low-cost air pollution sensors are a rapidly developing field in air monitoring. In recent years, numerous sensors have been developed that can provide real-time concentration data for different air pollutants at costs accessible to individuals and non-regulatory groups. Additionally, these sensors have the potential to improve the spatial resolution of monitoring networks and provide a better understanding of neighborhood- and local-scale air quality and sources. However, many new sensors have not been evaluated to determine their long-term performance and capabilities. In this study, nine different low-cost sensor models, including O3, NO2 and particle sensors, were deployed in Denver, CO from September 2015 to February 2016. Three sensors of each type were deployed to evaluate instrument precision and consistency over the time period. Sensors were co-located with reference monitors at the Denver NCore site in order to evaluate sensor accuracy and precision. Denver was chosen as the location for this study to evaluate sensor performance in dry, high altitude, and low winter temperatures. Sensors were evaluated for data completeness, performance over time, and comparison with regulatory monitors. This presentation will also address challenges and approaches to data logging and processing. Preliminary analysis revealed that most sensors had high data completeness when data loggers were operational (e.g., the Aeroqual O3 sensor ranged from 94-100%), and exhibited a wide range of correlations with reference monitors (e.g., PM sensors ranged from r = 0.01 to r = 0.86). In addition, most sensors of the same model showed strong correlation with themselves (e.g., AirViz Speck unit inter-correlation ranged from 0.77-0.92).