Extreme smoke event over the high Arctic

Abstract The intense western Canadian fires of August 2017 resulted in extreme, high-Arctic fine mode (FM) smoke AODs (aerosol optical depths) over a 2008 to 2017 (10-year) sampling period. The primary measurements employed to monitor smoke events were FM AODs derived from the measured AOD spectra of two AEROCAN/AERONET (CIMEL) sunphotometers at Eureka, Nunavut, Canada. The FM AOD attribution is argued to be a necessary condition for the presence of smoke. Various supporting information, including the correlation with smoke proxy (CO) retrievals, the high frequency (rapid diurnal variation) and the high amplitude nature of the FM AODs, ground-based backscatter lidar profiles, the redundancy of the double CIMEL retrievals, satellite remote sensing, aerosol modeling and backtrajectories indicated that the peak event was likely due to smoke from extreme pyroCb fires in British Columbia. The hypothesis that the FM AOD peak event was an extreme event was tested for a derived ensemble of fine mode events and their peaks over the 10-year sampling period. The results confirmed the hypothesis at the 0.001 level of significance. Important indicators that the 10-year ensemble of FM AOD events did indeed represent smoke were their high frequency and high amplitude FM nature, their occurrence during the Boreal forest fire and agricultural fire seasons in Canada and Asia, and their strong correlation with CO abundances retrieved from FTIR measurements (when sufficient FM AOD and CO statistics were available). In the process of accumulating climatological-scale, monthly-binned fine mode AOD statistics, we found moderate correlations with forest fire or agricultural fire emissions from the Boreal North American, Boreal Asia or Central Asia regions as well as with CO retrievals at Eureka. We argued that confounding factors constraining the monthly binned fine mode AOD vs emissions correlations were associated with the monthly-binned meteorological dynamics (with notable, event-level, exceptions) while confounding factors constraining fine mode AOD vs CO correlations included the different physio-optical nature of those smoke proxies (solar attenuation by fine mode particle scattering versus solar attenuation by molecular absorption). We also employed historic (2005-2010) AHSRL (Arctic High Spectral Resolution Lidar) profiles to estimate an optically averaged smoke plume height of ∼3 to 3 ½ km during the spring and summer seasons.