Abstract:
Background The two main seasons with highest concentrations of fine particulate matter (PM2.5) in the state of Washington (WA) are wintertime and periods of wildfire smoke. We examined wintertime PM2.5 through a research partnership with the Yakama Nation Air Quality Section, and wildfire smoke PM2.5 through research with schools in two different regions of WA, as well as school and childcare facility air quality decision-makers. Our research sought to address the following aims: 1) Characterize wintertime PM2.5 concentrations and chemical composition in the Yakama Nation reservation based on paired indoor and outdoor air sampling, 2) Examine application of low-cost sensors to understand indoor PM2.5 concentrations in four WA schools in areas impacted by wildfire smoke, and 3) Understand perspectives on feasibility, data-interpretability, and decision-making in using low-cost sensors for wildfire smoke response in schools and childcare facilities.
Methods For Aim 1, we collected PM2.5 onto filters during three 1-week periods indoors and outdoors at five locations. We quantified PM2.5 concentrations gravimetrically and analyzed the samples for levoglucosan – a biomass burning tracer compound – as well as select elements, ions, and polycyclic aromatic hydrocarbons (PAHs). We also measured continuous PM2.5 concentration using low-cost sensors. We quantified spatial variation in PM2.5 and analytes. PM2.5 temporal variation was observed by plotting diurnal patterns. For Aim 2, we measured PM2.5 concentrations indoors and outdoors at four schools in WA during wildfire smoke in 2020. This involved monitoring continuously using a low-cost sensor and gravimetrically. We randomly sampled 5-minute segments of low-cost sensor data to create hypothetical simulations of brief portable handheld measurements. For Aim 3, we conducted 15 semi-structured interviews with school, childcare, local health jurisdiction, air quality, and school district personnel regarding sensor use for wildfire smoke response. Interviews included sharing PM2.5 data collected at schools during wildfire smoke. Interviews were transcribed and transcripts were coded.
Results In the wintertime data, the PM2.5 concentration at one site was generally about half of the concentration measured at the closest regulatory monitor. While PM2.5 concentrations at other sites were similar, greater between-site variation was evident for analytes. Percent differences in PAHs relative to the median for at least one site within the same week were more than twice as high in 22 cases, and in two cases were more than 10 times as high. Gravimetric indoor/outdoor ratios varied overall from 0.3 to 2.6. Sites had varied diurnal patterns of peak PM2.5 concentrations which did not strongly resemble diurnal patterns typically associated with residential wood burning. During wildfire smoke events (lasting 4-19 days), median hourly PM2.5 concentrations at different locations inside a single facility varied by up to 50 µg/m3 during school hours over the same time period. Median hourly indoor/outdoor ratios during school hours ranged from 0.22 to 0.91. Within-school differences indicated that it is important to collect measurements throughout a facility. The simulation results suggested that making handheld measurements more often and over multiple days better approximates indoor/outdoor ratios for wildfire smoke. Three major themes were identified in the interview responses: 1) Low-cost sensors are useful despite data quality limitations, 2) Low-cost sensor data can inform decision-making to protect children in school and childcare settings, and 3) There are feasibility and public perception related barriers to using low-cost sensors.
Conclusion Spatial and temporal variation of PM2.5 in a rural area suggests that sparse regulatory monitors may misrepresent the range of PM2.5 exposures that people experience, as well as the multiplicity of PM2.5 sources. We found useful, practical information applicable for optimized sampling with low-cost sensors for wildfire smoke response in schools. Interview responses provided practical implications, including demonstrating a need for guidance that allows a variety of sensor preferences and addresses sensor uses outside of activity decisions, especially assessment of ventilation and filtration.