Long-term Evaluation of Low-Cost Air Sensors in Monitoring Indoor Air Quality at a California Community
- Author(s): Wang, Zemin
- Advisor(s): Zhu, Yifang
- et al.
Introduction: In response to substantial evidence showing adverse health effects of long-term or short-term exposure to air pollutants, interest has grown in real-time monitoring of air quality in fine-grained geographic detail. The development and application of low-cost air sensors enable high spatial resolution air quality monitoring. The validity and consistency of those devices, however, still needs to be investigated. Previous literature suggests potential inaccuracy due to environmental factors. Our study aims to provide evidence that low-cost air sensors could be applied at a community scale to monitor indoor air quality over time and to alert communities and residents about air pollution issues. Objectives: 1. Identify indoor particulate matter(PM2.5) sources using low-cost sensors at a community scale; 2. Identify and explore potential indoor PM mitigation measures; 3. Explore the impacts of ambient PM levels on indoor air quality; 4. Evaluate the long-term performance of low-cost air sensors Methods: We have operated 30 PurpleAir II (PA-II) sensors (12 outdoor and 18 indoor) at a California community located adjacent to a major interstate freeway vicinity since May 2017. PM2.5 data were recorded and uploaded automatically by the sensor network. We also collected one-week indoor human activity logs (i.e., whether cooking, opening windows, or using air purifiers at each hour) from 9 recruited residents using questionnaires during the PM monitoring campaign. The indoor PM data were matched with activity logs based on the monitoring location and time. We then assessed the impacts of ambient air quality, microclimatic factors (e.g., temperature and relative humidity (RH)), and indoor human activities on indoor PM2.5 concentrations using t-tests and a linear mixed-effects regression model. Results: Indoor sensors had greater data completeness than outdoor sensors. The average of PM2.5 Indoor/Outdoor (I/O) ratio during cooking hours was 7.3, significantly greater (p < 0.01) than the average of 1.4 during non-cooking hours. The fitted linear mixed-effects model can explain 86.4% of the variation in indoor PM levels. The model shows that indoor PM2.5 was positively influenced by ambient PM2.5 and indoor cooking and negatively influenced by window opening and using an air purifier. Moreover, ambient PM and window opening had an interaction effect on indoor PM levels. Conclusions: PA-II sensors could effectively monitor indoor PM concentrations over a long-time span and detect the impacts of indoor activities on IAQ. PA-II sensors deployed indoors performed better than those deployed outdoors. Ambient PM levels had a significant positive effect on IAQ. Residential cooking was a strong indoor PM emission source, which could be influenced by effective ventilation and mitigation measures.