Impact of Siberian forest fires on the atmosphere over the Korean Peninsula during summer 2014

Extensive forest fires occurred during late July 2014 across the forested region of Siberia, Russia. Smoke plumes emitted from Siberian forest fires underwent long-range transport over Mongolia and northeast China to the Korean Peninsula, which is located ∼ 3000 km south of the Siberian forest. A notably high aerosol optical depth of ∼ 4 was observed at a wavelength of 500 nm near the source of the Siberian forest fires. Smoke plumes reached 3–5 km in height near the source and fell below 2 km over the Korean Peninsula. Elevated concentrations of levoglucosan were observed (119.7± 6.0 ng m−3), which were ∼ 4.5 times higher than those observed during non-event periods in July 2014. During the middle of July 2014, a haze episode occurred that was primarily caused by the long-range transport of emission plumes originating from urban and industrial complexes in East China. Sharp increases in SO2− 4 concentrations (23.1± 2.1 μg m−3) were observed during this episode. The haze caused by the long-range transport of Siberian forest fire emissions was clearly identified by relatively high organic carbon (OC) / elemental carbon (EC) ratios (7.18± 0.2) and OC /SO2− 4 ratios (1.31± 0.07) compared with those of the Chinese haze episode (OC /EC ratio: 2.4± 0.4; OC /SO2− 4 ratio: 0.21± 0.05). Remote measurement techniques and chemical analyses of the haze plumes clearly show that the haze episode that occurred during late July 2014 was caused mainly by the long-range transport of smoke plumes emitted from Siberian forest fires.

During a severe forest fire episode in Moscow, Russia in August 2010, notably high concentrations of total carbon (mean of 202 µg m −3 ) and levoglucosan (3.1 µg m −3 ) were observed with an elevated organic carbon/elemental carbon (OC / EC) ratio of 27.4 (Popovicheva et al., 2014).Total carbon concentrations exceeded 10 times that during non-event periods in Moscow (Popovicheva et al., 2014).During severe forest fires in Siberia in May 2003, the surface PM 10 (particulate matter with a diameter of ≤ 10 µm) and O 3 concentrations in downwind areas increased by 5-30 µg m −3 and 3-20 ppbv, respectively, and this had important implications for air quality over East Asia (Jeong et al., 2008).
Russia is covered by over 800 million hectares of forest, which is equal to 50 billion tons of growing carbon stock, where annually about 1 % is damaged by fires (Bondur, 2010;Popovicheva et al., 2014).Russian boreal forests are sub-Published by Copernicus Publications on behalf of the European Geosciences Union.
Frequent forest fires over Siberia have an impact on downwind areas in Mongolia, China, Korea, and the Northwest Pacific through long-range atmospheric transport (Kajii et al., 2002;Kanaya et al., 2003;Lee et al., 2005;Jeong et al., 2008;Youn et al., 2011).In May 2003, intense forest fires occurred over Siberia (Lee et al., 2005;Jeong et al., 2008;Youn et al., 2011).Satellite observations clearly show the transport of smoke plumes emitted from Siberian forest fires through Mongolia and eastern China, south to the Korean Peninsula (Lee et al., 2005).Simulations by Youn et al. (2011) showed a significant surface cooling of −3.5 K over forested regions of Siberia.These simulations also showed that smoke aerosol affected large-scale circulation and resulted in an increase in average rainfall rates of 2.9 mm day −1 over the Northwest Pacific.Jeong et al. (2008) reported that smoke plumes from Siberian forest fires in May 2003 acted mainly as a cooling agent, resulting in a negative radiative forcing of −5.8 W m −2 at the surface over East Asia.
Severe wildfires occurred in the forested regions of Russia during summer 2014.The intensity of wildfires during this period was three times larger than in 2013.According to Russia's ITAR-TASS news agency, ∼ 12 600 forest fires had burned over 1.8 million hectares as of 16 July 2014.During this time, the most forest fires occurred in the Irkutsk and Yakutsk areas of Siberia.Over 200 forest fires covering 92 000 hectares occurred in Siberian forested regions as of 16 July 2014 (http://tass.ru/en/russia/740878).MODIS satellite RGB images clearly show that these smoke plumes lasted more than a week and were transported south to Mongolia, northern China, and the Korean Peninsula.
In this study, we investigate the smoke plumes emitted from Siberian forest fires during late July 2014 and their long-range atmospheric transport to the Korean Peninsula.The transport mechanism of these plumes is investigated based on satellite image analysis and satellite-based lidar observations.We also characterize the chemical composition of these plumes over the Korean Peninsula.Chemical characteristics of anthropogenic pollutants from East China transported to the Korean Peninsula in the middle of July 2014 are also investigated and compared with those originating from Siberian forest fires.

Atmospheric aerosol sampling and sample preparation
Table 1 summarizes the measurement parameters and conditions of this study.Daily PM 2.5 (particulate matter with a diameter of ≤ 2.5 µm) sampling was carried out at a regional air-quality monitoring station (36.19 Ultrapure water used in this study was prepared using a Labpure S1 filter and an ultra-violet (UV) lamp (ELGA, PureLab Ultra, USA).Resistivity and total organic carbon (TOC) values of the ultrapure water were maintained at 18.2 M cm −1 and 4 ppb, respectively.To measure carbohydrates and water-soluble ions, a quarter of each filter sample was extracted with 10 mL of ultrapure water under ultrasonication for 30 min, and then passed through a disk filter (0.45 mm, Millipore, Millex-GV, Germany).Water extracts were stored in a refrigerator at 4 • C before analysis.

Analysis of the chemical composition of fine particles
Mass concentrations of PM 2.5 were measured using a betaattenuation technique (BAM 1020, Met One Instruments), with an hourly averaging time resolution.The manufacturer reported the detection limit and measurement error of the beta-attenuation technique as 3.6 µg m −3 and 8 %, respectively.In addition to PM 2.5 mass concentrations, the dailyaveraged chemical composition of PM 2.5 was characterized through filter sampling and laboratory analysis.Because the PM 2.5 chemical composition measurements were made on a daily basis, daily-averaged PM 2.5 mass and chemical compositions were used in this study.

Levoglucosan and mannosan analysis
Levoglucosan and mannosan were analyzed using an improved high-performance anion-exchange chromatography (HPAEC) method with pulsed amperometric detection (PAD) (Engling et al., 2006;Jung et al., 2014).The HPAEC-PAD system uses an ion chromatograph consisting of an electrochemical detector and gold electrode unit, along with an AS40 auto-sampler (Dionex ICS-15000, Thermo Fisher Scientific, USA).Levoglucosan and mannosan were separated in a CarboPak MA1 analytical column (4 × 250 mm) using a sodium hydroxide solution as the eluent.The detection limits of levoglucosan and mannosan were 3.0 and 0.7 ng m −3 , respectively.The calculated values for analytical error, defined as the ratio of the standard deviation to the average value, obtained from triplicate analyses of filter samples, were 1.9 and 0.73 % for levoglucosan and mannosan, respectively.

Organic carbon/elemental carbon analysis
Carbonaceous PM 2.5 was measured using a semi-continuous organic carbon/elemental carbon (OC / EC) analyzer (Model RT3140, Sunset Lab).The air samples were drawn through a PM 2.5 sharp-cut cyclone at 8 L min −1 .The sampled aerosol was then passed through a multichannel parallel-plate denuder with a carbon-impregnated filter to remove semivolatile organic vapors, and then collected on a quartzfiber filter.The sampled aerosol was analyzed based on the thermal-optical transmittance (TOT) protocol for pyrolysis correction and the NIOSH (National Institute for Occupational Safety and Health) method 5040 temperature profile (Birch and Cary, 1996;Jung et al., 2010).External calibration was performed using known amounts of sucrose.The detection limit of both OC and EC is 0.5 µg C m −3 for a 1 h time resolution, according to the manufacturer.The uncertainty of OC and EC measurements has been reported as 5 % (Polidori et al., 2006).

Satellite aerosol optical depth and air mass backward trajectories
The NOAA/ARL HYSPLIT (HYbrid Single-Particle Lagrangian Trajectory) air mass backward trajectory analysis (Draxler and Rolph, 2015;Rolph, 2015) and Moderate Resolution Imaging Spectro-radiometer (MODIS) satellite im- age analysis were used to characterize potential source regions and the transport pathway of the haze plume.Air mass backward trajectories ending at the sampling site (36.19 • N, 127.24 • E) in Daejeon, Korea were computed for heights of 200, 500, and 1000 m a.g.l.using the HYSPLIT model.All back trajectories were calculated at 00:00 and 12:00 UTC (09:00 and 21:00 LT, respectively), extending back 96 h with a 1 h time interval.The calculated air mass pathways indicate the general airflow pattern rather than the exact pathway of air masses, because the typical error in traveled distance is up to 20 % for trajectories computed from analyzed wind fields (Stohl, 1998).
This study used aerosol optical depth (AOD) data retrieved using the NASA MODIS algorithm version V5.2, referred to as Collection 005 (C005) (Levy et al., 2007a, b), which are part of the MODIS Terra/Aqua Level-2 gridded atmospheric data product and are available on the MODIS web site (http://modis.gsfc.nasa.gov/).Cloud-screened level 1.5 sun-photometer data at sites in Yakutsk (61.66 • N, 129.37 • E; 118 m a.s.l.) and Ussuriysk (43.70 • N, 132.16 • E; 280 m above sea level) in Russia were obtained from the AERONET site (http://aeronet.gsfc.nasa.gov).This study used total column-integrated spectral AOD determined using the AERONET algorithm (Dubovik and King, 2000).
CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization) is a space-based lidar system onboard the Cloud Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite launched in 2006 (Winker et al., 2009).This study used version 2.30 data of total attenuated backscatter at 532 nm.Expedited CALIPSO images were obtained from the CALIPSO website (http://www-calipso.larc.nasa.gov/products/lidar/browse_images/show_calendar.php).Figure 2 shows temporal variations in the chemical composition of PM 2.5 at the Daejeon site throughout the entire measurement period.Daily average PM 2.5 mass concentrations ranged from 8.0 to 65.1 µg m −3 with an average of 26.8 ± 15.4 µg m −3 .Two peaks in PM 2.5 mass concentration were observed during 12-16 July (first episode) and 27-28 July 2014 (second episode).PM 2.5 mass concentrations reached 65.1 and 56.2 µg m −3 during the first and second episodes, respectively.The temporal variations in the sum of PM 2.5 chemical compositions show a similar pattern to that of total PM 2.5 mass (Fig. 2).The largest contribution to PM 2.5 mass during the measurement period came from SO 2− 4 , which had a mean of 8.8 ± 7.0 µg m −3 , followed by OC (4.3 ± 2.0 µg m −3 ), NH + 4 (4.3 ± 3.3 µg m −3 ), EC (1.1 ± 0.4 µg m −3 ), and NO − 3 (1.0 ± 1.1 µg m −3 ), with minor contributions from Ca 2+ , K + , and Na + .Figure 4 shows the horizontal distribution of AOD over East Asia from 23 to 28 July 2014.High values of AOD were observed over the Siberian forested region on 23 July, when forest fires occurred.Peak values of AOD were then observed to shift southward to northeast China and the Korean Peninsula from 23 to 28 July 2014 (Fig. 4).These horizontal distributions of AOD also support the transport of smoke plumes emitted from Siberian forest fires onto the Korean Peninsula during late July 2014.
Figure 5 shows temporal variations in AOD measured using a sun photometer at the Yakutsk and Ussuriysk sites.The Yakuksk site is located near the source of Siberian forest fire emissions, whereas the Ussuriysk site is located just to the north of the Korean Peninsula (Fig. 3).The measured AOD at the Yakutsk site started to increase from 23 July, and high AOD continued until 26 July 2014.The AOD dropped to < 0.5 during 06:00-10:00 UTC, 25 July and increased again during 26 July.Because high AOD at the Yakutsk site was caused by transport of the Siberian smoke plume (Fig. 3), the sharp drop in AOD observed during 25 July can be explained by a change in wind direction at the Yakutsk site.The maximum AOD (∼ 4) was observed at the Yakutsk site on 24 July 2014 during a Siberian forest fire event.High values for AOD were observed for 4 days at the Yakutsk site during the Siberian forest fire episode.Interestingly, a sharp increase in AOD was also observed at the Ussuriysk site on 24 July 2014.Spatial distributions of AOD from the MODIS satellite data (Fig. 4) clearly show that the Siberian smoke plumes extended over the Ussuriysk site during 24 July 2014.These results again suggest the transport of Siberian smoke plumes to the northern Korean Peninsula.
Figure 6 shows MODIS RGB images and vertical distributions of total attenuated backscatter at a wavelength of 532 nm measured by the CALIPSO satellite during 24, 25, and 27 July 2014.The left column in Fig. 6 shows MODIS RGB images taken during the Siberian smoke episode.These images show smoke plumes originating from the Siberian forest and being transported over northeastern China.The yellow lines over the images in the left column of Fig. 6 indicate the route of the CALIPSO satellite, and correspond to the x axis of the backscatter plots shown in the right column of Fig. 6.In the total attenuated backscatter measurement plots (Fig. 6, right), red and yellow represent atmospheric aerosol particles and white represents clouds.The spatial distribution of AOD obtained from the MODIS and CALIPSO satellite observations, and the HYSPLIT air mass backward trajectory analysis indicate that smoke plumes originated from Siberian forest fires between 23 and 24 July 2014 and were transported over 3000 km south to the Korean Peninsula between 27 and 28 July 2014.Groundbased AOD measurements using a sun photometer near the Siberian forest fire area and on the Korean Peninsula also support the transport of a smoke plume originating from Siberian forest fires onto the Korean Peninsula.Thus, the smoke episode observed between 27 and 28 July 2014 is hereafter referred to as the Siberian forest fire episode.

Long-range transported haze from Asian continental outflow
Besides the haze episode caused by the long-range transport of smoke emitted from Siberian forest fires during late July 2014, another haze episode was observed at the Daejeon site between 14 and 16 July 2014, as shown in Fig. 2. The MODIS RGB image from 14 July (Fig. 7) shows a severe haze plume originating from East China and extending to the Korean Peninsula across the Yellow Sea.HYSPLIT backward air mass trajectories also indicate the transport of air masses originating in East China to the Korean Peninsula over the Yellow Sea between 15 and 16 July 2014.The region of East China extending from Beijing to Shanghai consists of heavily populated and industrialized cities (Chan and Yao, 2008).Large amounts of anthropogenic pollutants are emitted from this region (L.Li et al., 2015).Figure 8 shows the horizontal distribution of MODIS AOD over East Asia from 13 to 16 July 2014.A trail of high AOD extending from East China to the Korean Peninsula over the Yellow Sea is evident, which suggests that the haze episode observed between 14 and 16 July 2014 was caused primarily by long-range transport of pollutants originating from East China.Thus, the haze episode observed between 14 and 16 July is hereafter referred to as the Chinese haze episode.It has been reported that biomass burning (including biofuel) contributed 14.1 % of the total VOC emissions in China during 2012, whereas in Anhui province the contribution of biomass combustion to VOC emissions was 28.7 % (Wu et al., 2016).B. Li et al. (2015) reported that biomass burning contributed 58 % of OC in Nanjing, China during summer 2012, suggesting that biomass burning is the dominant source of OC in this region.Du et al. (2011) classified the haze events in Shanghai, China during summer 2009 into three categories: biomass-burning induced (high K + , low SO 2− 4 and NO − 3 ), complicated (high SO 2− 4 and NO − 3 , good correlation between K + and SO 2− 4 and NO − 3 ), and secondary (high SO 2− 4 and NO − 3 , low K + ) pollution.Because Anhui, Nanjing, and Shanghai are located near the source of the long-range transported Chinese haze (Fig. 8), the chemical composition of pollution in those areas can be used to understand the Chinese haze episode observed in this study.Temporal patterns in K + concentration are similar to those of SO 2− 4 , and a sharp increase in SO 2− 4 concentration was observed during the Chinese haze episode (Fig. 9).This type of pollution episode is similar to the "complicated" pollution  described by Du et al. (2011), and suggests that the Chinese haze episode was caused mainly by secondary aerosol such as SO 2− 4 and NH + 4 , rather than by biomass burning emissions.
Figure 10 shows relative contributions to PM 2.5 mass during the Chinese haze and Siberian forest fire episodes.Concentrations of organic matter (OM) were reconstructed from measured OC concentrations by multiplying the OM / OC ratio of 1.8 that was measured using an aerosol mass spectrometer in Korea from spring to fall 2011 in the Asian continental outflow (Prof.T. Lee, personal communication, 2015).Huang et al. (2011) reported a similar OM / OC ratio of 1.77 ± 0.08 measured at a downwind site of the highly polluted Pearl River Delta cities in China during fall 2008.During the Chinese haze episode, SO 2− 4 was found to be the dominant species in PM 2.5 mass with an average contribution of 44.2 %, followed by OM (16.6 %) and NH + 4 (19.1 %).This result suggests that the Chinese haze episode can be attributed primarily to anthropogenic pollutants (pos-  sibly emissions from industrial complexes and urban cities in East China).However, during the Siberian forest fire episode, OM was the dominant species in PM 2.5 mass with an average contribution of 38.6 %, followed by SO 2− 4 (16.5 %) and NH + 4 (10.0 %).The high concentration of OM indicates that the Siberian forest fire episode originated primarily from biomass burning.

Comparison of biomass burning tracers during two haze episodes in the Daejeon atmosphere
Levoglucosan and K + are widely used as indicators of biomass burning.Levoglucosan is formed during pyrolysis of cellulose and hemicellulose, and is not emitted from the burning of other materials, such as fossil fuels (Simoneit et al., 1999;Caseiro et al., 2009;Elias et al., 2001).However, caution is required when using K + as a biomass-burning tracer because K + can also be emitted from sea salt and soil (Pio et al., 2008).The mass concentration of biomass burning tracers and their ratios during the Siberian forest fire and Chinese haze episodes are summarized in Tables 2 and 3. Significantly elevated concentrations of levoglucosan were observed during the Siberian forest fire episode, compared with smaller increases observed during the Chinese haze episode (Fig. 9).Concentrations of levoglucosan during the Siberian forest fire episode were measured to be 119.6 ± 6.0 ng m −3 , approximately 6 times higher than those   during the Chinese haze episode (22.3 ± 11.8 ng m −3 ), as listed in Table 2.However, similar levels of K + were obtained during the Chinese haze (0.27 ± 0.08 µg m −3 ) and Siberian forest fire (0.33 ± 0.07 µg m −3 ) episodes.Thus, relatively high levoglucosan / K + ratios were observed dur-ing the Siberian forest fire episode (0.37 ± 0.06) compared with those (0.08 ± 0.03) observed during the Chinese haze episode (Table 3).However, the levoglucosan / mannosan ratios observed during the Siberian forest fire episodes (3.43 ± 0.11) are similar to those observed during the Chinese haze episodes (4.81 ± 0.41), as shown in Table 3. OC concentrations increased as levoglucosan and K + concentrations increased during the Siberian forest fire episode (Fig. 11a).Elevated OC / EC ratios were also observed during the Siberian forest fire episode (7.18 ± 0.2).Simultaneous increases in K + , OC (Fig. 11b), and levoglucosan concentrations (Fig. 11c) during the Siberian forest fire episode suggest that the K + originated primarily from the smoke plume during the Siberian forest fire episode.
OC and levoglucosan concentrations observed during the Chinese haze episode are similar to those observed during the non-episode period, as shown in Fig. 11a.However, small increases in K + concentration were observed during the Chinese haze episode, as shown in Fig. 11b, resulting in relatively small levoglucosan / K + ratios during the Chi-nese haze episode (0.08 ± 0.03) compared with those during the Siberian forest fire episode (0.37 ± 0.06).This difference in levoglucosan / K + ratios can be explained as follows.First, different types of biomass burning might have occurred during the Chinese haze episode compared with the Siberian forest fire episode.It can be postulated that biomass-burning emissions with relatively low OC / K + and levoglucosan / K + ratios might have contributed to observations made on the Korean Peninsula during the Chinese haze episode.
Second, K + measured during the Chinese haze episode may have originated from sources other than biomass burning.Because OC is predominantly emitted from biomass burning, biomass-burning particles have relatively high OC / EC ratios and are generally well correlated with biomass burning tracers (Cao et al., 2008;Cheng et al., 2008;Popovicheva et al., 2014).The lack of significant increases in OC / EC ratio (2.4 ± 0.4), and OC and levoglucosan concentrations during the Chinese haze episode compared with nonepisode measurements suggests that the elevated K + concentrations observed during the Chinese haze episode might be due to emissions from other sources, such as soil, sea salt, or industrial complexes.Chow et al. (2004) reported that 3.9-12.5% of PM 2.5 consisted of K + in stack samples from cement kiln manufacturing processes.Positive correlations of K + with SO 2− 4 and EC concentrations during the Chinese haze episode (Fig. 9) also suggest that there were additional emissions of K + from anthropogenic sources other than biomass burning.
Elevated concentrations of levoglucosan and OC, and relatively high OC / EC ratios (7.18 ± 0.2) suggest that the haze episode that occurred during late July 2014 was caused primarily by the long-range transport of smoke emitted from Siberian forest fires.However, significantly elevated SO 2− 4 concentrations with relatively weak increases in OC and levoglucosan concentrations and lower OC / EC ratios indicate that the Chinese haze episode was caused primarily by anthropogenic pollutants emitted from industrial complexes and urban cities in East China, with relatively little contribution from biomass burning.
Because levoglucosan and mannosan are emitted from similar burning processes, the Levo / Man ratio can be used to track the type of biomass burning.Levo / Man ratios observed during the Siberian forest fire episode are similar to those obtained from the softwood and leaf burning experiments, and the smoke episode in Moscow, Russia during summer 2010.However, Levo / Man ratios during the Siberian forest fire episode are much lower than those reported for hardwood, grass, and crop residue burning.
Levo / K + ratios observed during the Siberian forest fire episode are close to those reported for grass, crop residue, and leaf burning, as well as to the ratios of the smoke episode in Moscow, but much lower than those from hardwood and softwood burning (Fig. 12b).Levoglucosan can be removed through photo-oxidative decay during atmospheric transport (Hennigan et al., 2010), but K + is relatively stable in the atmosphere.Thus, Levo / K + ratios can decrease during longrange atmospheric transport.The Levo / K + ratios observed during the Siberian forest fire episode were lower than those during the smoke episode in Moscow, Russia in summer 2010, which can be explained by photochemical degradation of levoglucosan during long-range atmospheric transport.
Based on a comparison of biomass burning tracers from various sources (Fig. 12), it is suggested that smoke aerosol emitted during the Siberian forest fire episode originated mainly from the burning of forest leaves in Siberia prior to their long-range atmospheric transport.Smoke aerosol observed during the smoke episode in Moscow, Russia in summer 2010 have similar Levo / Man and Levo / K + ratios to those from leaf burning (Fig. 12).These observations suggest that smoke episodes in the Russian forest originate primarily from the burning of forest leaves.

Conclusions
This study investigated the long-range transport of smoke plumes emitted from Siberian forest fires during late July 2014.Smoke plumes emitted from Siberian forest fires are generally transported to the Northwest Pacific by prevailing westerlies.However, the haze plume that occurred during late July 2014 had a significant impact on the Korean Peninsula, which is located ∼ 3000 km south of the Siberian forest.From the spatial distributions of AOD obtained from the MODIS satellite, CALIPSO satellite observations, and HYSPILT air mass backward trajectory analyses, it is evident that smoke plumes originating from Siberian forest fires between 23 and 24 July 2014 were transported over 3000 km south to the Korean Peninsula between 27 and 28 July 2014.During this episode, elevated concentrations of levoglucosan (119.6 ± 6.0 ng m −3 ) and K + (0.33 ± 0.07 µg m −3 ), and high OC / EC ratios (7.18 ± 0.2) were observed at a measurement site in Daejeon, Korea.These results suggest that the haze episode that occurred during late July 2014 was caused mainly by the long-range transport of smoke plumes emitted from Siberian forest fires.The Siberian smoke episode is clearly distinguished from a haze episode caused by the longrange transport of anthropogenic pollutants emitted from East China, which was characterized by elevated SO 2− 4 concentrations and weak increases in OC and levoglucosan concentrations.

Figure 2 .
Figure 2. Temporal variations in the chemical components of fine particulate matter (PM 2.5 ) at the Daejeon site during July 2014.

3. 2
Classification of haze episodes during summer 2014 3.2.1 Long-range transported smoke plumes from Siberian forest fires The MODIS RGB images clearly show severe smoke plumes over the Siberian forested region during late July 2014.Figure 3a shows a typical example from 25 July 2014 of satellite RGB images of the smoke plumes emitted from Siberian forest fires and their atmospheric transport to the south.Fire events over the Siberian forested region are indicated by red dots in Fig. 3a.It is clear that the smoke plumes originated in Siberia and were transported south to the Korean Peninsula across Mongolia and northeast China.HYSPLIT backward trajectory analyses (Fig. 3b) also indicate that the air masses originated in the Siberian forested region and were transported to the Korean Peninsula between 26 and 28 July 2014.

Figure 3 Figure 3
Figure 3 Figure 6a and b clearly show that between 24 and 25 July 2014, a smoke layer existed approximately 3-5 km in height near the source region of the Siberian forest fires.As shown in Fig. 6c, the height of the smoke layer decreased to below 2 km on 27 July 2014 as it reached the Korean Peninsula.

Figure 5 .
Figure 5. Temporal variations in AOD measured by a sun photometer at the Yakutsk and Ussuriysk sites in Russia during July 2014.

Figure 6 .
Figure 6.MODIS RGB images and vertical profiles of total attenuated backscatter at 532 nm measured by the CALIPSO satellite during (a) 24, (b) 25, and (c) 27 July 2014.Yellow lines in the MODIS RGB images indicate the route of the CALIPSO satellite, and correspond to the x axis in the vertical profiles of total attenuated backscatter.

RangeFigure 10 .
Figure 10.Average relative contributions to PM 2.5 mass during the (a) Chinese haze and (b) Siberian forest fire episodes.

Figure 11 .
Figure 11.Scatter plots of OC vs. (a) levoglucosan and (b) K + , and levoglucosan versus (c) K + and (d) mannosan between 4 and 31 July 2014.Filled black and red diamonds represent the Chinese haze and Siberian forest fire episodes, respectively.Open black circles represent the remaining sampling days in July 2014.

Table 1 .
Measurement parameters and conditions of the present study.

Table 2 .
Summary of fine particle (PM 2.5 ) mass, and organic and inorganic chemical composition of PM 2.5 particles during the Chinese haze and Siberian forest fire episodes measured at Daejeon, Korea during summer 2014.

Table 3 .
Summary of ratios of biomass burning tracers during the Chinese haze and Siberian forest fire episodes, as measured at Daejeon, Korea in summer 2014.