Molecular characterization of organic aerosols in the Kathmandu Valley, Nepal: insights into primary and secondary sources

Abstract. Organic atmospheric aerosols in the Hindu
Kush–Himalayas–Tibetan Plateau region are still poorly characterized. To
better understand the chemical characteristics and sources of organic
aerosols in the foothill region of the central Himalaya, the atmospheric
aerosol samples were collected in Bode, a suburban site of the Kathmandu
Valley (KV) over a 1-year period from April 2013 to April 2014. Various
molecular tracers from specific sources of primary organic aerosols (POAs)
and secondary organic aerosols (SOAs) were determined. Tracer-based
estimation methods were employed to apportion contributions from each source.
The concentrations of organic carbon (OC) and elemental carbon (EC) increased
during winter with a maximum monthly average in January. Levoglucosan (a
molecular tracer for biomass burning, BB) was observed as the dominant
species among all the analyzed organic tracers and its annual average
concentration was 788±685 ng m−3 (ranging from 58.8 to
3079 ng m−3). Isoprene-SOA (I-SOA) represented a high concentration
among biogenic-SOA tracers. For the seasonality, anhydrosugars, phenolic
compounds, resin acid, and aromatic SOA tracer showed similar seasonal variations with OC and EC while
monosaccharides, sugar alcohols, and I-SOA tracers showed lower levels during
winter. BB contributed a significant fraction to OC, averaging 24.9 %±10.4 % during the whole year, and up to 36.3 %±10.4 % in
the post-monsoon season. On an annual average basis, anthropogenic
toluene-derived secondary OC accounted for 8.8 % and biogenic secondary
OC contributed 6.2 % to total OC. The annual contribution of fungal
spores to OC was 3.2 % with a maximum during the monsoon season
(5.9 %). For plant debris, it accounted for 1.4 % of OC during the
monsoon. Therefore, OC is mainly associated with BB and other anthropogenic
activity in the KV. Our findings are conducive to designing effective
measures to mitigate the heavy air pollution and its impacts in the KV and
surrounding area.



Introduction
South Asia, especially the Indo-Gangetic Plain (IGP) region, is a global air pollution hotspot.
Atmospheric pollutants [e.g., organic carbon (OC), black carbon (BC), gaseous pollutants, etc.] from South Asia have been increasing during recent decades (Ramanathan et al., 2005;Muzzini and Aparicio, 2013;Lawrence and Lelieveld, 2010). While these pollutants are of concern locally near the emission 45 sources, they can also, in short span of time, be transported to rural and remote regions over a long distance.
This results in an annually recurring regional scale haze, referred to as atmospheric brown clouds (ABC), covers a large area from the Himalayan range to the Indian Ocean (Ramanathan et al., 2007). Until recently the emissions, types, levels, atmospheric transport and transformation, impacts and mitigation of various atmospheric pollutants were not well characterized in the vast mountain areas and the foothill 50 region in South Asia. In this context, the international project of "A Sustainable Atmosphere for the Kathmandu Valley (SusKat)" was launched, aiming to comprehensively understand the causes of the severe air pollution in the region, and identifying appropriate solutions to reduce its impacts . This paper presents analyses of samples collected as part of the SusKat field campaign.
The Kathmandu Valley (KV), the capital region of Nepal, is considered one of the most polluted 55 regions over South Asia and the largest metropolitan region in the foothills of the Hindu Kush-Himalayas-Tibetan Plateau (HKHTP) region, facing rapid but unplanned urbanization, with the current population of approximately 4 million (Muzzini and Aparicio, 2013). Additionally, the bowl-shaped topography restrict the free flow of air, resulting in poor air quality (Pudasainee et al., 2006;Panday and Prinn, 2009). Giri et al. (2007) showed PM10 concentrations in Kathmandu were about 2-4 times higher than the 60 guidelines prescribed by the World Health Organization (WHO) (PM10 24-hour mean: 50 µg m -3 ) (WHO, 2006). More recently, Shakya et al. (2017) reported that daily mean PM2.5 concentrations at seven locations in the KV during 2014 were about 5 times higher than the WHO guidelines (PM2.5 24-hour mean: 25 µg m -3 ) (WHO, 2006). Beside particulate matter, recent studies have pointed out that groundlevel ozone (O3) is also of concern (Mahata et al., 2017b;Bhardwaj et al., 2017). Ozone levels at Pakanajol 65 in the city center exceeded the WHO's 8-hour maximum O3 guidelines of 100 µg m -3 on 125 days in a year (Putero et al., 2015), while Mahata et al. (2017b) reported such exceedance was for nearly 3 months at Bode (where sampling for this study was conducted) and 6 months at Nagarkot, a hilltop site downwind The KV is a round flat basin with the bottom of an elevation of approximately 1300 meters above sea level (a.s.l.) in the southern foothills of the Himalayas. It is encircled by green mountains (elevation: 1500 m to 2800 m a.s.l.) (Panday and Prinn, 2009). Our sampling was performed during April 2013 to April 2014 in Bode (27.67°N, 85.38°E, 1320 m a.s.l.), a suburban site to the east of Kathmandu city in the valley (Fig. 1). There are two major wind flows in the KV: (a) west to east, from Nagdhunga-100 Bhimdhunga mountain pass in the west to Nagarkot-Sanga mountain pass in the east, (b) south to north, from Bagmati River corridor to the northeast direction through the central-eastern part of the Valley.
These two airflows meet around central-eastern part of the Valley and move eastward towards the Nagarkot-Sanga passes (Panday and Prinn, 2009). The Bode area receives these two air flows, and hence it is downwind of Kathmandu city and Lalitpur or Patan city located in southwest, west and northwest 105 direction during the daytime, and Bhaktapur city located in east and southeast during nighttime Mahata et al., 2017a;Rupakheti et al., 2018). In addition, it is situated in a residential area with urban buildings and houses scattered across agricultural fields with paddy, wheat, corn and vegetable farms. Some small industries (plastics, electronics, wood, fabrics, etc.) and Bhaktapur Industrial Estate are located in the south-eastern direction of Bode, as well as several brick kilns that use low quality coal 110 during January to April (Sarkar et al., 2016). The Tribhuvan international airport in the west of Bode (∼ 4 km) may have potential impacts when there is westerly wind. Approximately 1.5 and 7 km to the north, there are two reserve forests, consisting of a mix mainly broad-leaved deciduous trees and evergreen conifer trees (Department of Plant Resources, 2015). BC and O3 measurements in the two major SusKat-ABC sites (Paknajol and Bode) in the Valley show similar levels (Putero et al., 2015;Mahata et al., 2017b). 115 Therefore, Bode site can be taken as a representative site for the KV .

Sample collection
The total suspended particulates (TSP) were continuously collected for 23 h (day and night time) every five days by a medium-volume sampler (model: KC-120H, Laoshan Co., China), which was installed on the rooftop of a building, approximately 20 m above ground. The flow rate was100 L min -1 .

120
Overall, 82 aerosol samples were successfully obtained using 90 mm diameter quartz filters (Whatman PLC, UK). The filters were pre-baked (550 °C, 6 h) to remove all organic material and weighed by a microbalance (sensitivity: ±0.01 mg) before and after sampling. Before each weighing, they were equilibrated at constant temperature (25±3 °C) and humidity (30±5%) conditions for 24 h. Finally, the filters were preserved at 20 °C below zero until laboratory analysis. Field blanks (one blank filter each 125 month) were also collected, briefly putting a filter onto the instrument without drawing air to assess potential contamination. There may be positive and negative artifacts during the sample handling/conditioning due to the adsorption/evaporation processes of organic aerosols Boreddy et al., 2017;Oanh et al., 2016). In a comparable study, Ding et al. (2013) reported the positive artifacts for OC and organic tracers were 10−20% and up to 16%, respectively.

Chemical Analysis
The aerosol samples were analyzed for major ions, OC, EC, and organic molecular tracers in the laboratory. Major ions (Ca 2+ , Na + , K + , Mg 2+ , NH4 + , Cl -, SO4 2and NO3 -) were measured using an ion chromatography (Dionex, USA) with ICS-320 and ICS-1500 . The detection limit (LOD) of all the major ions was 0.01 µg m -3 . They denoted less than 5% of the real sample concentrations 135 in the field blank filters . Non-sea-salt Ca 2+ (nss-Ca 2+ ) and K + (nss-K + ) was estimated according to the method from George et al. (2008). OC and EC were determined by a thermal/optical reflectance analyzer (Model 2001A, USA) (Wan et al., 2015). The field blank OC (0.59±0.13 μg m -3 ) were subtracted from the filter samples. EC in the field blank sample was 0.00 μg m -3 .

140
Detailed analytical method of organic molecular tracers was described previously by Wan et al. (2017). A trace gas chromatography coupled to a PolarisQ mass spectrometry detector (GC-MS, Thermo Scientific) was used for analysis. Briefly, small filter portions (1.13-3.39 cm 2 ) were cut, spiked with appropriate amounts of methyl-β-D-xylanopyranoside (MXP, 99%, Sigma) and D3-malic acid (DMA, CDN isotopes, 99%) as internal recovery standards. Each filter portion was then extracted three times 145 with a mixture of 20-ml dichloromethane/methanol (2:1, v/v) at room temperature for 30 minutes. The solvent extracts in total of 60 ml were combined and successively filtered through quartz wool, concentrated, dried over ultrapure nitrogen gas and then reacted with 50 μl of 99% N,O-bis-(trimethylsilyl)trifluoroacetamide (BSTFA, with 1% trimethylsilyl chloride) and pyridine (v/v=2:1) at 70 °C for 3 h. n-Hexane of 150 μl was added after derivatization. A TG-5MS (30 m × 0.25 mm × 0.25 150 μm) was used for separation according to the GC temperature program. The oven temperature was initially held at 50 °C for 2 min, increased to 120 °C at 30 °C min -1 , then to 300 °C at 6 °C min -1 and finally held for 16 min. The MS was operated in electron ionization (EI) mode at 70 eV with a scan range of 50-650 Da.

Estimation of measurement uncertainty
The application of surrogate standards for the quantification of most SOA tracers excluding PNA 175 and PA) could cause additional errors to the measurements. Error in analyte measurement (EA) is propagated from the standard deviation of the field blank (EFB), error in spike recovery (ER) and error from surrogate quantification (EQ): = √ 2 + 2 + 2 EFB was 0 in this study due to the undetectable SOA tracers in the field blanks. To estimate the ER 180 of tracers, the spike recoveries of surrogate standards within the range of 9.2% (erythritol) to 26.1% (PNA) were used. EQ was estimated by an empirical approach according to Stone et al. (2012). The relative error introduced by each carbon atom (En), oxygenated functional group (Ef) and alkenes (Ed) was estimated to be 15 %, 10% and 60%, respectively. Therefore, the EQ are calculated as: where Δn, Δf and Δd are the difference between a surrogate and an analyte of carbon atom number, oxygen-containing functional group and alkene functionality, respectively.
Currently, EQ was calculated in the range of 15% (2-MTLs) to 120% (ß-CA) and the estimation of EA ranged from17.6% to 122.4%. The estimated uncertainties for the measurement of the SOA tracers is presented in Table S2.

Meteorological parameters
The meteorological parameters (e.g., temperature (T), relative humidity (RH), etc.) used in this study were derived from Tribhuvan International Airport (www.wunderground.com), which was located west of Bode (approximately 4 km). Mixing layer height (MLH) data was measured with a Vaisala ceilometer at Bode site . The meteorology of KV and its surrounding regions is controlled by the 195 South Asian monsoon circulations in the wet season (monsoon, June-September). Westerlies dominate the atmospheric circulation patterns during the dry seasons including pre-monsoon (March-May), postmonsoon (October-November) and winter (December-February) with limited precipitation (Pudasainee et al., 2006;Mues et al., 2017). Additionally, it is also influenced by local mountain valley circulation .

Results and discussion
A statistical concentration summary of major ions, OC, EC, and organic tracers identified in TSP samples collected at the Bode site is presented in Table 1. Tracers for six classes of organic compounds were detected: anhydrosugars, monosaccharides, sugar alcohols, phenolic compounds and resin acid, phthalic acid esters, and secondary organic aerosol (SOA) tracers.

Aerosol loadings
The TSP samples at Bode site exhibited daily mass concentrations from 32.0 to 723 µg m -3 (255±167 µg m -3 ) during April 2013 to April 2014 (Table 1). Putero et al. (2015) reported 195±83 µg m -3 of online PM2.5 concentration in Pakajol site (also one of SusKat-ABC sites), accounting for roughly 80% of TSP in our study. The TSP concentrations were comparable to those reported over other heavily polluted 210 regions in South Asia, including Islamabad in Pakistan (Shah and Shaheen, 2008), Kolkata (Gupta et al., 2007) and Agra (Rajput and Lakhani, 2010) in India. Compared to the remote sites such as Lulang in the Tibetan Plateau  and Manora Peak in the central Himalaya (Ram et al., 2010), the TSP in Bode shows significantly higher mass concentrations. We found a clear seasonal variation in TSP mass concentrations (Fig. 2a), higher in pre-monsoon season (381 ± 366 μg m -3 ) and winter (353 ± 348 μg m -

215
± 213 μg m -3 ). It generally corresponded to the build-up of the atmospheric brown clouds (ABCs), which engulfed most of South Asia and the Northern Indian Ocean extending from November to May (Ramanathan et al., 2005).
Meteorological parameters may also affect the TSP concentrations. The highest TSP concentration 220 observed during the pre-monsoon can be caused by the fugitive dust which is been blown up by strong wind and the absence of wet-precipitation (Fig S1a and c). The lower TSP concentration in the monsoon was likely related to increased precipitation (Fig S1c) after the onset of the South Asian monsoon. During this season, nearly 80% of the annual precipitation falls in the KV, which flushes out pollutants from the atmosphere Wester et al., 2019). During winter, an inversion layer often occurs in 225 the KV owing to its bowl-shaped topography (Pudasainee et al., 2006). The existence of an inversion layer with the lower temperature (12.0±2.41°C), wind speed (2.86±1.34 km h -1 ), and MLH (0.34±0.08 km)   (Fig S1a,c and d) altogether reduced the pollution dispersion mechanism resulting in increased levels of pollutants close to the ground surface.  (Table 2), which hinted that dust may co-exist with SO4 2-, NO3and NH4 + in the KV .

Major ions and OC/EC
Carbonaceous aerosols (OC: 38.7±32.7 μg m -3 and EC: 9.92±5.33 μg m -3 ) accounted for 19.2%± 5.48% of TSP mass through the sampling period at the Bode site, which was higher than that of the major ions. OC alone accounted for 14.6%±4.81% of the TSP mass. During winter and pre-monsoon seasons,

255
OC and EC showed much higher concentrations than that during wet season (Fig. 2b and c). In this study, we found that the daily OC to EC mass ratios (OC/EC) varied from 0.77 to 15.8 (annual mean: 3.78±2.73) and seasonal mean ratios of 4.44, 2.71, 3.31, and 5.86 during pre-monsoon, monsoon, post-monsoon, and winter seasons, respectively (Table 1 and Fig. 2d). The OC/EC ratios of more than 2.0 indicates the BB aerosols or the formation of secondary organic matter (Cao et al., 2007). Their influence and contribution will be discussed in the following sections. The OC/EC ratios found in this study for the KV were similar to other sites in South Asia, like Lumbini (5.16±2.09, 2.41-10.03) (Wan et al., 2017), Delhi (5.86±0.99, 2.9-9.2) (Bisht et al., 2015) and Lahore (3.9 ± 1.6,1.5-7.2) (Alam et al., 2014).

265
Anhydrosugars of levoglucosan (1,6-anhydro-β-D-glucopyranose) and its two isomers (mannosan and galactosan) have been used as ideal molecular tracers for BB emissions (Simoneit, 2002;. They are exclusively emitted from the combustion and pyrolysis of cellulose and hemicelluloses. In the current study, the annual average concentration of levoglucosan was 788±685 ng m -3 , ranging from 58.8 to 3079 ng m -3 , which was the dominant species of the total identified tracer 270 compounds (Table 1).
For the seasonality, levoglucosan showed significantly higher levels during winter, pre-monsoon and post-monsoon seasons ( (Fig. 4). This indicates that OC and EC in KV's aerosols are strongly related to BB source (Kim et al., 2015).
The ratio of levoglucosan to mannosan (Lev/Man) has been applied to distinguish the possible categories of biomass burnt. Previously, higher Lev/Man ratios were reported for emissions from combustion of hardwood (ranging from 12.9 to 35.4 with an average of 21.5±8.3) and agricultural 285 residues (range from 12.7 to 55.7 with an average of 32.6±19.1) (Sang et al., 2013;Bhattarai et al., 2019).
For the softwood burning, the average ratio was 4.0±1.0 (ranging from 2.5 to 5.8). In current study, the annual mean ratio of Lev/Man was 16.3±5.96 ranging from 9.13 to 33.1 with only 9 samples <10. It can be inferred that the combustion of crop residues and hardwood is likely to be one of the major sources of atmospheric pollution in this region. A previous study also reported that the combustion of wood fuel for 290 cooking and heating is common during wintertime in Nepal while there is much more crop residue combustion during both pre-and post-monsoon seasons (Stockwell et al., 2016). This is not only a local but also a regional phenomenon; for example, Bhardwaj et al. (2017) and Wan et al. (2017) pointed out emissions from crop residue burning during the pre-and post-monsoon periods from western India and eastern Pakistan impact the air quality in Nepal. Similarly,  also showed that the 295 combustion of agricultural residues and forest fires over the northwestern IGP region are causes of the air pollution episodes over the foothills of the central Himalayas. In addition, brick kilns mainly operated during January-April, burned substantial quantities of low-grade coal, mixed crop wastes and firewood (Kim et al., 2015;Wester et al., 2019). Such emissions may also lead to the high levels of levoglucosan observed at Bode. We must point out that the incense burning in KV may also influence the levoglucosan 300 concentration.

Monosaccharides
Primary biological aerosol particle (PBAP) tracers, commonly known also as bioaerosols, were analyzed in the Bode aerosol samples, including five monosaccharides of glucose, fructose, trehalose, sucrose and xylose. PBAP is derived from fungal spores, vegetative debris, pollen, bacteria and viruses.
In current work, total monosaccharides had an annual mean concentration of 298±127 ng m -3 .
Glucose was the predominant species among monosaccharides (124±60.0 ng m -3 ), followed by fructose  (Table 1). Except xylose, they all presented higher concentrations in pre-monsoon period while being lower in winter (Fig 3h, i, j and k). There were significant linear correlations between glucose and fructose (R 2 = 0.77, p<0.001), trehalose and glucose (R 2 = 0.30, p<0.001), trehalose and fructose (R 2 = 0.23, p<0.001), sucrose and glucose (R 2 = 0.55, p<0.001), sucrose and fructose (R 2 = 0.55, p<0.001), and sucrose and trehalose (R 2 = 0.28, p<0.001) (Table 3). Therefore, the strong correlations indicated that they were derived from common sources, e.g. from local forests in the KV during the period of high productivity of plants. In addition, the pollen produced from the flowering of local vegetation also largely contribute to glucose, fructose, trehalose, sucrose. The flowering of trees and crops peaks during the premonsoon season. The similar phenomenon was also reported in deciduous forests in northern Japan (Miyazaki et al., 2012).

Sugar alcohols 330
Total concentration of sugar alcohols (arabitol, sorbitol, erythtitol and mannitol) was 213±126 ng m -3 , and thus lower than that of total monosaccharides (  Fig. 3m, n, o and p). They also showed significant correlations with each other, implying their common sources (Zhu et al., 2015).
Mannitol and arabitol have been mostly associated with fungal spores, along with vegetation and mature leaves and algae (Yttri et al., 2007;Myriokefalitakis et al., 2017). Recent studies proposed elevated concentrations of mannitol and arabitol were usually observed augmentation after rain events and also 340 highly correlated with relative humidity (Yue et al., 2016;Zhu et al., 2016). Therefore, at Bode, sugar alcohols were likely emitted by plants in nearby forest and agriculture fields, especially during the monsoon with the higher relative humidity (Fig. S1b). In addition, the higher temperatures (Fig. S1a) were conducive for more active microbial activities. Notably, the levels of PBAP discussed above were much higher than other sites in the world (Zhu et al., 2015;Liang et al., 2016;Yttri et al., 2007), indicating 345 the strong fungal spore production in the KV during the wet season.

Phenolic compounds and resin acid
Phenolic compounds (e.g., vanillic, syringic and p-hydroxybenzoic acids) derived from lignin pyrolysis and resin acid (e.g., dehydroabietic acid) from burning of conifer plants can be also used as biomarkers for BB. Syringic acid is prevalent in hardwood smoke while vanillic acid is dominant both in  Besides the information revealed by anhydrosugars discussed in section 3.3.1, lignin and resin biomarkers further confirmed that BB emissions play a significant role in contributing to organic aerosols 375 in the KV, particularly during winter and pre-monsoon periods.

Phthalic acid esters
Phthalates or phthalic acid esters are extensively utilized as key additives in the manufacture and processing of plastic products. As they are physically rather than chemically bonded to the polymer, they can be easily released into the environment. There are potential adverse effects on ecological system and 380 human health due to their toxicity, e.g., carcinogenicity and endocrine disruption (Fu et al., 2010;Li et al., 2016). Diethyl (DEP), di-n-butyl (DnBP) and bis-(2-ethylhexyl) (DEHP) phthalates were analyzed in current study. The annual average concentration of phthalates was 510±230 ng m -3 (165-1520 ng m -3 ) (Table 1). They showed higher concentration during pre-monsoon (Fig. S4).  TgC/yr) (Guenther et al., 1995). It should be noted, besides biogenic emissions, combustion of biomass and fossil fuel also contribute to the isoprene, monoterpenes and sesquiterpenes (Jathar et al., 2014;Sarkar et al., 2016;Sarkar et al., 2017). The investigation of gaseous VOCs during winter (December 2012 to February 2013) air in the KV during SusKat-ABC campaign also showed high levels of isoprene and it was attributed (at least during high isoprene periods) mostly to biogenic emissions (Sarkar et al.,400 2016; Sarkar et al., 2017). It is difficult to appropriately quantify the fractions of biogenic and anthropogenic emissions of these compounds, based on ambient measurement of these species alone without measurement of BB tracers such as acetonitrile and furan. The budget of isoprene emissions (500 Tg y -1 ) on a global scale is dominated by vegetation (Guenther et al., 2006). Therefore, in our study, we considered the oxidation products of isoprene, monoterpenes and sesquiterpenes as the tracers of biogenic 405 emissions and attribute their main source as biogenic emissions. This may lead to some overestimation of their contributions to SOA formation.  (Table 1). During the post-monsoon and pre-monsoon periods, their concentrations were similar, and a little lower than those during the monsoon (Fig. 5d) while being the lowest during winter. Their seasonal variation was in agreement with the ambient temperature (Fig S1a), which can influence the isoprene emissions and the photochemical processes (Shen et al., 2015;Wang et al., 2008). The annual average
For the seasonal variation, relatively high concentrations of M-SOA tracers occurred during premonsoon and post-monsoon seasons (Fig. 5e,f,g,h,and i (Jathar et al., 2014;Ding et al., 2013). Good correlations were obtained between the BB tracer i.e. levoglucosan and the 465 higher generation oxidation products (e.g., 3-HGA and MBTCA, R 2 =0.32 and R 2 =0.53 respectively) in the Bode aerosols (Fig. S6). The forests in the KV mainly consist of broad-leaved evergreen mixed forest, oak-laurel forest and oak forest as well as the conifer tree species (Department of Plant Resources, 2015; Sarkar et al., 2016). Monoterpenes were mainly released from coniferous trees .
Therefore, it suggested that the atmospheric aerosol compositions especially of SOA tracers over KV

Aromatic SOA tracer
Anthropogenic SOA is also an important OC source. DHOPA is a tracer of anthropogenic SOA from aromatics. In this study, DHOPA showed higher levels in winter and pre-monsoon periods while lower in monsoon season (Fig. 5). Though the major emissions of aromatics come from fossil sources, BB is 480 also considered as an possible source in some sites of the world (Shen et al., 2015). There was a good correlation between DHOPA and levoglucosan (Fig. 9), especially during pre-monsoon with the value of R 2 at 0.73. This indicated BB emissions are also significant source of DHOPA in Bode.

Estimation of the contributions of different sources to OC
As discussed above, both the primary and secondary sources have influence on OC in the 485 atmospheric aerosols of the KV. In this part, we will apply the tracer-based methods to evaluate the different sources' contributions to OC. It should be noted here that tracer methods can provide a reasonable estimation, but uncertainties are introduced considering the site differences and the lack of representative source profiles for the given study location. The contribution evaluated from each source to OC in the current study is still inferable.

BB-derived OC
Levoglucosan to OC ratios (Lev/OC) detected in source samples has been used in a wide range for quantitative estimation of the BB contribution to OC (Stone et al., 2012;Zhang et al., 2015;Wan et al., 2017), although the ratios vary among different types of biomass burnt and combustion conditions ). An average of 8.14% (8.0% to 8.2 %) for Lev/OC from the burning sites of biofuel, 495 savanna, crop residues, tropical forests, and so on was documented in (Andreae and Merlet, 2001). Zhang et al. (2007) obtained the Lev/OC ratios ranging from 5.4%-11.8% (an average of 8.27%) from the aerosols of cereal straw (wheat, corn and rice) combustion. Sheesley et al. (2003) reported an average of 7.94% of levoglucosan from the combustion of biomass (including rice straw, biomass briquettes, dried cow-dung patties, etc.) indigenous to South Asia. However, the ratio acquired from the hardwood burning 500 in fireplaces and stoves in the US was 14%, which was applied at the background sites in Europe (Fine et al., 2004). Stone et al. (2012) used the Lev/OC ratio of 12%±0.2% during the burning of acacia wood at Godavari in the KV for the CBM profile source apportionment. The mean value of Lev/OC value of BB from main biomass types was 10.1%. In this study, we choose the mostly used values of 8.14% for BB estimation (Graham et al., 2002;Fu et al., 2014;Ho et al., 2014;Sang et al., 2011;Zhu et al., 505 2016; Mkoma et al., 2013). In addition, the uncertainties of using different ratios were also calculated (see Table S3). The diagnostic ratios among molecular tracers and OC (e.g., Lev/OC) from direct emissions are critical for more precise results. It's meaningful to understand the emission characteristics for individual OC emission categories, as well as in different locations, especially in South Asia.
These results indicate that BB severely affect the air quality in the KV, especially during the post-monsoon period. Similarly, Stone et al. (2010) reported 21±2% of OC in PM2.5 from Godavari rural site in the outskirts of the KV during 2006, was also attributed to the primary BB sources.

520
Primary biological aerosol particle (PBAP) has been identified as an important source using tracers (section 3.3.2). They are likely to have a big contribution to the aerosols in Bode. In order to reveal how much they are contributing to organic aerosols, "total" plant debris was calculated based on glucose following the equation ( As shown in Fig. 11a, fungal spore-derived OC and plant-debris-OC annually contribute to 3.15±2.86% and 1.42±1.03% of OC, respectively. The contributions were both higher in the monsoon season, with 5.85±2.50% for fungal spore-derived OC and 2.29±0.79% for plant-debris-OC to OC, respectively (Fig. 11c). During winter, the contributions were the lowest due to the inactive vegetation.

535
There are also some similar results from the literatures. For example, Zhu et al. (2016) reported the plant debris contribution to OC was 5.6% and 4.6% during nighttime and daytime respectively from aerosols in a mid-latitudinal forest. Szidat et al. (2006) reported the plant debris contributed to 3.2% of OC during summer in urban aerosols collected in Zurich, Switzerland. The contributions of fungal aerosol to OC was 8% in the aerosols from a Brazil urban site (Emygdio et al., 2018). Liang et al. (2017) reported the 540 fungal aerosol contributions of 3.5 ± 3.7% in aerosols from a rural site in Beijing, China. In marine aerosols, the fungal spores was documented the major contributor to total OC with 3.1 % (0.03 %-19.8 %) over the East China Sea . All above strengthened the importance of plant-debris and fungal spores to the aerosol burden in the atmosphere.

555
The total calculated concentrations of B-SOC vaired from 0.41 to 2.77 μg m -3 with an annual mean concentration of 1.36±0.49 μg m -3 , a higher concentration of 1.43±0.48 μg m -3 in monsoon and lower concentration of 0.86±0.20 μg m -3 in winter (Fig 10g). The B-SOC/OC showed a higher average percentage of 10.1%±3.34% in the monsoon season (Fig 11c), indicating B-SOC was an important OC sources in Bode during this period. During post-monsoon, B-SOC/OC declined to 5.36% (Fig 11d). The 560 B-SOC/OC showed the lowest value of 1.52%±0.70% in winter (Fig 11e), indicating that B-SOC had minor contributions to elevated OC in winter. The annual average concentration of A-SOC was 2.45± 1.45 μg m -3 , which is higher than the B-SOC. The highest A-SOC concentration was obtained in winter (3.27±1.25 μg m -3 ) (Fig.10h). A-SOC was the second most important contributor to OC after BB-OC. It is not only derived from increased fossil fuel combustion and the subsequent oxidation, but also from BB 565 emissions.
In total, SOC (including B-SOC and A-SOC) reconstructed using the formula above in this section was 3.81±1.63 μg m -3 , accounting for 15.0%±8.99% of OC.

Possible sources of the unidentified OC
On the whole, BB contributed one-fourth (24.9%±10.4%) of the OC in Bode, followed by A-SOC 570 (8.82%±5.55%), B-SOC (6.19%±4.49%), fungal-spores (3.15%±2.86%) and plant-debris (1.42%± 1.03%) (Fig. 11a). Nevertheless, there is still part of OC (55.5%) that we were not able to be attributed to any specific sources based on the tracers analyzed in current study. There are partly uncertainties caused by the organic tracer analyses (estimation of measurement uncertainty was shown in Table S2).
Furthermore, fossil fuel combustion and soil dust could be also notable fractions of OC in Bode aerosols.

575
Additionally, low molecular weight (LMW) dicarboxylic acids from both primary and secondary sources is also a remarkable contributor to atmospheric organic aerosols (Kawamura and Bikkina, 2016). Humiclike substances and amines can constitute another fraction of OC, but not well studied Laskin et al., 2015). Therefore, the possible contributions of the unidentified OC (55.5%) from various sectors need further investigation, which is better to comprehensively understand the sources of 580 South Asian aerosols and will be very useful for the targeted pollution control measures in this region.

Summary and conclusions
Field measurements of atmospheric aerosols were conducted in a semi-urban site ( Understanding OC's climate impacts is a frontier area of research, because a large uncertainty still exists in the estimation of OC radiative forcing. Our study implies that since BB is a major source of ambient 605 OC, the fraction of OC that absorbs light (referred to as brown carbon) and also acts as cloud condensation nuclei, needs to be further studied in order to better understand radiative effects of OC on regional climate change. The current source contribution estimates from the tracer-based methods do not accurately evaluate the large temporal variations from all kinds of sources. Contributions from other sectors (ca 55.5%), including low molecular weight dicarboxylic acids (Kawamura and Sakaguchi, 1999;Kawamura 610 and Bikkina, 2016), need further investigation to better understand the atmospheric aerosols from both urban and rural sources such as the KV and other sites in the Himalayan foothills and the Indo-Gangetic Plain regions. These observations of the severe air pollution, particularly the particular matter pollution, provide valuable support for air pollution control measures, especially in determining which sources and sectors to first focus on the KV and the surrounding region, in order to reduce the air pollution from being 615 severe to become much cleaner in the near future. In addition, the current study based on the molecular level-source apportionment of OC in heavy polluted region from South Asia provides a much more specific quantification of source estimation for OC, which is different from previous studies based on the bulk carbonaceous aerosol using radiocarbon ( 14 C) measurements, PMF and CBM.
There are additional improvements for future studies to be addressed in the supporting information