A Low-Cost Passive Monitor for Black Carbon Air Pollution

Abstract

TPS 651: Air pollution exposure modeling 1, Exhibition Hall, Ground floor, August 27, 2019, 3:00 PM - 4:30 PM Background: Developing-country households using solid fuels often have high concentrations of black carbon air pollution. Monitoring, especially in rural areas, is expensive, uncertain, and logistically challenging, and often reflects short-term (~hours or ~day) rather than long-term (~months or longer) conditions. We developed a low-cost (<$10) passive sampler for long-term average black carbon concentration. Method: We tested samplers for multiple months in 20 indoor locations in rural India to determine reproducibility and limits of detection. All locations received duplicate samplers (i.e., two per location). Our approach measures the change over time in sampler filter surface reflectance (pixel intensity [PI]) relative to baseline and to an unexposed reference. Results: Our study is ongoing: to date (March 2019), samplers have been exposed for ~2 months. All samplers (n=40) experienced a measurable change in reflectance (>1.0% relative to reference); none of the samplers have fully saturated yet. Duplicate samplers’ measurements were highly correlated (r>0.99), suggesting excellent reproducibility. Variability in reflectance among locations suggests strong ability to differentiate between higher versus lower long-term average black carbon concentration: specifically, differences between duplicate samplers averaged 1.1% (range: 0.02% to 2.9%) [720 PI (range: 12 to 1900)], whereas the difference among locations was 73% (range: 1.1% to 74%; mean: 18%) [48,000 PI (range: 730 to 49,000; mean: 12,000)]. Conclusions: Results thus far for 20 locations suggest robust reproducibility and precision. For indoor locations investigated, in a community that uses solid fuels, the minimum time-scale for sampling (which reflects the lower limit of detection) appears to be <2 months; the maximum time-scale for sampling (which reflects the upper limit of measurement) appears to be >2 months. Ongoing work will continue investigating reproducibility and limits of detection.