Chemical Characterization of Wintertime Aerosols over Islands and Mountains in East Asia : Impacts of the Continental Asian Outflow

This study aimed to characterize the wintertime surface aerosol chemistry over islands and mountains sites (i.e., Cape Fuguei, Mt. Bamboo, Mt. Lulin, Cape Hedo, and Kumamoto) in East Asia. Aerosols were sampled over a 24-h period as part of an intensive observational period (IOP) in winter 2015. Aerosol samples were analyzed for water-soluble inorganic ions (WSIIs), organic carbon (OC), and elemental carbon (EC). PM2.5 mean concentration (in μg m) was found the highest over Kumamoto (22 ± 7), followed by Cape Fuguei (20 ± 9), Cape Hedo (11 ± 5), Mt. Bamboo (10 ± 13), and Mt. Lulin (4 ± 3). Strong correlations (r > 0.91) in ion charge balance suggested the good quality of data-sets and the ions share common source origins. Larger variations in (non-sea-salt-sulfate) nss-SO4 and NH4 over all the sites indicated the significant contribution of anthropogenic emissions from continental Asian outflow. OC was found the most abundant resolved component in PM2.5 over Mt. Lulin (37.58 ± 25.90%) and Mt. Bamboo (33.24 ± 24.11%) than that over Cape Fuguei (11.94 ± 3.48%). OC3 (evolved at 280–480°C) was the most abundant in OC over all the sites, indicating the possible contributions of biomass-burning (BB) while EC1-OP (EC evolved at 580°C minus the pyrolized OC) was found the highest in EC over Cape Fuguei. Analysis of back-trajectories and fire-counts suggested the influence of long-range transported BB from continental Asian outflow. Higher concentrations of PM2.5 along with the BB tracers (i.e., NO3, OC3, and EC1-OP) were also observed during the influence of continental Asian outflow. This study provides needful information to understand the effect of continental Asian outflow on the air quality over East Asia. We propose the longterm and extensive field measurements to upgrade our knowledge on continental Asian outflow BB influence over islands and mountains in East Asia.


INTRODUCTION
Atmospheric aerosols influence solar radiations (Charlson et al., 1992), cloud microphysics (Penner et al., 2001), precipitation efficiency and photochemistry of troposphere (IPCC, 2001), global climate (Charlson et al., 1992), and human health (IPCC, 2007).Arising from both natural (e.g., soil dust, sea-salt, botanical debris, volcanic dust, forest fires, and gas-to-particle conversion) and anthropogenic (e.g., industrial processes, vehicular emissions, fossil-fuel and open burning) sources (IPCC, 2001), aerosol plays a crucial role among their sources, transport, and deposition mechanisms of various pollutants in the atmosphere.Aerosol distribution is controlled by emissions, transport, transformation and loss processes in the atmosphere (Sahu et al., 2009).Aerosols can be transported from areas of high emissions to relatively clean remote sites regionally.Atmospheric transport is the extensive pathway of various chemical species from land to open ocean (Simoneit, 2002(Simoneit, , 2004) ) and remote locations (Lee et al., 2011).The chemical characterization of atmospheric aerosols is vital to determine their effects on climate and human health (Watson, 2002) and also to understand their various source origins.
Emission inventories show large regional differences mainly due to contributions from various sectors with different emissions factors in East Asia (Woo et al., 2003;Bond et al., 2004).A number of aerosol measurements have been made to investigate the aerosol transport and transformation processes over the downwind regions of the East Asian continent (Sahu et al., 2009;Kundu et al., 2010a;Kunwar and Kawamura, 2014;Shimada et al., 2015;Zhu et al., 2015;Shimada et al., 2016).The most systematic and extensive measurements were made during the Transport and Chemical Evolution over the Pacific (TRACE-P) campaign in the spring of 2001 (Jacob et al., 2003) and the Aerosol Characterization Experiment (ACE)-Asia 2001 (Huebert et al., 2003).Rapid economic growth and urban development cause large scale anthropogenic emissions of aerosol and trace gases in East Asia (Streets and Waldhoff, 2000;Streets et al., 2003).Apart from anthropogenic emissions, East Asia is one of the most active biomass burning (BB) regions in the world with significant aerosol loadings (van der Werf et al., 2006).BB is a common energy source for cooking and space heating in East Asia (Verma et al., 2015).
Regionally, the largest contributions of BB are from boreal Asia (Russia), South East Asia (SEA; comprising Vietnam, Laos, Combodia, Thailand, Myanmar, and west Malaysia), China, and Mongolia (Streets et al., 2003;van der Werf et al., 2006van der Werf et al., , 2010)).Open BB (both forest fires and open burning of crop residues) is a dominant source in boreal Asia and SEA (Streets et al., 2003).Intensive BB emissions frequently occur in the SEA during every spring and investigated elsewhere (Lin et al., 2013(Lin et al., , 2014;;Lee et al., 2016;Pani et al., 2016a, b;Khamkaew et al., 2016;Tsay et al., 2016 and references therein).Burning of grassland is dominant in Mongolia while burning of crop residues is dominant in China (Streets et al., 2003).Large amounts of crop residues are burned in the field during the postharvest seasons (i.e., May-June and October-November) in eastern China (Streets et al., 2003;Yan et al., 2006).The open BB could cause severe regional air pollution and haze issue over Beijing-Tianjin-Hebei areas of China, the Pearl River delta, and the Yangtze River delta (Zhu et al., 2010;Cheng et al., 2014 and references therein).Several studies reported significant impact of BB emissions on air quality over northern China and Mongolia in winter and remarkably in Beijing (Wang et al., 2006;Yu et al., 2013;You et al., 2015 and references therein).Atmospheric aerosols were greatly influenced by the burning of agriculture crop residue and fallen leaves in Northern China during late September to early November (You et al., 2015).The prevalence of dry weather since late autumn at most regions of northern China (Zhai et al., 2005) increases the risks of forest and grassland fires (You et al., 2015).
Emissions from such burning activities can affect regional air quality and climate through long distance transport (Mochida et al., 2010;Zhu et al., 2015).Springtime regional transport of SEA BB downwind to southeastern China, South China Sea (SCS), East Asia, western and central Pacific Ocean, northeastern China, even over Japan were also investigated within the Seven South East Asian Studies (7-SEAS) framework (Lin et al., 2013).Plume signatures were also investigated above boundary layer over Hong Kong (Chan et al., 2003;Deng et al., 2008), at Mt. Lulin in central Taiwan (Sheu et al., 2010;Lee et al., 2011;Lin et al., 2014;Chuang et al., 2014), and possibly even near the surface in the southern Taiwan (Tsai et al., 2013;Yen et al., 2013).Liu et al. (2003) analyzed the contributions from various sources (i.e., fossil fuel, biofuel, and BB) to the Asian outflow over the West Pacific by modelling studies.Anthropogenic emissions of aerosol and trace gases from Asian outflows (i.e, central, northern and northeastern China) were investigated over Jeju Island, Gosan, Korea in spring of 2005 (Sahu et al., 2009).These outflow emissions from the Asian continent were associated with a high-pressure system centered over the northern region of China and Mongolia (Sahu et al., 2009).Seasonal variations of diacids, ketoacids, and αdicarbonyls in long-range transported aerosols from Asian continent were investigated at Jeju Island, Korea during April 2003-April 2004(Kundu et al., 2010a).Kundu et al. (2010a) demonstrated that the Asian outflows from northeast China most likely affect the air quality of Jeju Island and the surrounding ocean in western North Pacific Rim (NPR).Kunwar and Kawamura (2014) investigated the carbonaceous and nitrogenous components as well as major ions in aerosols from subtropical Okinawa Island, Japan and reported important contribution from Asian dusts and anthropogenic sources including BB, vehicular emissions and coal combustions.Zhu et al. (2015) studied the wintertime maxima of anhydrosugars in ambient aerosols from Okinawa Island and reported the effect of East Asian BB emissions on the air quality in the western NPR.BB organic tracers (i.e., levoglucosan, mannosan, and galactosan) in Asian aerosols transported downwind to the western NPR by westerly winds were investigated and studied over Chichijima Island in the western NPR (Verma et al., 2015).Similarly, Deshmukh et al. (2016) extensively investigated the size distributions and formation processes of various water soluble diacids, carbon fractions and inorganic ions in spring aerosols at Okinawa Island, Japan and demonstrated the significant influence of anthropogenic and BB burning aerosols emitted from East Asia on the molecular compositions of watersoluble organic aerosols in the western NPR.
The abovementioned studies investigated the chemical profile of ambient aerosols and identified some tracers to understand the influence of long-range transported anthropogenic and BB emissions from Asian outflows over NPR.Influence of BB emissions from Asian outflows were identified and reported most significant in winter over the outflow region of East Asia (e.g., Korea and Japan) and western NPR.However, the influence of BB emissions from Asian outflows on the aerosol chemical characteristics over islands and mountains in East Asia (e.g., Taiwan) are not yet thoroughly investigated and well understood.Aerosol chemical characterization is essential in identifying the source influence in the process of source apportionment.The main goal of this work is to characterize the signature profiles of anthropogenic or BB emissions from continental Asian outflow, with reference to the composition of water soluble inorganic ions (WSIIs) and carbonaceous fractions in wintertime aerosols collected over islands and mountains in East Asia.

Site Description, Aerosol Sampling, and Chemical Analysis
Aerosol sampling was conducted at three distinct sites i.e., Cape Fuguei (25.29°N, 121.53°E, 10 meter above mean sea level (amsl)), Mt. Bamboo (25.18°N, 121.53°E, 1103 m amsl), andMt. Lulin (23.47°N, 120.87°E, 2862 m amsl) in East Asia (Fig. 1) during the winter period (25 October-10 November 2015; 17 days) as a part of an intensive observational period (IOP).Cape Fuguei is located on the northwestern tip of Taiwan and mainly covered with dwarfed and twisted local vegetation due to the constant sea winds.Owing to its location, it is free from all types of local activities for most of the time, particularly for northeast monsoon prevailing period.Mt.Bamboo is located at an isolated peak inside the Yangmingshan National Park in northern Taiwan and about 17 km away from the northern Taiwan coastline.This site primarily receives the northeasterly monsoonal flow and frontal passages during the winter months and frequently immersed in clouds.Moreover this site is occasionally located above the planetary boundary layer and the boundary layer dynamics regulate the vertical transport of aerosols from the sea surface to this site like the Yangmingshan mountain (826 m amsl) site (Chou et al., 2017).A more detailed description about Mt.Bamboo site can be found elsewhere (Lin and Peng, 1999;Wang et al., 2010;Sheu and Lin, 2011).Mt.Lulin is a high-elevation background station located inside the Jade Mountain National Park in Taiwan and a part of the Central Mountain Range.This site is also free from direct influence of local pollutions and lies in the free troposphere.Moreover, this is an ideal site to conduct measurements of long-range transported air pollutants in the free troposphere in East Asia (Wai et al., 2008;Sheu et al., 2010;Lee et al., 2011;Chuang et al., 2014).More detailed descriptions of the Mt.Lulin site are available in the literature (Sheu et al., 2010).
Sampling of PM 2.5 (cut sizes ≤ 2.5 µm) aerosols for chemical analysis was made continuously over Cape Fuguei (12 days; 25 October-3 November and 9-10 November, 2015), Mt. Bamboo (14 days;28 October-10 November, 2015), and Mt. Lulin (14 days;28 October-10 November, 2015).Ambient PM 2.5 and PM 2.5-10 (2.5 ≤ cut sizes ≤ 10 µm) aerosol samples over Cape Fuguei were manually collected for 24 hours (08:00 a.m. to 08:00 a.m.local time) by using a Dichotomous Sequential Air Sampler (R&P Partisol 2025; Rupprecht & Patashnick, Co.).The mass concentrations of both PM 2.5 and PM 2.5-10 were weighted after appropriate conditioning for each samples and used for further chemical analysis.However, 24 hourly PM 2.5 samples over Mt.Bamboo (08:00 a.m. to 08:00 a.m.local time) and Mt.Lulin (09:00 a.m. to 09:00 a.m.local time) were collected by using collocated R&P ChemComb Model 3500 Speciation Sampling Cartridges (Thermo Fisher Scientific Co., Inc., Waltham MA, USA) supported by tripods, as described in Chuang et al. (2013b).After the sampling collections, the collected filters (both Teflon and quartz fibers) were stored in refrigerator at 4°C to maintain their chemical stability, and transported back to the laboratory in Taiwan for filter weighing, conditioning, and subsequent chemical analysis.Proper care was taken during collection and preservation of the samples until the analysis.
to ensure the correctness of the analyzed signals.More details of the sampling procedure and analytical methods for WSIIs can be found elsewhere (Lee et al., 2011;Chuang et al., 2013a, b;Lee et al., 2016).
The quartz filters (TISSUQUARTZ 2500QAT-UP, PALL Life Sciences, Inc., Ann Arbor, MI, USA) were analyzed for carbonaceous contents (OC and EC) by using a carbon analyzer (DRI Model 2001A, Atmoslytic Inc., Calabasas, CA, USA).The filter punch (0.5 cm 2 ) was analyzed for 8 carbonaceous fractions (OC1, OC2, OC3, OC4, OP, EC1, EC2, and EC3) following the IMPROVE (Interagency Monitoring of Protected Visual Environments) protocol with the TOR (thermal optical reflectance) correction scheme (Chow et al., 1993;Chow et al., 2004;Watson et al., 2005).Four OC fractions (OC1, OC2, OC3, and OC4) were determined at 140°C, 280°C, 480°C and 580°C, respectively in a pure He atmosphere.EC1, EC2, and EC3 were determined at 580°C, 740°C, and 840°C, respectively in a 2% O 2 /98% He atmosphere.The residence time of each heating step was defined by the flattening of the carbon signal.The pyrolyzed carbon fraction (OP) was determined when reflected laser light returns to its initial value after oxygen is introduced.Generally, OC is defined as OC1 + OC2 + OC3 + OC4 + OP and EC is defined as EC1-OP + EC2 + EC3.The detection limit for the carbon analyzer was 0.05 µg C cm -2 for both OC and EC.The resolved EC fractions were further classified into char-EC (EC1-OP) and soot-EC (EC2 + EC3) (Han et al., 2007;Chuang et al., 2013b) for source identifications.More details of the sampling procedure and analytical methods for carbonaceous fractions can be found elsewhere (Chuang et al., 2014;Lee et al., 2016).
In order to extend our knowledge about East Asian aerosols, PM 2.5 aerosols were also sampled over Cape Hedo,Okinawa Island,Japan (26.87°N,128.25°E;60 m amsl) and Kumamoto,Kyushu Island,Japan (32.8°N,130.7°E) in East Asia during the same study period in addition to the aforementioned sites.PM 2.5 sampling was made continuously over Cape Hedo (13 days;26 October-07 November, 2015) and Kumamoto (7 days;26-31 October, 2015).Detailed descriptions of these sites can be found elsewhere (Moreno et al., 2013;Shimada et al., 2015;Misawa et al., 2017).In brief, Cape Hedo is situated at the northern end of Okinawa Island, about 100 km from Naha, the largest city on the island, and about 650 km from Shanghai, a major city in mainland China.There are no such large residential or industrial areas near this site.Kumamoto sampling site is situated on the island of Kyushu in Western Japan (Moreno et al., 2013, Fig. 1) approximately midway between Tokyo (~1000 km in northeast direction) and Shanghai (~1000 km southwest direction).This site is not impacted by any nearby heavy industrial point sources and is an excellent site for investigating the characteristics of transboundary aerosol intrusions (Moreno et al., 2013).24 hourly PM 2.5 aerosols over Cape Hedo and Kumamoto were sampled with six-stage cascade impactor (Model 3180, KANOMAX, Osaka, Japan) and analyzed only for WSIIs.More detailed about the sampling and analysis procedures can be found elsewhere (Itahasi et al., 2017;Tatsuta et al., 2017).

Estimate of Secondary Organic Carbon and Organic Matter
The EC-tracer method (e.g., Turpin and Huntzicker, 1991) was used to estimate the contributions of the primary organic carbon (POC) and secondary organic carbon (SOC) to total carbonaceous aerosols.EC and POC commonly have the same source origins (Pavuluri et al., 2011); hence the EC-tracer method assumes the relatively constant OC/EC ratios for given area, season and local meteorology (Kunwar and Kawamura, 2014).The minimum OC/EC ratio was used to estimate the SOC formation widespread (Turpin and Huntzicker, 1995;Pavuluri et al., 2011).In this study, the concentrations of SOC and POC were estimated by the following equations min where OC meas is the measured OC concentration, (OC/EC) min is the estimated minimum OC/EC ratio during the sampling period, k is the parameter for non-combustion sources contributing to the POC that is assumed to be negligible.The minimum OC/EC ratios in PM 2.5 were 3, 1.1, and 2.3 over Cape Fuguei, Mt.Bamboo, and Mt.Lulin, respectively (supplement Fig. S1).
All the aforementioned sampling sites in this study are free from local anthropogenic activities; located in the Asian outflow region and receiving the long range transported aerosols.Aerosols are subjected to aging processes during the atmospheric transport and result in more oxygenated organic species (Kundu et al., 2010;Kunwar and Kawamura, 2014).Turpin and Lim (2001) reported that organic matter (OM) in the atmosphere estimated by the product of measured OC concentration with a conversion factor of 1.6 ± 0.2 (urban aerosols) and 2.1 ± 0.2 (aged aerosols).We used the conversion factor of 2.1 in this present study to estimate the OM concentrations.

Backward Air Mass Trajectories and Fire Counts
Trajectory is the time-integration of the change in position of an air parcel as it is transported by the wind.Air mass trajectories were calculated in a backward mode (path of air movement arriving at a receptor location) by using the HYSPLIT-4 (Hybrid Single-Particle Lagrangian Integrated Trajectory -Version 4) model developed by NOAA (National Oceanic and Atmospheric Administration, USA) Air Resources Laboratory (http://www.arl.noaa.gov/ready/hysplit4.html)during the study period (Draxler and Rolph, 2003).The seven-day back-trajectories (BT) ending at ground level of all the three aforementioned study sites at 00/06UTC daily for the measurement days were analyzed in order to investigate the influence of continental Asian outflows (anthropogenic/BB) on ambient aerosols.A total number of BTs studied (NB) during the sampling period (1 BT daily) over Cape Fuguei, Mt.Bamboo, and Mt.Lulin were 12, 14, and 14, respectively and their relative frequency distributions (e.g., Pani and Verma, 2014;Verma et al., 2014) were calculated.Burning activities in East Asia were illustrated by fire-counts and were obtained from the MODIS (Moderate Resolution Imaging Spectroradiometer) Terra satellite (http://disc.sci.gsfc.nasa.gov/neespi/data-holdings/mod14cm1.shtml).

Air Mass Origins and Fire Counts
Representative BT paths (Figs.2(a)-2(c)), indicating the influence of air masses originating from different source regions on ambient aerosols at Cape Fuguei, Mt.Bamboo, and Mt.Lulin sites.The details of possible pathways and their relative frequency (%) are presented in Table 1.BT pathways to Cape Fuguei (NB: 12) and Mt.Bamboo (NB: 14) were found very similar but distinct over Mt.Lulin (NB: 14).Entrainment of air pollutants along each BT was considered more alike.Originating from the Yellow Sea (YS) or East China Sea (ECS), the Path A carries mainly marine aerosols to Cape Fuguei (33%), Mt.Bamboo (21%), and Mt.Lulin (14%).Similarly, Path B influences Cape Fuguei and Mt.Lulin site with mostly marine aerosols.Path C carries mostly continental Asian outflow to Cape Fuguei (42%) only.Path D carries both oceanic and continental aerosols to Mt. Bamboo (29%) only and Path E carries mainly mixed aerosols only to Mt. Lulin (29%).BT pathways to Mt. Lulin during October to November indicated that most of the trajectories traverse over a marine atmosphere and the coastal areas of China and Korea (Wai et al., 2008, Fig. 3(c)).Path F mostly carries Asian outflow to all the sites i.e., Cape Fuguei (17%), Mt.Bamboo (21%), and Mt.Lulin (14%).Cape Fuguei was influenced by the Asian outflow through both Paths C and F, whereas Mt.Bamboo and Mt.Lulin were only through Path F. Similar BT results were derived for investigation the origin of air masses over the Dongsha Island (Pani et al., 2016c) in the northern SCS and Yangtze Delta region of China (Liu et al., 2012).Zhu et al. (2015) reported the influence of 65-70% of air masses originated in western to central Russia and passed through Mongolia and northern and northeastern China to Cape Hedo, while another 30-35% of air masses originating from the northern and northeastern China with shorter transport distances.Path G carries mostly marine aerosols to Mt. Bamboo (29%) and Mt.Lulin (29%) from the western Pacific.Cape Hedo was also influenced by both oceanic (from western Pacific) and continental (from Asian outflows) air masses (Kunwar and Kawamura, 2014) during autumn (September, October and November).Strong influence of continental air masses was detected over Cape Fuguei and Mt.Bamboo as compared to Mt. Lulin in this study.Cape Hedo receives the air masses exclusively originating from the Asian continent during autumn (Zhu et al., 2015, Figs. 2(k)-2(l)) and winter (Zhu et al., 2015, Figs. 2(a)-2(b)) with strongest influence of Asian outflow dominated by the winter Asian monsoon.Air mass was frequently transported from the Asian continent to the Kumamoto during October (Misawa et al., 2017, Figs. 5(a)-5(b)).
Vander Werf et al. (2006) reported a bimodal pattern with two maxima of BB emissions in spring and autumn in central Asia (Mongolia, China, and Japan) on the basis of satellite observation and model simulation.Monthly value of total fire-counts in middle-to-north Asia (30-60°N, 80-130°E) showed a secondary peak (total counts 17838) in October (Zhu et al., 2015, Fig. S1 (Zhu et al., 2015).On the basis of the BT paths and fire-counts analyzed, it could be summarized that aerosols over the sampling sites in the present study were influenced by the BB emissions from open burning of wheat and maize straw in the northern and northeastern China.

Distribution of Aerosol Mass
The simultaneous measurement of PM 2.5 concentrations obtained gravimetrically from Dichotomous Sequential Air Sampler (R&P Partisol 2025) and from a semi-continuous particulate monitor i.e., Tapered Element Oscillating Microbalance (TEOM 1405-DF, Thermo Scientific Inc.) were strongly correlated (r = 0.99) over Cape Fuguei.For comparison purpose only, we used the PM 2.5 mass from TEOM with that collected from Cape Hedo (supplement Fig. S2) during the same sampling period.PM 2.5 (from TEOM) mass levels were 18 ± 12 µg m -3 (range: 2-44 µg m -3 ) and 13 ± 6 µg m -3 (6-30 µg m -3 ) over Cape Fuguei and Cape Hedo, respectively.PM 2.5 mass over Cape Fuguei was found higher than that of Cape Hedo in most cases and also similar/comparable in some cases.The difference in mass concentrations over two locations (r = 0.02) may be due to the alternation of air masses and the relative contribution from the continental Asian outflows.Both the sampling sites were free from local sources and generally received the long-range-transported pollutants from continental outflows associated with a high-pressure system centered over the northern region of China and Mongolia.Recently, Jeong and Park (2016) investigated the effects of East Asian winter monsoon on wintertime aerosol concentrations over East Asia by using GEOS-Chem 3D model and reported that higher concentrations over the southern parts of East Asia during strong winter monsoon years and opposite pattern during weak monsoon years.Shimada et al. (2015) also inferred that during spring and winter, atmospheric pollutants were transported from the Asian continent to Cape Hedo by migratory high-pressure systems and cold fronts.The mean aerosol mass loading (total suspended particles, TSP) at Okinawa was 74 µg m -3 (range: 19-286 µg m -3 ) during one-year observation period of October 2009-October 2010 (Kunwar and Kawamura, 2014).
Moreover, we also compared the gravimetric PM 2.5 mass over Cape Fuguei with that over Cape Hedo and Kumamoto (supplement Fig. S3) during the same sampling period.The average PM 2.5 mass concentration (Tables 2(a) and 2(b)) over Cape Fuguei (20 ± 9 µg m -3 ) was found comparable with Kumamoto (22 ± 7 µg m -3 ) but relatively lower PM 2.5 mass was observed over Cape Hedo (11 ± 5 µg m -3 ).Shimada et al. (2015) also reported the mass of PM 2.5 (10 ± 5 µg m -3 ) at Cape Hedo, Okinawa in during 23-26 October, 2010.Maximum (~80% of the total) air masses in spring and winter were transported from China to Cape Hedo as estimated from the back trajectory analysis (Shimada et al., 2011).However, higher PM 2.5 (~30 µg m -3 ) mass were also observed in winter and spring over Kumamoto (Misawa et al., 2017, Fig. 2(a)) mainly due to the combustion of coal with higher sulfur contents for the household heating in the northeast area of China (Kurokawa et al., 2013).Transboundary air pollution from the Asian continent due to seasonal monsoons also found as a significant contributor to higher aerosol loading over Kumamoto (Misawa et al., 2017).
During the sampling period, the average PM 10 mass concentration was 47 ± 10 µg m -3 (number of sampling days, N = 10, range: 29-63 µg m -3 ) over Cape Fuguei.The mean mass concentrations were 20 ± 9 µg m -3 (N = 10, range: 8-37 µg m -3 , and 42 ± 10% in PM 10 ) and 27 ± 4 µg m -3 (N = 10, range: 21-35 µg m -3 , 58 ± 10% in PM 10 ) for PM 2.5 and PM 2.5-10 , respectively over Cape Fuguei (Table 2(a)), indicating the presence of more coarse aerosols.The aerosol mass was varied widely over this site.The ratio of PM 2.5 to PM 10 (PM 2.5 /PM 10 ) was 0.4 ± 0.1, which also indicates relatively higher dominance of coarse aerosols.The PM 2.5 level over Cape Fuguei was within the limit of World Health Organization (WHO) 24-hour guideline value (25 µg m -3 ) on 8 out of 10 days during sampling period.This finding may not suggest health risk of human exposure but climatic effect of aerosols at this rural remote site can't be avoided.The mean levels of PM 2.5 and PM 2.5-10 mass over Cape Fuguei were found comparable with those reported in other studies conducted over nearby islands and marine locations.Chou et al. (2010) reported similar mass of PM 2.5 (23 ±   (5 µg m -3 ) and 10 µg m -3 (6 µg m -3 ), respectively during January 2003 (Cruise I) and April 2003 (Cruise II) over Pearl Estuary and waters near Hong Kong (Zhang et al., 2007), clearly indicating the presence of higher fine aerosols.PM 2.5 mass concentrations were 20 ± 16 µg m -3 and 13 ± 11 µg m -3 over Cheji Island, near southern Korea and Sado Island, Japan, respectively during 1 January-30 June 2002 (Cohen et al., 2004).The average PM 2.5 and PM 2.5-10 mass concentrations were 13 ± 6 and 14 ± 15 µg m -3 , respectively during the spring 2010 over Dongsha Island, a remote area over the northern SCS (Chuang et al., 2013a).The PM 2.5 mass concentrations over Mt.Bamboo and Mt.Lulin were 10 ± 13 µg m -3 (N = 14, range: 1-38 µg m -3 ) and 4 ± 3 µg m -3 (N = 14, range: 0-9 µg m -3 ), respectively.It is worth to note that the PM 2.5 mass over Mt.Bamboo was 3 ± 3 µg m -3 (N = 11, range: 1-8 µg m -3 ) and found similar to Mt. Lulin, excluding the episodic days (PM 2.5 mass was 34, 27, and 38 µg m -3 on 29 October, 7 and 8 November, 2015, respectively and all were exceeding the WHO guidelines) with more influence by either local or longrange transported aerosols.The mean levels of PM 2.5 mass were comparable with those reported over other high altitude stations.PM 2.5 mass concentrations over Mt.Lulin were 18 ± 8 µg m -3 (for the air masses transported from the BB area; BB group), 5 ± 2 µg m -3 (from BB source areas during the non-BB period; SNBB group), 10 ± 4 µg m -3 (from anthropogenic sources; Anthropogenic group), and 4 ± 2 µg m -3 (from the oceanic area and the free troposphere above the 700-hPa pressure level over the Asian continent; FT group) during April 2003-April 2009 (Lee et al., 2011).PM 2.5 mass over Mt.Lulin in this study was found closer to the values reported earlier by Lee et al. (2011).PM 2.5 levels over high altitude stations in China i.e., Qilian Shan Station of Glaciology and Ecologic Environment, Qinghai-Tibet Plateau (4180 m amsl) and Quinghai Lake (3200 m amsl) were 3 and 6 times higher than that of Mt.Lulin.Similar PM 2.5 mass over Jungfraujoch, Switzerland (4 ± 2 µg m -3 ) was reported by Krivacsy et al. (2001).PM 2.5 mass level around 4 µg m -3 can probably represent global clean and pristine area in free troposphere (Lee et al., 2011).
Aerosol mass concentrations varied greatly for different samples.Temporal variations in the PM 2.5 bulk contents of WSIIs and carbonaceous fractions during the study periods over all the study sites are shown in Fig. 3. Bulk anions were found to dominate the distribution, followed by cations and OC over Cape Fuguei, indicating the influences of mixed pollution (from anthropogenic, BB, biogenic, and natural activities like dust and sea-salt) activity on the PM 2.5 chemical components.However, OC was found the most dominant over Mt.Bamboo and Mt.Lulin followed by anions, EC, and cations.EC fractions were found lesser than OC fractions in the PM 2.5 , providing other distinction of carbonaceous contents.

Ion Balance, Neutralization Factors, Ionic Ratios and Correlations in WSIIs
Among the WSIIs, anions were found dominant over the cations at all the study sites (Tables 2(a) and 2(b); Figs 3(a)-3(e)).Significant correlations in mass concentrations (Supplement Fig. S4) were estimated between sum of cations (Na + , NH 4 + , K + , Mg 2+ , and Ca 2+ ) and sum of anions (Cl -, NO 3 -, and SO 4 2- ) over Cape Fuguei (r = 0.99 in PM 2.5 and r = 0.91 in PM 2.5-10 ), Mt.Bamboo (r = 1.00 in PM 2.5 ), Mt.Lulin (r = 0.96 in PM 2.5 ), Cape Hedo (r = 0.89 in PM 2.5 ), and Kumamoto (r = 0.99 in PM 2.5 ), indicating good ion balance for samples.Moreover, in order to determine the quality of the analysis, we carried out an ion balance estimation using cations and anions with the assumption that most of the ions were in the solutions.The sum of total anions (µeq m -3 ) should be equal to the sum of total cations (µeq m -3 ) in the solution as per the electro neutrality principle and this ratio can be used as a fit indicator to study aerosol acidity over the sampling sites (Boreddy and Kawamura, 2015).The charge balance between cations and anions were estimated as following equations: Cation Equivalent 23 18 39 12 20 Similarly, significant correlations in charge balance (Supplement Fig. S5) were also found between sum of cations to anions (µeq m -3 ) over Cape Fuguei (r = 1.00 in PM 2.5 and r = 0.99 in PM 2.5-10 ), Mt.Bamboo (r = 1.00 in PM 2.5 ), Mt.Lulin (r = 0.98 in PM 2.5 ), Cape Hedo (r = 0.90 in PM 2.5 ), and Kumamoto (r = 1.00 in PM 2.5 ), indicating good quality of data-sets and the ions allocate a common source origin.The ∑ + /∑ -ratio is a good indicator to study the acidity of the environment.Average ∑ + /∑ -ratios > 1 over Cape Fuguei (1.12 ± 0.06 in PM 2.5 and 1.11 ± 0.04 in PM 2.5-10 ), Mt.Bamboo (1.10 ± 0.23 in PM 2.5 ), and Kumamoto (1.11 ± 0.05 in PM 2.5 ) indicated that the aerosol samples were evidently alkaline in nature and the anion deficits were possibly due to the CO 3 2-, HCO 3 -, and other organic anions (such as acetate and formate).But, the ∑ + /∑ -ratio < 1 over Mt.Lulin (0.85 ± 0.31 in PM 2.5 ) and Cape Hedo (0.93 ± 0.13 in PM 2.5 ) suggested the acidic nature of aerosols probably due to the lack of H + ions.We also estimated the ion difference by using the following equation The average ion difference over Cape Fuguei (5.77 ± 2.58% in PM 2.5 and 5.19 ± 1.95% in PM 2.5-10 ), Mt.Bamboo (3.96 ± 9.55% in PM 2.5 ), and Kumamoto (5.22 ± 2.22% in PM 2.5 ) were found well within acceptable range (WMO, 1994).
Neutralization factor (NF) is the acid neutralization capacities of different cations.Acidic nature of aerosols is mainly due to the anions like SO 4 2-and NO 3 -(formed by secondary oxidation of NO, NO 2 , and SO 2 ) and mainly neutralized by NH 4 + , K + , Mg 2+ , and Ca 2+ cations.The role of Cl -(in acid production) and Na + (in alkalinity) were negligible owing to their marine origins in the form of seasalt, which is neutral.The NF for different cations was estimated as by following equations (Keene et al., 1986): where the concentrations of all WSIIs are in μeq m -3 .The NF values of NH 4 + , K + , Mg 2+ , and Ca 2+ are presented in supplement Table S1.The orders in NF were found as NH 4 + > K + > Ca 2+ > Mg 2+ over Cape Fuguei in PM 2.5 , NH 4 + > Ca 2+ > K + > Mg 2+ over Cape Fuguei in PM 2.5-10 and over Mt.Bamboo, Mt.Lulin, and Kumamoto in PM 2.5 , and NH 4 + > Mg 2+ > K + > Ca 2+ over Cape Hedo in PM 2.5 .NH 4 + was found to be the dominant acid neutralizing agent over all the sites.However, it can be noted that nss-Mg 2+ had a negligible role on acid neutralizing except the Cape Hedo site.
It is worth to note that over Cape Fuguei nss-SO 4 2-was mainly distributed in PM 2.5 while lower values were observed in PM 2.5-10 .The larger variations in nss-SO 4 2-and NH 4 + fractions of the total PM 2.5 indicate that the aerosol collected at the Cape Fuguei site have contributions from anthropogenic sources from continental Asian outflows.The variation in NH 4 + was consistent with that in nss-SO 4 2-on the linear correlation coefficients in PM 2.5 (r = 0.98) and PM 2.5-10 (r = 0.79), indicating the association of both ions over Cape Fuguei.Though NO 3 -, NH 4 + , and nss-SO 4 2-were mainly from anthropogenic sources (Seinfeld and Pandis, 2006), NO 3 -exhibited moderate correlation with nss-SO 4 2-in PM 2.5 (r = 0.42) and higher correlation in PM 2.5-10 (r = 0.82).This finding implied that NO 3 -ions did not exist in fine particles but in coarse particles (Chuang et al., 2013a).Among the WSIIs, Na + and Cl -contributed moderately to the PM 2.5 while much higher in PM 2.5-10 , indicating that sea-salt aerosol appeared to be the main constituent of coarse particles.Similar result was reported over Dongsha Island (Chuang et al., 2013a) in northern SCS.Na + and Cl -exhibited consistent variations in PM 2.5 (r = 0.84) and PM 2.5-10 (r = 0.97) caused by the moderate effect of chlorine loss in PM 2.5 aerosols (as discussed in next sub-section).Mg 2+ and Ca 2+ were also found higher in coarse aerosols.The nss-Ca 2+ in PM 2.5-10 (0.33 ± 0.29%) was found too low and possibly came from the soil dust.The nss-K + and nss-Ca 2+ showed significant correlation (r = 0.91, Table 3(b)) in PM 2.5-10 and indicated the notable influence of soil dust.
However, similar findings in distributions of WSIIs were seen over Cxape Hedo and Kumamoto sites.Similar in findings to other locations, nss-SO 4 2-was also found the most abundant over Cape Hedo (25.44 ± 9.45%) and Kumamoto (18.90 ± 5.51%) followed by NH 4 + and NO 3 -(Table 2(b)).Larger variations of nss-SO 4 2-and NH 4 + over these two locations (Figs. 4(e) and 4(f)) revealed the contributions from anthropogenic emissions mainly from the continental Asian outflow.The variation in Na + and Cl -were found significant over Cape Hedo owing to the vicinity of the oceanic air masses.Strong correlations (r > 0.94, see Tables 3(e)-3(f)) between nss-SO 4 2-and NH 4 + revealed strong neutralization of acidic sulfate.The nss-SO 4 2-and nss-K + showed higher correlation over Kumamoto (r = 0.95) than Cape Hedo (r = 0.72) and suggested the significant formation of secondary aerosols over Kumamoto.Both Cape Hedo and Kumamoto were influenced by eastward drift of transboundary aerosol intrusions from the Asian continent into the NPR region and resulted in worse air quality in Japan (Moreno et al., 2013;Kunwar and Kawamura, 2014;Shimada et al., 2015;Zhu et al., 2015;Misawa et al., 2017).The cause findings over these two locations in this study were found corroborated with the previous results in literatures.

Chloride Deficit
The chloride deficit (Cl loss in %) decreases as the distance from the sea or the coast increases due to the depletion of Cl - (Park et al., 2016).Cl loss was estimated by using the following equation (Yao and Zhang, 2012).
where, [Na + ] and [Cl -] represents the mass concentrations of Na + and Cl -in µg m -3 , respectively.The mean Cl loss was found more substantial over Cape Fuguei i.e., 80.75 ± 23.53% (range: 39-99%) and 23.93 ± 10.99% (12-54%) for PM 2.5 and PM 2.5-10 , respectively.Relatively lower mean of Cl loss i.e., 74.88 ± 21.64% (24-98%) and 60.35 ± 15.84% (44-84%) were estimated over Cape Hedo and Kumamoto, respectively.The Cl loss was not found significant over Mt.Bamboo and Mt.Lulin sites and hence not discussed here.The Cl loss (%) estimated in this study was compared with marine aerosols sampled at coastal regions and open sea locations in East Asia and over the world (supplement Table S2).Fine mode aerosols showed higher Cl loss as compared to the coarse mode.Park et al. (2004) accounted for Cl loss (55% for fine mode particles) and (16% for coarsemode particles) during the normal periods over Jeju Island in southern Korea.The estimated Cl loss was up to 50% in spring over Okinawa Island (Kunwar and Kawamura, 2014).Cl loss was nearly 86-98% for PM 2.5 and 29-30% for PM 2.5-10 over the sea waters between Taiwan and Dongsha Island in SCS (Hsu et al., 2007).The Cl loss in PM 2.5 were 77.3 ± 9.9% (Asian Continent Trajectory; ACT); 77.9 ± 23.4% (Philippine Trajectory; PPT), and 59.9 ± 10.0% (West Pacific Trajectory; WPT) over Dongsha Island in northern SCS (Chuang et al., 2013a).Likely, the Cl loss in PM 2.5-10 were 51.4 ± 10.4% (ACT); 95.6 ± 5.3% (PPT), and 54.6 ± 24.5% (WPT) over Dongsha Island (Chuang et al., 2013a).The Cl loss in PM 2.5 at the offshore site and over sea in the Taiwan Strait ranged in 16.2-19.2% and 7.6-20.4%, respectively (Li et al., 2016).The higher Cl loss over East China Sea was too much lower than those on the SCS.Moreover, the East China Sea and the Taiwan Strait received abundant amount of acidic gases from Asian outflows (Streets et al., 2000;Carmichael et al., 2002;Li et al., 2016), whereas air masses transported toward the SCS were not only from South China but also from BB emitted particularly from the Indochina Peninsula (Arndt et al., 1997;Chuang et al., 2013a).In the Antarctic coastal site, Cl loss was observed only in the polluted summer but not in the clean winter (Jourdain and Legrand, 2002).Cl loss was observed only for the finer aerosols over the Arctic (Hara et al., 2002).Lesser values of Cl loss indicated that the strength of emissions of acid gases would not be the most critical controlling factor Cl loss of over East coast of US (Bermuda), NE Pacific Ocean, tropical northern Atlantic Ocean, northern Indian Ocean, and NW Mediterranean Sea (Keene and Savoie,  1998; Johansen et al., 1999, 2000; Sellegri et al., 2001;  Newberg et al., 2005).Contrarily, the tropical Arabian Sea showed higher Cl loss as compared to the abovementioned oceanic sites i.e., 89 ± 9% and 25.6 ± 21.3% for PM 2.5 and PM 2.5-10 (Johansen and Hoffmann, 2004).

Carbonaceous Contents
The distributions of eight temperature-resolved carbonaceous fractions of PM 2.5 collected over the study sites are shown in Fig. 5. OC was found the second most abundant resolved component in both PM 2.5 and PM 2.5-10 (11.94 ± 3.48% in PM 2.5 ; 2.66 ± 1.90% in PM 10-2.5 ), followed by the nss-SO 4 2-over Cape Fuguei (Table 2(a)).By contrast, EC exists mostly in PM 2.5 (1.77 ± 1.59%) as compared to PM 2.5-10 (0.19 ± 0.16%).However, OC comprised the dominant bulk resolved component in PM 2.5 (Table 2(a)) over Mt.Bamboo (33.24 ± 24.11%) and Mt.Lulin (37.58 ± 25.90%) followed by the water soluble anions (26.61 ± 8.39% for Mt.Bamboo and 28.27 ± 6.07% for Mt.Lulin) and EC (13.80 ± 13.56% for Mt.Bamboo and 11.26 ± 7.55% for Mt.Lulin).Excess concentration of OC was reported in winter in urban regions of China (Wang et al., 2011).Our study sites were receiving most of the air parcels from the northern China suggesting the additional contribution of OC from BB emissions as burning of crop residues (wheat straw and corn stalk) and fire-woods were significant in urban Beijing from late autumn to winter (Wang et al., 2009;Yu et al., 2013).Moreover, high concentrations of OM were observed over Beijing during winter time (Wang et al., 2006;Zhang et al., 2008).Among the 8 carbonaceous fractions analyzed in PM 2.5 (PM 2.5-10 ), the OC3 > OC2 > OC4 > OP > OC1 (OC3 > OC2 > OP > OC4 > OC1) order in OC fractions and EC1-OP > EC2 > EC3 (EC1-OP > Fig. 5. Distributions of carbonaceous fractions as weight % of total (a) PM 2.5 over Cape Fuguei, (b) PM 2.5-10 over Cape Fuguei, (c) PM 2.5 over Mt.Bamboo, and (d) PM 2.5 over Mt.Lulin.EC2 > EC3) order in EC fractions were observed over Cape Fuguei (Table 2(a) and Figs.5(a)-5(b)).The least amount of OC1 fraction may be due to the rapid evaporation of this fraction in BB smoke after emission into the atmosphere (Chuang et al., 2013b;Lee et al., 2016).It is well documented that OC3 is often used as tracer for BB (Chuang et al., 2013b;Lee et al., 2016).Similarly, nonvolatile EC1-OP (char-EC) is one of the major components in BB emissions (Chuang et al., 2013b;Lee et al., 2016) and used for source apportionment of BB (Cao et al., 2005;Han et al., 2007;Han et al., 2009).Both, OC3 and EC1-OP were found the most abundant fractions of OC and EC contents, respectively in both PM 2.5 and PM 2.5-10 , over Cape Fuguei and revealed the significant contribution of BB from continental Asian outflow.The highest EC1-OP in both PM 2.5 and PM 2.5-10 suggested the char dominated EC over this site from the BB emissions in continental Asian outflow.The sequence order of OCs showed the relatively higher contribution of OP in PM 2.5-10 than that of PM 2.5 .OP was also reported as the most abundant carbonaceous fraction in PM 2.5-10 over a near-source BB region at Chiangmai, northern Thailand during March-April, 2010 (Chuang et al., 2013b).Similarly, OP was also observed the most abundant carbonaceous fraction in both PM 10 and PM 2.5 , in a rural area during smoke haze events in northwestern China (Zhu et al., 2016).Likely, the orders of fractions in OC (EC) were OC3 > OC1 > OC2 > OC4 > OP (EC2 > EC1-OP > EC3) and OC3 > OC1 > OC2 > OC4 > OP (EC2 > EC1-OP > EC3) for Mt.Bamboo (Fig. 5(c)) and Mt.Lulin (Fig. 5(d)), respectively.EC3 was found almost negligible over all the sites.The highest OC3 fraction in OCs suggested the significant influence of BB emissions, while the highest EC2 fraction in ECs suggested the soot dominated EC from anthropogenic emissions over Mt.Bamboo and Mt.Lulin sites.Overall these two sites were found influenced by the both anthropogenic and BB from the Asian outflows during the IOP.
Table 5 summarizes the OC and EC levels observed at the Cape Fuguei, Mt.Bamboo, and Mt.Lulin with other background, rural, and high-elevation sites.The average concentrations of OC and EC at Cape Fuguei (OC: 2.81 ± 1.68 µg m -3 and EC: 0.42 ± 0.55 µg m -3 ) were found comparable with those at islands and back-ground/rural stations in Japan, Korea, China, and other parts of the world.OC and EC concentrations were ranged 0.89-6.07µg m -3 and 0.02-2.01µg m -3 at Cape Fuguei.These background EC concentrations were also found similar with 0.13-0.30µg m -3 of EC measured at Mt. Waliguan (Tang et al., 1999), 0.08-0.43µg m -3 of EC from the Gosan marine GAW station of Korea (Kim et al., 1999), 0.53 µg m -3 of EC from S. Pietro Capofiume in Europe (Decesari et al., 2001), and 0.8 µg m -3 of EC from the Galveston background site in north America (Fraser et al., 2002).The mass range of OC and EC were 1.6-5.2µg m -3 and 0.2-1.3µg m -3 , respectively over the rural and background locations (Table 5).
A higher value of OC/EC (32.4; estimated from the reported OC and EC values) was reported over Okinawa Island, Japan in spring, 2008 due to the long-range-transport BB from East Asia (Deshmukh et al., 2016) and well compared with our results.Moreover, the OC/EC ratios were 4.8 for BB aerosols in Phimai, Thailand (Li et al., 2013) and 5.7 in Chiangmai, Thailand (Chuang et al., 2013b), 6.8 in Sonla, Vietnam (Lee et al., 2016), 9.1 in Tanzania, East Africa (Mkomo et al., 2010), 18.9 in Rondonia, Brazil (Kundu et al., 2010b), 15.4 in Georgia, USA (Lee et al., 2005), and 17.8 in Wyoming, USA (Pratt et al., 2011).Table 5 also compares the OC/EC ratio over Cape Fuguei together with those from different back-ground and rural locations in East Asia and in the world.Chou et al. (2010) reported the annual range of 2.6-3.0 (OC/EC) over Cape Fuguei in a previous study.Shimada et al. (2016) reported a value of 3.0-4.3for OC/EC ratio over Okinawa, Japan.Kim et al. (2000a) estimated comparably higher value of 14.4  ± 3.5 for OC/EC ratio over a rural location in Gosan, Korea.OC/EC ratio was 10 in Gosan, Jeju Island in December 1996 (Lee et al., 2001), 11.9 in Shangri-La, southwest tip of China (Zhang et al., 2008), and 12.2 in Akdala, northwest tip of China (Zhang et al., 2008) over back-ground remote locations.OC/EC ratios in back-ground stations in Europe (Table 5) were comparably lower (2.8-6.8)than East Asia and attributed mainly to anthropogenic emissions.The mean OC/EC ratio was 3.01 ± 1.37 (range: 1.15-5.16)and 3.56 ± 1.20 (range: 2.32-7.06)over Mt.Bamboo and Mt.Lulin, respectively during the study period.OC/EC ratio was 4.8 ± 1.5 (BB group), 3.4 ± 2.0 (SNBB group), 3.8 ± 1.5 (anthropogenic group) and 4.3 ± 1.9 (FT group), during April 2003-April 2012 over Mt.Lulin (Chuang et al., 2014).The OC/EC ratios over mountains in the present study were normally lower than most OC/EC ratios observed at other high-altitude mountain sites, as shown in Table 5. Lower OC/EC values of 3.6 in Jungfraujoch, Switzerland (Krivacsy et al., 2001), 2.6 in Tengchong, China (Engling et al., 2011), 3.1 in Puy de Dome, France (Gelencser et al., 2007), 4.9 in Southwestern Germany (Gelencser et al., 2007), and 4.3 in Qinghai Lake (Li et al., 2013a) over high altitude stations were influenced by free tropospheric air masses of different origins with respect to seasonal variations.Higher OC/EC ratios obtained in the present study than those reported over the high altitude stations i.e., Namco (31.9 ± 31.1),Muztagh Ata (2.9-32.1),Lulang (1.7-58.4)mountain (Table 5) may be attributed to the very low EC level.
The ratio of char-EC to soot-EC in PM 2.5 is more effective indicator for BB contribution of carbonaceous aerosols than the previously used OC/EC ratio because the later can be affected by SOC formation (Han et al., 2007(Han et al., , 2009)).Char-EC is formed during relatively low combustion temperatures from BB activity, whereas soot-EC is formed at high combustion temperatures from coal combustion or from vehicle exhaust (Zhu et al., 2010).The char-EC/soot-EC ratios can be as high as 22.6 for BB activity and as low as 0.6 for vehicle exhaust (Chow et al., 2004).The char-EC/soot-EC ratios were 11.6 and 1.9 for BB and coal combustion, respectively, in Xi'an City, China (Cao et al., 2005).(Lee et al., 2016) and suggested relatively more influence of BB emissions.Both the sites, Mt.Bamboo and Mt.Lulin, showed relatively lower values of char-EC/soot-EC ratio may be due to the presence of mixed type aerosols (both anthropogenic and BB) from the air masses coming from urban China.It is worth to note that the mean concentration (in µg m -3 ) of char-EC (Mt.Bamboo: 0.24 ± 0.29; Mt.Lulin: 0.12 ± 0.07) was more or less same with soot-EC (Mt.Bamboo: 0.27 ± 0.09; Mt.Lulin: 0.14 ± 0.08).This ratio was 3.9 ± 3.5 (BB group), 0.4 ± 0.4 (SNBB group), 0.9 ± 0.9 (anthropogenic group) and 0.3 ± 0.4 (FT group), during April 2003-April 2012 over Mt.Lulin (Chuang et al., 2014).

Long-Range Transported Haze from Continental Asian Outflow
We further investigated the relationship among WSIIs and carbonaceous species to verify the influence of anthropogenic and BB emissions from Asian outflow.NO 3 -is a tracer of anthropogenic (Kundu et al., 2010a) and BB (Lee et al., 2016), mainly derived from coal combustions, BB, and vehicular emissions (Kundu et al., 2010a).EC is a tracer of incomplete combustion of fossil fuels and BB (Kundu et al., 2010a).Positive correlation between NO 3 -and nss-K + over Cape Fuguei (r = 0.4 in PM 2.5 and r = 0.96 in PM 2.5-10 ), Mt.Bamboo (r = 0.70 in PM 2.5 ), and Mt.Lulin (r = 0.33 in PM 2.5 ) suggested the association of NO 3 -with BB and anthropogenic emissions from continental Asian outflow.Kunwar and Kawamura (2014) also reported a positive correlation between NO 3 -and nss-K + over Okinawa (r = 0.65) in winter and suggested the influence of BB emissions.
From the above analysis and discussions, we found the possible and significant influence of continental Asian outflow on the ambient aerosols in East Asia.However, in order to quantify the possible contribution from the continental Asian outflow, we analyzed the PM 2.5 mass concentration and surface chemistry characterization over the abovementioned sites (Cape Fuguei, Mt.Bamboo, and Mt.Lulin) under different BT pathways (as presented in Table 1 and Figs.2(a)-2(c)).Higher concentrations of PM 2.5 , WSIIs, OC, and EC were observed for the representative pathway of Asian outflow over Cape Fuguei (Path C + Path F), Mt.Bamboo (Path F), and Mt.Lulin (Path F) during the IOP (supplement Table S3).Fig. 6 shows the relative mass fraction (%) of PM 2.5 and its various components for different BT pathways during the IOP.Moreover, higher concentrations of BB tracers (i.e., NO 3 -, OC3, and EC1-OP) were also observed when these sites were under the influence of continental Asian outflow (supplement Table S3).The order of relative mass fraction (%) for different BT pathways in ).In addition to above, the OC/EC and char-EC/soot-EC ratios were also found the highest for the Asian outflow path over all the sites.These results suggested that the continental Asian outflow significantly influence the overall PM 2.5 mass concentrations and its BB components over the islands and mountains in East Asia.
We also selected some episodic days as representative of BB emissions from continental Asian outflow over all the above mentioned sites during the IOP.PM 2.5 aerosol mass was observed the highest over Cape Fuguei (37 µg m -3 ) on 29 October, when it was the second highest over Mt.Bamboo (34 µg m -3 ; the highest was 38 µg m -3 on 8 November) and Mt.Lulin (8 µg m -3 ; the highest was 9 µg m -3 on 28 October).The PM 2.5 mass was found lesser with distance from the Asian continent and sea-level altitude.Fig. 7 shows the PM 2.5 chemical species distributions over three aforementioned locations on 29 October.Among the three sites, SO 4 2-, K + , OC, and EC fractions were observed the highest over Cape Fuguei and indicated the higher influence of both anthropogenic and BB emissions.The highest PM 2.5 mass over Cape Fuguei was associated with the highest mass of SO 4 2-(12.51µg m -3 and 34.04% in PM 2.5 ), OC (6.07 µg m -3 and 16.52% in PM 2.5 ), and EC (2.01 µg m -3 and 5.47% in PM 2.5 ).It is also worth to note that, on that particular day, NO 3 -(1.62µg m -3 and 4.40% in PM 2.5 ), OC3 (2.31 µg m -3 and 6.30% in PM 2.5 ), EC1-OP (1.94 µg m -3 and 5.28% in PM 2.5 ), and char-EC/soot-EC ratio (27.54) were also found the highest (for both mass concentrations and % contributions in PM 2.5 ) over Cape Fuguei with respect to other days.These results suggested 29 October as a representative day of both anthropogenic and BB influence over Cape Fuguei during the IOP.Similarly, the highest mass concentrations of PM 2.5 (38 µg m -3 ), OC (6.36 µg m -3 ), EC (1.26 µg m -3 ) and all the BB tracers (NO 3 -: 3.67 µg m -3 ; nss-K + : 0.07 µg m -3 ; OC3: 1.63 µg m -3 ; EC1-OP: 0.92 µg m -3 ) were recorded over Mt.Bamboo on 8 November, indicating a representative BB influenced day.Similarly, the highest concentration of all the BB tracers i.e., NO 3 -(0.24µg m -3 ), OC3 (0.43 µg m -3 ), and EC1-OP (0.25 µg m -3 ) were observed on 10 November over Mt.Lulin.This observation suggested the influence of long-range transported BB emissions from continental Asian outflow to the background and free-tropospheric aerosols over Mt.Lulin.

Reconstruction of PM 2.5 Mass
We reconstructed the surface PM 2.5 mass concentrations by using the IMPROVE algorithm (Hand et al., 2011) and Fenn, 1979), and [Soil] = 1.63 × [Ca] (Patterson, 1981).In this study, for soil estimation we used [Ca 2+ ] in place of [Ca] as per the availability with an assumption that both were from the crustal influences and other metals (Al, Si, Ti, and Fe) were not used because of their unavailability of measurements.The observed and reconstructed PM 2.5 mass exhibited satisfactory correlations (Figs.8(a)-8(c)) over Cape Fuguei (r = 0.98) and Mt.Bamboo (r = 0.99) whereas little poor over Mt.Lulin (r = 0.89).The unavailability of other crustal metals may be introduced some uncertainty into the calculation of the reconstructed PM 2.5 mass.
Relative contribution (%) of different components in PM 2.5 reconstructed mass (Figs.

CONCLUSIONS
This study is a component of an IOP carried out during 25 October-10 November, 2015 over various sites in East Asia.Ambient aerosols were collected over islands and mountains sites (i.e., Cape Fuguei, Mt.Bamboo, Mt.Lulin, Cape Hedo, and Kumamoto) in East Asia and analyzed for mass concentrations, WSIIs and carbonaceous fractions on the basis of the US-IMPROVE protocol and correction method.The main goal of this study was to characterize the wintertime surface aerosol chemistry over remote islands and mountains in East Asia.
The highest PM 2.5 aerosol mass was found over Kumamoto (22 ± 7 µg m -3 ) and followed by Cape Fuguei (20 ± 9 µg m -3 ), Cape Hedo (11 ± 5 µg m -3 ), Mt.Bamboo (10 ± 13 µg m -3 ), and Mt.Lulin (4 ± 3 µg m -3 ).The PM 2.5 collected in this study showed that WSIIs accounted for ~65% of the total aerosol mass over Cape Fuguei with ~14% of carbon fractions (the sum of OC and EC).On the contrary, carbon fractions were the higher contribution (~48%) over Mt.Bamboo and Mt.Lulin as compared to WSIIs (38%).WSIIs accounted for ~40% in PM 2.5 over Cape Hedo (42.55 ± 9.09%) and Kumamoto (37.17 ± 7.59%).Chemical analysis showed that nss-SO 4 2-was the most abundant species followed by NH 4 + , Na + , Cl -, NO 3 -, and nss-K + .Significant correlations (r > 0.9) in charge balance of ions suggested the good quality of data-sets and the ions shares common source origins.The study showed more influence of (NH 4 ) 2 SO 4 and NH 4 HSO 4 than NH 4 NO 3 .Mass ratio of [NO 3 -]/[SO 4 2-] in PM 2.5 were found < 1, indicating the dominance of stationary sources (i.e., coal combustion and BB) during the IOP.The nss-SO 4 2and NH 4 + exhibited wider ranges, thereby indicating the influence of anthropogenic sources over all the study sites.Bulk dominant of nss-SO 4 2-in SO 4 2-and nss-K + in K + in PM 2.5 over all the sites, suggested the significant influence of long-range transported anthropogenic and BB in East Asia.Higher correlations (r ≥ 0.88) between nss-SO 4 2-and nss-K + might be attributed to formation of secondary aerosols.
For carbonaceous contents, the OC concentration was much higher than that of EC in the collected aerosols and among OCs; the OC3 was the most abundant over all the sites (Cape Fuguei, Mt.Bamboo, and Mt.Lulin), indicating the significant influence of BB emissions too.EC1-OP was found dominant in EC over Cape Fuguei.Significant correlations (r > 0.74) between OC and EC suggested common emission sources and transport processes.Higher OC/EC and char-EC/soot-EC ratio over Cape Fuguei implies that most carbonaceous components were contributed directly from the similar BB emissions.The analyses of BT pathways and fire-counts demonstrated that the temporal variations in PM 2.5 and its constituents (WSIIs and carbon fractions) were caused by long-range transported anthropogenic and BB from the Asian continental outflow.Higher concentrations of PM 2.5 , WSIIs, OC, and EC along with the BB tracers (i.e., NO 3 -, OC3, and EC1-OP) were observed during the influence of continental Asian outflow.This study demonstrates that both anthropogenic and BB emissions from continental Asian outflow plays an important role in regulating chemical properties of wintertime aerosols over remote islands and mountains in East Asia.

Fig. 2 .
Fig. 2. Possible pathways based on 7-day BTs over (a) Cape Fuguei, (b) Mt.Bamboo, and (c) Mt.Lulin in East Asia.The blue dot and its size denote the annual CO emission based onStreets et al. (2003).Also shown are the gridded fire-counts during October-November, 2015 from the MODIS Terra satellite (http://disc.sci.gsfc.nasa.gov/neespi/data-holdings/mod14cm1.shtml).The paths with occurrence % of BT ending at each location are denoted, as described in Table1.For the interpretation of references to the color for BT paths in Figs.2(a)-2(c), the reader can refer to the web version of this article.

Feb
NO 3 -was found to be as Path C + Path F > Path A > Path B (Cape Fuguei), Path F > Path A > Path G > Path D (Mt.Bamboo), and Path F > Path A > Path B > Path G > Path E (Mt.Lulin).Likely, the sequence of relative mass fraction (%) for different BT pathways in OC3 was found to be as Path C + Path F > Path B > Path A (Cape Fuguei), Path F > Path D > Path A > Path G (Mt. Bamboo), and Path F > Path G > Path B > Path A > Path E (Mt.Lulin).Additionally, the sequence of relative mass fraction (%) for different BT pathways in EC1-OP was found to be as Path F > Path B > Path A (Cape Fuguei), Path F > Path A > Path D > Path G (Mt. Bamboo), and Path F > Path G > Path A > Path B > Path E (Mt.Lulin

Fig. 6 .
Fig. 6.Relative mass fraction (%) of different BT pathways in PM 2.5 and its components over Cape Fuguei (top-panel), Mt.Bamboo (mid-panel), and Mt.Lulin (bottom-panel) during the IOP.Similar colors for the BT paths are used here as Figs.2(a)-2(c).For the interpretation of references to the colors in this figure legend, the reader can refer to the web version of this article.

Table 2 (
a). Measurements of PM 2.5 , WSIIs, carbonaceous contents, and ratios of different components as mass concentrations (µg m -3) and weight percentages (%) of the total mass collected over Cape Fuguei, Mt.Bamboo, and Mt.Lulin.

Table 2 (
b). Same as Table 2(a), but for PM 2.5 and WSIIs collected over Cape Hedo and Kumamoto.

Table 4 .
Comparison of WSIIs mass concentrations from this study with other islands and high-altitude Mountain sites in East Asia and worldwide.

Table 5 .
Comparison of OC and EC mass concentrations and OC/EC ratio from this study with other rural/back-ground sites and high-altitude Mountain sites in East Asia and worldwide.