In situ measurements of trace gases, PM, and aerosol optical properties during the 2017 NW US wildfire smoke event

Vanessa Selimovic, Robert J. Yokelson, Gavin R. McMeeking, Sarah Coefield

Research output: Contribution to journalArticlepeer-review

41 Scopus citations


In mid-August through mid-September of 2017 a major wildfire smoke and haze episode strongly impacted most of the NW US and SW Canada. During this period our ground-based site in Missoula, Montana, experienced heavy smoke impacts for ∼ 500h (up to 471μ-3 hourly average PM2.5). We measured wildfire trace gases, PM2.5 (particulate matter ≤2.5μm in diameter), and black carbon and submicron aerosol scattering and absorption at 870 and 401nm. This may be the most extensive real-time data for these wildfire smoke properties to date. Our range of trace gas ratios for δNH3 δCO and δC2H4 δCO confirmed that the smoke from mixed, multiple sources varied in age from ∼ 2-3h to ∼ 1-2 days. Our study-average δCH4 δCO ratio (0.166±0.088) indicated a large contribution to the regional burden from inefficient smoldering combustion. Our δBC δCO ratio (0.0012±0.0005) for our ground site was moderately lower than observed in aircraft studies (∼ 0.0015) to date, also consistent with a relatively larger contribution from smoldering combustion. Our δBC δPM2.5 ratio (0.0095±0.0003) was consistent with the overwhelmingly non-BC (black carbon), mostly organic nature of the smoke observed in airborne studies of wildfire smoke to date. Smoldering combustion is usually associated with enhanced PM emissions, but our δPM2.5 δCO ratio (0.126±0.002) was about half the δPM1.0 δCO measured in fresh wildfire smoke from aircraft (∼ 0.266). Assuming PM2.5 is dominated by PM1, this suggests that aerosol evaporation, at least near the surface, can often reduce PM loading and its atmospheric/air-quality impacts on the timescale of several days. Much of the smoke was emitted late in the day, suggesting that nighttime processing would be important in the early evolution of smoke. The diurnal trends show brown carbon (BrC), PM2.5, and CO peaking in the early morning and BC peaking in the early evening. Over the course of 1 month, the average single scattering albedo for individual smoke peaks at 870nm increased from ∼ 0.9 to ∼ 0.96. B/scat401 B/scat870 was used as a proxy for the size and "photochemical age" of the smoke particles, with this interpretation being supported by the simultaneously observed ratios of reactive trace gases to CO. The size and age proxy implied that the Ångström absorption exponent decreased significantly after about 10h of daytime smoke aging, consistent with the only airborne measurement of the BrC lifetime in an isolated plume. However, our results clearly show that non-BC absorption can be important in "typical" regional haze and moderately aged smoke, with BrC ostensibly accounting for about half the absorption at 401nm on average for our entire data set.

Original languageEnglish
Pages (from-to)3905-3926
Number of pages22
JournalAtmospheric Chemistry and Physics
Issue number6
StatePublished - Mar 26 2019


Dive into the research topics of 'In situ measurements of trace gases, PM, and aerosol optical properties during the 2017 NW US wildfire smoke event'. Together they form a unique fingerprint.

Cite this