Arctic black carbon during PAMARCMiP 2018 and previous aircraft experiments in spring

1Institute for Space–Earth Environmental Research, Nagoya University, Nagoya, Aichi, Japan 2Institute for Advanced Research, Nagoya University, Nagoya, Aichi, Japan 3Department of Earth and Planetary Science, Graduate School of Science, University of Tokyo, Tokyo, Japan. 4National Institute of Polar Research, Tokyo, Japan 10 5Meteorological Research Institute, Tsukuba, Japan 6Graduate School of Environmental Studies, Nagoya University, Nagoya, Japan. 7Johannes Gutenberg University of Mainz, Institute for Atmospheric Physics, Mainz, Germany 8Max Planck Institute for Chemistry, Particle Chemistry Department, Mainz, Germany 9Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research (AWI), Bremerhaven, Germany 15 10LISA, UMR CNRS 7583, Université Paris-Est-Créteil, IPSL, Créteil, France

reported in this paper were obtained by integrating the mass of BC particles within the detectable diameter range and are expressed in units of ng m -3 under standard temperature and pressure (STP, 0ºC and 1013.25 hPa) conditions. The accuracy of the MBC measurements estimated from the uncertainty of the calibration and operational conditions of SP2 was about 10%.
As described in Appendix A, MBC measurements made with the University of Tokyo SP2 during 130 ARCTAS (Sahu et al., 2012) were compared with NOAA SP2 measurements (NOAA-ARCPAC experiment, Spackman et al., 2010) and they agreed within 20%. The systematically higher U. Tokyo values might partly be attributed to a wider size range of U. Tokyo BC measurements and a difference in corrections for undetected BC between the measurements. Because MBC measurements were consistently and traceably made by the individual groups, the systematic differences in absolute MBC 135 values are expected to be smaller than 20% at least for PAMARCMiP and ARCTAS datasets (U. Tokyo) and HIPPO dataset (NOAA).
Aerosols were also sampled on filters onboard the aircraft during PAMARCMiP and analyzed using transmission electron microscopy (TEM, Adachi et al., 2021). In this study, we used potassiumcontaining particles as an indicator of influences from biomass burning. 140 Gaseous carbon monoxide (CO) concentration was measured using an Aerolaser AL5002 CO monitor (Gerbig et al., 1999;Scharff et al., 2012), which detects fluorescence in vacuum ultraviolet radiation.
For the aerosol measurements, ambient air was drawn into the aircraft cabin using a forward facing iso-145 kinetic inlet with a tip hole diameter of 4.6 mm designed and made by DMT, Inc. The sample air speed at the inlet tip and true air speed of the aircraft at various altitudes agreed within 10%, indicating that iso-kinetic sampling was achieved during the experiment. For the gas measurements, two backward facing sampling inlets (Ehrlich et al., 2019) were used.
PAMARCMiP aircraft experiment was carried out with the Alfred Wegener Institute (AWI) research  Table 1. The data obtained during 5 min after taking off and 5 min before landing were not used to avoid possible influences of local pollutants emitted from the base Station Nord. Vertical profiles of BC and other atmospheric species were measured up to 5 km during most of the research flights. Appendix B presents 6-day kinematic back trajectories of air parcels measured onboard the aircraft.

Ground-based MBC measurements at Ny-Ålesund and Barrow
Along with year-to-year variations observed during aircraft experiments, those observed from groundbased MBC measurements at the high Arctic sites Ny-Ålesund in Svalbard (78. 9°N, 11.9°E) and Barrow in Alaska (71.3°N, 156.6°W) are also presented for comparison in this paper. At both sites, MBC data https://doi.org/10.5194/acp-2021-349 Preprint. Discussion started: 28 April 2021 c Author(s) 2021. CC BY 4.0 License. prescribed sea surface temperature and sea ice data (Ishii et al., 2005). In the simulations, horizontal wind fields were nudged toward 6-hourly Japanese 55-year Reanalysis data (Kobayashi et al., 2015). 195 Simulations used monthly AN emission from the MACCity emission dataset (Lamarque et al., 2010) and daily BB emission from the Global Fire Assimilation System dataset (GFAS, Kaiser et al., 2012).
We also performed a model simulation that omitted the BB emission of BC to estimate the separate contributions of AN and BB sources to total BC concentrations.  Table 2. Median MBC values were between 7 and 18 ng m -3 at altitudes below 4.5 km. At altitudes around 5 km, enhancements of MBC up to 250 ng m -3 (one-minute data) were observed. As 205 described in Sect. 5, these high MBC values were likely due to influences from BB BC emissions.

Year-to-year variation in BC and influences from BB
In Fig. 2b Using the four aircraft datasets, BC column amounts at altitudes between 0 and 5 km (COLBC, in units of µg m -2 ) were calculated by vertically integrating median MBC values in each altitude obtained at latitudes north of 66.5°N (Table 4). Although data were obtained at altitudes higher than 5 km during 220 the ARCTAS and HIPPO experiments, they were not used. Fig. 3 compares COLBC values among the four experiments. In accordance with the vertical MBC profiles (Fig. 2b) A part of the differences in the observed BC levels could be due to the different locations where the measurements were made. As shown in Fig. 1a, the PAMARCMiP and NETCARE aircraft experiments were conducted mainly at latitudes around 80°N, while ARCTAS and HIPPO data obtained at latitudes between 66.5 and 90°N were used in this study. In Fig. 3

Biomass burning fire counts
In Fig. 3, the daily averaged numbers of fire counts (counts day -1 , indicator of BB activities) detected by the MODIS satellite (MCD14DL products, https://earthdata.nasa.gov/active-fire-data) are shown for the four aircraft experiments. In this figure, we show fire counts at latitudes north of 50°N for the time period between 14 days before the first day of the aircraft experiment and the last day of the experiment.

240
The averaged fire counts for HIPPO and PAMARCMiP are similar, whereas those for NETCARE and ARCTAS are, respectively, factors of around 5 and 14 higher than that for PAMARCMiP ( Fig. 3 and Table 4). Similar tendencies can be also seen in averaged GFAS and GFED BB BC emissions (Fig. 3) that were created using the fire counts and were used for numerical simulations in this study (section 3).
Consequently, the relative changes in COLBC were generally consistent with those in BB activities.

245
These results suggest that BC emissions from BB likely contributed to the increased COLBC during the NETCARE and ARCTAS experiments, at least to some extent. In other words, year-to-year variations of tropospheric BC amounts in the Arctic in spring were primarily affected by transport of BB BC, at least during these four periods. As described below in Sect. 4.3, results from numerical model simulations indicate that the year-to-year variation of AN COLBC (>66.5°N) was smaller than 30-40%, 250 thus supporting that year-to-year variation in COLBC was mostly due to BB BC. 2005; Park et al., 2005;Evangeliou et al., 2016). Even when this criterion was changed to 0, 7, and 21 260 days before the first day of the experiment, the relative changes in averaged fire counts shown in Fig. 3 were not strongly affected. A time series of fire counts in individual latitude ranges is presented in Fig.   C1 in Appendix C. As seen in this figure, fire counts at latitudes > 40°N and > 50°N started to increase in mid-to late March during each of the four years examined in this study. Because time periods of 7 or 14 days generally captured this onset of fire activities, the year-to-year variations of fire counts were 265 generally similar irrespective of the chosen time period. Notably, time series of the fire counts at latitudes > 40°N and > 50°N were generally similar (Fig. C1) and year-to-year variations were also similar when either of these criteria was adopted. Furthermore, the fire counts at latitudes > 60°N were negligible in all the cases and the correspondence to their year-to-year variation was less significant.
Consequently, transport of air influenced by BB at latitudes between 45°N and 60°N is likely 270 responsible to the increased BC level in the Arctic spring.
As described in Sect. 2, during the PAMARCMiP experiment, air parcels sampled at altitudes below 1, 3, and 5 km had generally originated from north of 70°N, 60°N, and 50°N, respectively (Fig. B1). Most of the air parcels sampled at altitudes above 3 km had not passed over high BB areas and had maintained their altitudes, thus, they were likely not significantly influenced by BB BC emissions.

275
Consequently, air parcels sampled during the PAMARCMiP experiment were likely not influenced by recent BC emissions, except for the plumes observed at altitudes around 5 km (Sect. 5).

Evaluation of BB BC using numerical model simulations
In Fig Table 3, and (C) area-weighted averages within the entire region at latitudes north of 66.5°N for the time period of an aircraft experiment (Table 5)  suggest that BB BC emissions could be largely underestimated or the removal of BC from BB 315 emissions was overestimated, although the level of these uncertainties cannot be evaluated in this study. for HIPPO and NETCARE, respectively), medians were also calculated for the 31-day period for which 325 the median date was chosen to be the median date of the experiment ( Fig. 6 and Table 6). The corresponding average fire counts (31 + 14 days) are also shown in Fig. 6 and Table 6. Previous long time period studies showed that MBC values in Barrow were about a factor of 1.4 higher than in Ny-Ålesund in March-April (Sinha et al., 2017). This tendency is also recognized in the ARCTAS and PAMARCMiP periods shown in Fig. 6, although it is not clearly seen in the other two propose that these enhanced MBC values were likely due to influences from BB emissions. We then examine possible differences in microphysical properties between these BC particles and other BC particles that were least affected by BB.

350
The plumes with enhanced MBC values were observed at similar locations and altitudes on 2, 3, and 4 April. In total, 29 one-minute data with MBC > 50 ng m -3 were obtained. Although the horizontal wind direction was different on these three days, wind speeds of 2.5 to 6.5 m s -1 suggest that the plumes could extend over 1000 km. Because the research aircraft could not fly at altitudes above 5.2 km, we could not observe the vertical extent of these enhancements. However, the elastic air-borne mobile 355 aerosol Lidar (AMALi) installed on board of the research aircraft observed a well-defined aerosol layer at altitudes between 5.15 and 6.8 km on 2 April (Nakoudi et al., 2020), suggesting that the plumes with enhanced MBC values could extend over this vertical range. A photo taken from the research aircraft POLAR 5 on 3 April is also shown in Appendix D (Fig. D1), where a haze layer with reddish color, which may correspond to the plumes, was captured. When MBC enhancements were observed, increased aerosol potassium concentration was also observed, suggesting that these plumes had likely been affected by BB emissions (Adachi et al., 2021).
Enhancements in CO and CO2 concentrations were also observed in these plumes. The ratio of their increases (DCO/DCO2), derived from a slope of a least squares fit to the scatter plots of CO and CO2 concentration data during the plume periods, was 12.3 ppbv ppmv -1 , suggesting that these plumes were 365 likely due to incomplete combustion. This result is also in accordance with influences from BB. In fact, this ratio is similar to the DCO/DCO2 ratio of 15±5 ppbv ppmv -1 observed in air affected by Asian BB observed during the ARCTAS experiment (Kondo et al., 2011a).  Fig. 8a shows a vertical profile of the mass averaged diameter of BC particles (DBC), which was defined

Microphysical features of BB BC
where rBC is the void-free density of BC ( The median value of the DBC in the enhanced MBC air was 145 nm (Table 2), which is only slightly 415 higher than that of other BC particles sampled at this altitude range. Various factors could cause the observed differences. The BC particles could have already been larger upon BB emission than those from anthropogenic sources or they could have been removed less efficiently. However, considering the relatively large change in DBC with altitude (Figs. 8a and 9), the similarity in DBC between recent BB plumes and aged AN air is remarkable. The shell-to-core diameter ratios of the enhanced MBC data were only slightly higher than for other data (Table 2). These air parcels were likely affected by recent (within a week) wet removal and, therefore, one may expect a smaller coating thickness because BC particles with thicker coating were likely removed due to higher CCN activities. Weaker updraft speed (lower supersaturation) or less hydrophilic 430 coating compositions could be possible explanations. Cloud-precipitation processes during transport such as the aqueous-phase reactions or evaporation of precipitating particles after accretion might also increase the coating thickness of BC. The similarity in the shell-to-core diameter ratios between recent BB plumes and aged AN air provide useful constraints to validate numerical model simulations. in other air sampled, which were considered to be mostly aged AN BC.
Direct radiative forcing of BC in the Arctic is considered to be highest in spring when MBC is the largest and incoming solar radiation is increasing. BC in spring is also important because slight changes in the time of snow/ice melt can influence the radiation budget in the Arctic. The observations presented in this study provide useful bases to improve and evaluate numerical model simulations that assess the BC 480 radiative effect in the Arctic spring. April 2008. This figure also shows that time series of one-minute MBC data agreed well between the two measurements. Fig. A1b shows a scatter plot between these two one-minute data. They were highly correlated (R 2 = 0.92) with a slope of 0.80. The systematically higher U. Tokyo values might partly be attributed to a wider size range of U. Tokyo BC measurements and a difference in corrections for 490 undetected BC between the measurements as described below.

Appendix A. Comparison between the University of Tokyo and NOAA SP2 measurements
During the ARCTAS campaign, the U. Tokyo SP2 measured BC particles for the Dm range of 80-860 nm. Although the estimated mass fraction of the undetected BC was small (4% for 50-80 nm), the MBC for the Dm range of 50-900 nm was obtained by integrating the single-mode lognormal fit function of  (Spackman et al., 2010, Schwarz et al., 2010. These differences in measurements and data reductions could result in some systematic differences between the two MBC datasets.

500
Because MBC measurements were consistently and traceably made by the individual groups, the systematic differences in absolute MBC values obtained by them are expected to be smaller than 20%.

Appendix B. Back trajectories of sampled air
Six-day kinematic back trajectories of air parcels measured onboard the aircraft were calculated every minute based on the method described by Tomikawa and Sato (2005). For calculating the trajectories, 6-505 hourly meteorological data from the National Centers for Environmental Prediction (NCEP) Final (FNL) operational global analysis were used. As shown in Fig. B1, air parcels sampled at altitudes below 1 km and between 1 and 3 km had originated mostly from north of 70°N and 60°N, respectively.
Air parcels sampled at altitudes between 3 and 5 km had occasionally originated at latitudes as low as 50°N, in central and eastern Eurasia and north-western Canada.

Appendix D. Photo taken from POLAR 5
A photo taken from the research aircraft POLAR 5 on 3 April is shown in Fig. D1. A haze layer with reddish color, which may correspond to the plumes, was captured. Although not so apparent as this photo, a haze layer with dark color was also sighted on April 2 and 4, when the plumes with enhanced 520 MBC values were observed.

Data availability
The observational data set used in this publication will be available online (https://ads.nipr.ac.jp).

Competing interests
The authors declare that they have no conflicts of interest.  (Table 3).