Measurement report: Quantifying source contribution and radiative forcing of fossil fuel and biomass burning black carbon aerosol in the southeastern margin of Tibetan Plateau

Black carbon (BC) aerosol plays a vital role in disturbing the balance of ecosystem and climate stability of Tibetan Plateau (TP). An intensive campaign was carried out from 14th March to 12th May 2018 in the 15 southeastern margin of TP to investigate the sources of BC and their radiative effects. To do so, an improved ‘aethalometer model’ was used to distinguish and apportion BC into fossil fuel combustion source and biomass burning source. To minimize the uncertainty associated with the ‘aethalometer model’, a receptor model coupling multi-wavelength absorption with chemical species was used to retrieve the site-dependent Ångström exponent (AAE) and BC mass absorption cross-section (MAC). The results 20 show that the AAEs and BC MACs at wavelength of 880 nm were 0.9 and 12.3 m2 g-1 for fossil fuel source and 1.7 and 10.4 m2 g-1 for biomass burning, respectively. Based on these parameters, the fossil fuel source-related BC (BCfossil) was estimated 43% of the total BC and the rest 57% was from biomass burning (BCbiomass) during the campaign. The results from a regional chemical dynamical model reveal that high BCbiomass was contributed from the northeastern India and northern Burma, and the Southeast 25 Asia can explain 40% of BCbiomass. The high BCfossil was mainly identified from the southeast of sampling site. A radiative transfer model estimated that the atmospheric directive radiative forcing of BC was +4.6 ± 2.4 W m-2 on average, including+2.5 ± 1.8 W m-2 from BCbiomass, and +2.1 ± 0.9 W m-2 from BCfossil, https://doi.org/10.5194/acp-2020-408 Preprint. Discussion started: 23 June 2020 c © Author(s) 2020. CC BY 4.0 License.

which correspond to and heating rates of 0.07 ± 0.05 and 0.06 ± 0.02 K day -1 , respectively. Our study will be useful for improving our understanding in BC sources on the TP and their climatic effect.

Introduction
The Tibetan Plateau (TP) is considered as a regulator of climate change in the northern hemisphere and plays a crucial role in global ecosystem and climate stability (Yang et al., 2014). It covers one of the 5 largest ice masses of the Earth system, which is known as the water tower of Asia (Immerzeel et al., 2010).
The glaciers on the TP recently shows are rapidly retreating, which could lead to adverse effects on Asian hydrological cycle and the formation of the Asian monsoon (Wu et al., 2015). Although the causes of glacier melting are complex, the increasing black carbon (BC) loading on the TP has an inescapable role in this process (Yang et al., 2015). The BC deposited on glaciers increases ice melting by the enhanced 10 radiation absorption on the surface, which further influences the mass balance of glaciers (Ming et al., 2009). The main paths for BC in glaciers are scavenged from atmosphere through dry and wet deposition (Ménégoz et al., 2014). Thus, exploring the source features of atmospheric BC on the TP is critical for further understanding its environmental and climatic effects.
As well known, BC is a byproduct from incomplete combustion process of carbon contained fuels (e.g., its robustness depends on the selection of absorption Ångström exponent (AAE) and BC mass absorption cross-section (MAC) used in the model. In current studies, most of them cited source-specific AAEs from 10 previous publications and assumed the same BC MACs regardless of its difference in sources (e.g., Healy et al.;2017;Zhu et al., 2017;Martinsson et al., 2017;Rajesh and Ramachandran, 2018;Dumka et al., 2018;Liu et al., 2018;Helin et al., 2018). This introduces great uncertainties, because even for the same type of emission source, the AAE and MAC can vary with specific fuel subtypes and combustion conditions (Wang et al., 2018;Tian et al., 2019). To improve the accuracy, several studies tried to 15 constrain the AAE values by a comparison between different source-generated BC and the source markers (e.g., levoglucosan) or by the 14 C method (Sandradewi et al., 2008;Helin et al., 2018;Zotter et al., 2017). However, several studies have shown that the 'aethalometer model' is not as sensitive to the AAE variation of biomass burning source as to that of fossil burning source (Dumka et al., 2018;Titos et al., 2017), thus, the comparison method may be less effective to nail down the accurate source-specific AAE. 20 In addition, all the 'aethalometer model' users ignore the contribution of light absorption formed by the secondary processes which may account for a substantial fraction at short wavelengths (e.g., 370 and 470 nm) (Wang et al., 2019a;2019b). Consequently, this will lead to an overestimation of BC source apportionment results.
BC is recognized as the second largest anthropogenic warming agent globally after carbon dioxide, and 25 its radiative effect varies largely with different areas (Bond et al., 2013). Currently, regional modeling and observation-based methods have been used in assessing BC radiative forcing on the TP. In this regard, most studies concentrate on the areas of longitude between 72 -86° (e.g., Rajesh and Ramachandran, https://doi.org/10.5194/acp-2020-408 Preprint. Discussion started: 23 June 2020 c Author(s) 2020. CC BY 4.0 License. 2018; Bhat et al., 2017;Dumka et al., 2013;Singh et al., 2018). These studies focus on the radiative effect caused by bulk BC aerosol and some of them indicate that BC can account for 55 -70% of the composite aerosol atmospheric radiative forcing (Panicker et al., 2010;Srivastava et al., 2012;Sreekanth et al., 2007).
As summarized above, owing to the lack of studies regarding BC source apportionment on the TP, to our knowledge, there is no literature referring to the impacts of source-specific BC on the radiative forcing at 5 present.
In this study, a receptor model coupling with multi-wavelength absorption and chemical species was utilized to retrieve the site-dependent AAEs and BC MACs, and then an improved aethalometer model was used to determine the mass portions of BC aerosol from fossil fuel combustion source and biomass burning on the TP. Regional transport of source-specific BC was further explored by models. Finally, the 10 radiative effect caused by BC from different sources were evaluated using a radiative transfer model. Our study provides insights into the BC sources on the TP and their impacts on regional climate.

Sampling site
Intensive observation from 14 March to 13 May 2018 were conducted on the rooftop of. a 10m height 15 building above the ground (3260 m a. s.l.) in the Lijiang Astronomical Station, Chinese Academy of Sciences (100.03°E, 26.70°N), which was located in Gaomeigu County, Yunnan Province, China (Fig.   S1). The location connects high altitude TP with low altitude Yungui Plateau forming a transportation channel for pollutants from Southeast Asia (Wang et al., 2019a). The population of Gaomeigu County is small, and limited anthropogenic activities were found around the sampling site.
Briefly, light at λ = 370, 470, 520, 590, 660, 880, and 950 nm from diodes were utilized to irradiate the filter deposition spot, and the produced light attenuation is detected by optical detectors. Non-linear loading and filter matrix scattering effects are common issues of filter-based absorption measurements (Coen et al., 2009). The former issue was resolved using a dual-spot compensation technique embedded in AE33 aethalometer while a factor of 2.14 was used to correct the latter issue caused by quartz filters 5 (Drinovec et al., 2015).
A photoacoustic exinctionmeter (PAX, Droplet Measurement Technology, Boulder, CO, USA) was used to determine the aerosol light scattering coefficient (bscat) and babs as well as obtaining the single scattering albedo (SSA = bscat/(bscat+babs)) at wavelength of 532 nm. The bscat was measured using a wide-angle (5 -175 degree) integrating reciprocal nephelometer in the scattering chamber; and the babs was measured 10 simultaneously with an intracavity photoacoustic technique in the acoustic chamber. More detailed description of PAX can be found in Carrico et al. (2018). During the campaign, the bscat and babs measurements were calibrated by different concentrations of ammonium sulphate and freshly-generated propane soot, respectively. Detailed calibration procedure has been elaborated in Wang et al. (2018).
Twenty-four-hour PM2.5 samples were collected from 10:00 local time to 10:00 the next day on quartz 15 fibre filters (8 × 10 inches, QM/A™; Whatman, Middlesex, UK) by a high-volume sampler (Tisch Environmental, Inc., USA) with an operating flowrate of 1.05 m 3 min -1 . The blank filters before sampling and the loaded filters after sampling were weighted to obtain PM2.5 mass in a thermostatic room using a Sartorius MC5 electronic microbalance (Sartorius, Göttingen, Germany). The sampled filters were wellstored in a refrigerator at -4°C to reduce the loss of volatile substances prior to chemical analysis. The 20 chemical species involved carbonaceous matter (i.e., organic carbon (OC) and elemental carbon (EC)), water-soluble inorganic ions (i.e., K + ), and inorganic elements (i.e., S. Ca, Ti, Mn, Fe, Cu, As, Br, Pb, Zn), which were determined by thermal/optical carbon analyser, ion chromatograph, and energydispersive X-ray fluorescence spectrometry, respectively. Furthermore, organic markers of levoglucosan and benzothiazolone were measured using an ion chromatograph and a high-performance liquid 25 chromatography (HPLC), respectively. More detailed information regarding above measurements is shown in Text S1 in the supporting information.

BC source apportionment
The aethalometer model (Sandradewi et. al., 2008) was improved by considering the babs caused by the secondary formation and soil dust. Therefore, the babs of the BC generated from fossil fuel combustion and biomass burning were calculated as follows: (1) where AAEfossil and AAEbiomass are the AAEs corresponding to fossil fuel contribution and biomass burning emissions, which were derived from an optical source apportionment method as discussed in represents babs caused by secondary processes at λ = 370; babs(370)dust represents babs associated with soil dust at λ = 370, which was calculated in section 3.1. In this study, a BC-tracer method coupled with a 15 minimum R-squared approach was used to retrieve the babs(370)secondary (Wang et al., 2019a): where ( where MACBC(880)fossil and MACBC(880)biomass are the MAC of BC generated from fossil fuel combustion and biomass burning, respectively, which were retrieved from optical source apportionment as discussed in section 3.1.

Optical source apportionment
The optical source apportionment was realized via a positive matrix factorization (PMF) model. The fundamental principle of PMF is to resolve the chemical mass balance by separating data matrix into factor contributions and factor profiles as follows: where 0B represents the input matrix elements; p is the number of sources; 0? is the source contribution of the th factor to the th sample; ?B is the factor profile of th species in the th factor; and 0B is representative of the residual. 0? and ?B are non-negative. The two matrices are resolved by minimizing the sum of squares of the normalized residuals as follows: where represents the object function; and 0B denotes the uncertainties of 0B . The PMF version 5.0 (PMF5.0) was utilized, and the optical parameters (primary babs at different wavelengths) and chemical species (including carbonaceous matter, inorganic elements, K + , and organic markers) were both used as model inputs to perform the optical source apportionment. 20 To determine the influences of BC regional transport to Gaomeigu, cluster analysis was based on the hourly three-day backward trajectories at 500 m above the ground level using the Hybrid Single-Particle Lagrangian Integrated Trajectory model (Draxler and Hess, 1998). Since we focused on differentiating https://doi.org/10.5194/acp-2020-408 Preprint. Discussion started: 23 June 2020 c Author(s) 2020. CC BY 4.0 License. and clustering the main directions of coming trajectories, an angle-oriented distance definition was adopted in the cluster analysis. The detailed description of the cluster analysis can be found in Wang et al. (2018). To distinguish the pollution direction, the trajectory with BC mass concentration over 75th percentile was referred as a polluted trajectory, otherwise was a clean one.

Trajectory-related analysis
The potential source contribution function (PSCF) was further used to identify the likely pollution regions 5 that influenced BC loadings at Gaomeigu based on the backward trajectories. The geographic region covered by the trajectories was overlapped by a of 0.5 × 0.5°(i, j) grid layer. The PSCF value of each grid was calculated as follows: where 0B is the number of the endpoints associated with BC mass concentration higher than the set 10 criterion; and 0B is the total endpoints of the th cell. To improve the resolution of PSCF source identifications (Cheng and Lin, 2001), the 75th percentile of each source's BC mass concentration was set as the pollution criterion (i.e., 0.6 μg m -3 for BCbiomass and 0.45 μg m -3 for BCfossil). Further, an arbitrary weighting factor ( 0B ) was applied to different 0B ranges to reduce the uncertainty caused by the small 0B (Polissar et al., 1999), and it was defined as follows (Polissar et al., 2001):

Regional chemical dynamical model
The Weather Research and Forecasting model coupled with chemistry (WRF-Chem) model was used to quantify the biomass-burning contribution of Southeast Asian to BC mass at Gaomeigu. More detailed descriptions of the model configurations are described in our previous publication (Xing et al., 2020). 20 Briefly, the model resolution was 3 km × 3 km, and there were 320 grids in each of the latitude and longitude. The domain was concentrated in the southwest of China, South Asia, and Southeast Asia, with the central location at 100.03°E, 26.70°N. Thirty-five vertical layers has been set in the model from the ground surface to 50hPa. A BC emission inventory of 0.25° × 0.25°spatial resolution was used in model, including industry, power, transportation, and residential sources (e.g., fossil fuel and biofuel), which was based on the Asian anthropogenic emission inventory (that is MIX) for the year of 2010 .
The FINN fire inventory was used for the biomass-burning emission during the simulation. 5 The direct radiative forcing (DRF) of source-specific BC was estimated by the widely used Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART) model. The detailed description of SBDART model is elaborated in Ricchiazzi and Yang, (1998). The important input parameters include aerosol optical depth (AOD), extinction coefficient, SSA, asymmetric parameter (ASP), and visibility. These aerosol optical parameters were estimated by Optical Property of Aerosol and Cloud (OPAC) model using 10 Mie theory (Hess et al., 1998). The measured BC, water-soluble matter (including measured water-soluble inorganic ions and water-soluble organic matter that assumes accounting for 79% of OC loading on the TP, Xu et al., 2015), and water insoluble matter (calculated by the mass concertation of PM2.5 minus BC and water-soluble matter) were used in the OPAC model to retrieve the number concentrations of these matters which were then further used to obtain the optical parameters at the nearest observed relative 15 humidity. These number concentrations were tuned till the modelled bscat, babs, SSA were close to the ones measured by PAX, with a difference within ±5% (Srivastava et al., 2012) (see Table S1). The underlying assumption is when the modelled bscat, babs are very close to their measured counterparts, the rest derived optical parameters are assumed reasonable for the measured aerosols. This assumption has been widely accepted in previous studies (Dumka et al., 2018;Panicker et al., 2010;Rajesh et al., 2018). Finally, the 20 DRF induced by source-specific BC alone (or PM2.5) at the surface atmosphere (SUF) and the top of the atmosphere (TOA) were estimated by the difference in the net flux with and without BC (or PM2.5) under cloud-free conditions.

Estimations of direct radiative forcing and heating rate
Atmospheric forcing (DRF at TOA subtracts DRF at SUF) leads to solar heating rate change, which can be calculated as follows (Ramachandran and Kedia, 2010): where IJ IK is the heating rate (K day -1 ); is the acceleration due to gravity (9.8 m s -2 ); L is the specific heat capacity of air at constant pressure, ∆ )M> is the atmospheric forcing; and ∆ is the atmospheric pressure between the ground and 3 km above.

5
Based on the optical source apportionment, four sources were identified contributing to primary babs(λ) are widely used markers for biomass burning (Urban et al., 2012). Owing to the presence of BrC, the absorption fraction increased toward to short wavelengths, which is consistent with the optical feature of biomass burning (Forello et al., 2019). Therefore, this source factor was assigned to biomass burning.
Based on the contributions of biomass burning to babs(370) and babs(880), the AAEbiomass was estimated to be 1.7, which is within a relative boarder range of AAEbiomass (1.2 -3.5) reported by previous studies 15 obtained via comparison with other measurements (e.g., △ 14 C and organic tracers) in the atmosphere (Sandradewi et al., 2008;Helin et al., 2018;Harrison et al., 2012;Zotter et al., 2017). The estimated average MACBC(880)biomass was 10.4 m 2 g -1 , which was over two times larger than the value of uncoated BC particles suggested by Bond and Bergstrom, (2006)  The second source factor was characterized by large loadings of benzothiazolone (54%), Pb (46%), Br (40%), Cu (35%), Zn (27%), EC (36%), and OC (30%). The benzothiazolone is a substance released from to motor vehicle emission (Guo et al., 2009), and Zn and Cu are associated with the combustion of lubricating and wearing of brake and tires (Lough et al., 2005;Song et al., 2006). Meanwhile, EC and OC also can be used to denote motor vehicle emissions (Cao et al., 2013). Although unleaded gasoline has been extensively used in China since 2005, a considerable portion of Pb is still found in vehicle-related particles, which may be relation with the wear of metal alloys (Hao et al., 2019). Therefore, this source 5 factor was identified as traffic emissions. This source constitutes a moderate percentage of primary babs(λ) (15-30%). The estimated traffic emission-related AAE (AAEtraffic) was 0.8, consistent with the feature that BC is the dominant light-absorbing carbon for traffic emissions (Kirchstetter et al., 2004). The AAEtraffic here is also close to those obtained with △ 14 C approach (Zotter et al., 2017). The estimated BC MAC(880) of traffic emissions (MACBC(880)traffic = 9.1 m 2 g -1 ) was similar with MACBC(880)biomass, 10 indicating that traffic emission-related BC particles were also subjected to substantial aging.
The third factor was dominated by high loadings of As (70%), S (37%), and Cu (47%), which is a typical feature of coal combustion (Hsu et al., 2016;Kim and Hopke, 2008). Although coal is a scarce energy used on the TP, it can be influenced by coal combustion from surrounding areas (e.g., East Asia, Li et al., 2016). This source contributed only 12-17% of primary babs(λ), less than biomass burning and traffic 15 emissions. The obtained AAE of coal combustion (AAEcoal = 1.1) was similar to AAEtraffic, suggesting that BC was also a dominant light-absorbing carbon in coal combustion emissions. The AAEcoal is close to the value of chunk coal combustion (1.3) but lower to briquettes coal (2.6) (Sun et al., 2017), which may reflect the coal type that affected BC particles at Gaomeigu. The estimated BC MAC(880) of coal combustion (MACBC(880)coal = 15.5 m 2 g -1 ) was larger than MACBC(880)biomass and MACBC (880)traffic. 20 Numerous studies have confirmed that aged BC could result in MAC increased by a factor of 1. 5 -3.5 relative to uncoated one Ma et al., 2020). The enhance factor for MACBC(880)coal (3.3) falls near the upper limit of this range. This large enhancement should be related to the aging processes of BC particles during their long-range transport to Gaomeigu, although more work is need to explain in the future. 25 The fourth source factor was enriched of high loadings of Ca (35%), Ti (66%), Mn (47%), and Fe (61%), consistent with the characteristics of crustal elements (Guo et al., 2009). Thus, it was assigned to soil dust.
The light absorption of soil dust is mainly due to the presence of iron oxides and varies with the diversity https://doi.org/10.5194/acp-2020-408 Preprint. Discussion started: 23 June 2020 c Author(s) 2020. CC BY 4.0 License. of iron contents (Alfaro, 2004;Valenzuela et al., 2015). The mineral dust here contributed minor to primary babs(λ) (6-9%) due to the low content of iron oxides, which was consistent with the result obtained on the southeastern TP (Zhao et al., 2019). The estimated AAE of mineral dust (AAEdust) was 1.5, which is within the range of 1.2-3.0 obtained from multi-non-oceanic sites (Dubovik et al., 2002). 5 As fossil fuel-related BC aerosol involves BC emitted from traffic and coal combustion, the AAEfossil (0.9) and MACBC (880) (7)). As 10 shown in Fig. S3a, no correlation (r = 0.01, p = 0.02) was found between BCbiomass and BCfossil, implying that BC from these two sources can be separated reasonably by the improved aethalometer model. The biomass-burning tracer (levoglucosan) and traffic-related tracer (benzothiazolone) were used to further verify the results of BC source apportionment. As expected, BCbiomass and BCfossil correlated significantly with levoglucosan (r = 0.75, p < 0.01, Fig. S3b) and benzothiazolone (r = 0.67, p < 0.01, Fig. S3c), 15 respectively. These results indicate that the source-specific AAEs and MACs(880) obtained from optical source apportionment were appropriate for the improved 'aethalometer model'. an arithmetic mean (± standard deviation) of 0.7 (± 0.5) μg m -3 , which was lower than those in the western 20 TP but higher than in the northern TP (Wang et al., 2018, and references therein). The larger BC loading in the western TP is mainly attributed to the influences of Southeast Asia, where anthropogenic actives are intensive (Kurokawa et al., 2013). In terms of different BC sources, the mass concentrations of BCbiomass (0.4 ± 0.3 μg m -3 , 57% of total BC) was slightly higher than BCfossil (0.3 ± 0.2 μg m -3 , 43% of total BC) on average. The mass fraction of BCbiomass enhanced as the BC loading increased while BCfossil 25 mass fraction showed an inverse trend (Fig. S4). This suggests and biomass-burning emissions were responsible for the high BC loading episode during the campaign.

BC source apportionment
Distinct diurnal variations in mass concentrations of BCbiomass and BCfossil were observed (Fig. 3a). For BCbiomass, it started to increase after midnight reaching a small peak at 05:00 and then remained at a constant level before sunrise (~08:00). This may be attributed to the accumulative effect caused by planetary boundary layer (PBL) height (https://rda.ucar.edu/datasets/ds083.2), which showed a declining trend during this period (Fig. 3b). Thereafter, the BCbiomass increased again and reached the maximum 5 value of the day at midday. This enhancement was accompanied by the increased PBL height and wind speed (https://rda.ucar.edu/datasets/ds083.2) (Fig. 3b). Generally, deeper PBLs and stronger winds accelerate local pollutants diffusion and as a result lower their loadings . However, the increased trend here was more likely influenced by regional transport of BCbiomass from upwind region.
After sunrise, the PBL height began to rise and accompanied with west/southwest winds from 08:00 to 10 12:00 (Fig. S5). These favourable meteorological conditions provided good transport advantages for pollutants (including BC) from high-density biomass-burning emission areas to the sampling site (Chan et al., 2017). After the midday peak, BCbiomass decreased sharply until midnight. The initial decrease (13:00 -18:00) was mainly attributed to the continuous increases in PBL height and wind speed, which promoted BCbiomass diffusion. The further reduction at night was due to the scarcity of local biomass- 15 burning activities, although the PBL height and wind speed decreased during this period.
As shown in Fig. 3a, different from the unimodal diurnal pattern of BCbiomass, the BCfossil showed a roughly opposite trend. Owing to the influence of diffusion effect caused by increases in PBL height and wind speed, BCfossil decreased from 09:00 to 15:00. The initial morning decrease in BCfossil was just opposite to the increasing trend of BCbiomass, reflecting minor regional transport contributing to BCfossil. Further, 20 because of the small contribution of coal combustion to EC (12%, Fig.1c), the BCfossil was mainly contributed by motor vehicle emissions from the surrounding areas of southeastern TP. The subsequent increase in BCfossil from 17:00 to 20:00 was attributed to the reduction of PBL height and as a result accumulated pollutant in the southeastern TP. Since there were scarce traffic activities at night, the BCfossil remained at a steady concentration from 21:00 to 08:00. The stable nocturnal BCfossil may reflect the 25 impact of fossil fuel emissions on BC in the southeastern margin of TP due to the accumulation effect driven by the low PBL height.

Regional influences of BCbiomass and BCfossil
To investigate the regional impacts on BC, three groups of air masses were identified based on their transport pathways (Fig. 4a). Cluster #1 originated from northeastern India and then passed over Bangladesh before arriving Gaomeigu. The average BC mass concentration of this cluster was the highest (0.8 ± 0.4 μg m -3 ) among the three clusters. About 74% of total trajectories were associated with Cluster 5 #1, of which 22% was identified as polluted ones with an average BC loading of 1.3 ± 0.5 μg m -3 . Cluster #2 derived from Burma with an average BC loading of 0.7 ± 0.7 μg m -3 . This cluster constituted only 24% of total trajectories, but among them, about 37% was referred to pollution with BC reaching as high as 1.6 ± 0.9 μg m -3 . The air masses associated with Cluster #3 originated from interior of mainland China, which had the lowest BC mass concentration of 0.4 ± 0.1 μg m -3 . This cluster comprised small fraction 10 of total trajectories (2%), and none of them was identified as pollution, suggesting minor influence from the mainland China during the campaign.
The diurnal patterns of BCbiomass and BCfossil from different clusters were further used to investigate the impacts of regional transport. As shown in Fig. 3c and e, a similar diurnal variation of BCbiomass was found for Cluster #1 and Cluster #2, with larger values during daytime compared with the night-time. The 15 enhancements of daytime BCbiomass (8:00 -12:00) were related to the regional transport from northeastern India (Cluster #1) and Burma (Cluster #2). For Cluster #3, the diurnal variation of BCbiomass decreased during the daytime and increased at night (Fig. 3g), which was mainly controlled by the daily variability in PBL height. Compared to other two clusters, the inverse diurnal variation of BCbiomass for Cluster #3 indicates the influences of biomass-burning activities from the surrounding areas rather than regional 20 transport. However, it should be noted that these cases were scarce because of only 2% of air-masses associated with Cluster #3. For BCfossil, a similar diurnal pattern was found for Cluster #1 and Cluster #2 ( Fig. 3c and e), which was mainly associated with the influences of traffic emissions in surrounding areas as well as daily cycles of PBL as discussed in section 3.2. The BCfossil of Cluster #3 (Fig.3g) exhibited a relatively stable diurnal variation expect for sporadic fluctuations. Unlike the declining trend of BCfossil 25 during the daytime in the other two clusters, the relative stable variation in Cluster #3 indicates that there were fossil fuel sources to offset the effect of pollutant diffusion caused by increased PBL height. Owing to the high-density transportation network in mainland China (Liu, 2019), the regional transport of traffic emissions may be the cause.
Further, the PSCF model was applied to investigate the likely spatial distribution of pollution source regions for BCbiomass and BCfossil. As shown in Fig. 4b, a low PSCF value of BCbiomass was found near Gaomeigu while high values were concentrated in the northeastern India and northern Burma, consistent 5 with intensive fire spots in these areas (Fig. S6). This indicates that large BCbiomass loadings at Gaomeigu were more likely influenced by cross-border transport of biomass burning rather than local emissions. For BCfossil (Fig. 4c), the most possible impact region was located near the southeast of Gaomeigu, where there are a few villages scattered around and two highways (e.g., Hangzhou-Ruili Expressway and Dali-Nujiang Expressway). Owing to the low consumption of coal in the southeastern TP (Li et al., 2016), the 10 high PSCF values of BCfossil may be mainly contributed by traffic emissions rather than coal combustion.
Moreover, sporadic high PSCF values of BCfossil were also found in the northern Burma, indicating possible influences of fossil fuel emissions.
To further quantify the contributions of Southeast Asia transported to BC at Gaomeigu, a high BC episode (23 -27 March) was arbitrary selected and simulated by the WRF-Chem model. Two scenarios of 15 emissions were simulated: one involved all BC emission sources, and the other turned off biomassburning emissions in Southeast Asia. The variation of modelled BC mass concentration shows an acceptable consistency with the measured one (r = 0.63, p < 0.01, Fig. S7), and the index of agreement was estimated to be 0.77, indicating that the formation process of this BC episode was captured by the WRF-Chem model. However, it should be noted that the normalized mean bias between the measured 20 and modelled BC values was estimated to be 24%, suggesting an overestimation of simulation. This discrepancy could be attributed to the simulation uncertainties resulted from the emission inventory and meteorological conditions. Fig. 5a shows the spatial distributions of BC loadings in Gaomeigu and surrounding areas. The mass concentrations of BC can exceed 15 μg m -3 in Burma and northern India, which was over one order of magnitude higher than that in the southeastern margin of TP (0.7 μg m -3 ). 25 After turning off the biomass burning emission source in Southeast Asia, the BC loading at southeastern TP dropped over 40% (Fig. 5b), suggesting a substantial impact of biomass-burning activities in Southeast Asia countries on BC at the sampling site, which was consistent with cluster analysis and PSCF results. Fig. 6 shows the average DRFs of PM2. 5 and BC (including BCbiomass and BCfossil) at the TOA and SUF during the campaign. The average DRF of PM2.5 at the TOA was +0.03 ± 1.1 W m -2 , implying the importance of light-absorbing carbon. Actually, the BC can produce +1.6 ± 0.8 W m -2 at the TOA on average. At the SUF, BC forcing (-3.0 ± 1.5 W m -2 ) contributes near half of PM2.5 forcing (-6.3 ± 4.5 W 5 m -2 ). The net forcing trapped in the atmosphere was +4.6 ± 2.4 W m -2 for BC aerosol, which comprised 73% of that caused by PM2.5 (6.3 ± 4 W m -2 ), suggesting a substantial radiative effect caused by BC in the southeastern margin of TP, although its mass fraction is small in PM2.5 (3.3%).

Radiative forcing and heating rate
From the perspective of BC sources (Fig. 6), the average BCbiomass (BCfossil) DRF was +0.8 ± 0.6 W m -2 (+0.7 ± 0.4 W m -2 ) at the TOA, and -1.7 ± 1.2 W m -2 (-1.4 ± 0.6 W m -2 ) at the SUF. This causes an average 10 atmospheric radiative forcing of +2.5 ± 1.8 W m -2 (+2.1 ± 0.9 W m -2 ). Owing to the impacts of regional transport, the atmospheric forcing of BCbiomass was more fluctuation relative to BCfossil (Fig. S8). For example, the average atmospheric forcing of BCbiomass can reach as high as +5.6 W m -2 when the coming air masses passed over the biomass-burning regions in Southeast Asia.
The atmospheric forcing indicates the trapping of energy in the atmosphere leading to an increase in 15 atmospheric heating in the southeastern margin of TP. The heating rate caused by BC varied from 0.02 to 0.3 K day -1 with an average of 0.13 ± 0.07 K day -1 . In terms of the DRF efficiency, the heating rate caused by a unit BC mass concentration in this region is found higher (0.19 (K day -1 ) (μg m -3 ) -1 ) compared with that in Qinghai Lake, northeastern TP (0.13 (K day -1 ) (μg m -3 ) -1 ) , but it was generally lower than the values in the southwest of Himalaya regions (Fig. S9). The heating rate caused 20 by the BCbiomass is a slightly higher (0.07 ± 0.05 K day -1 ) compared with BCfossil (0.06 ± 0.02 K day -1 ).
The heating rate of BCbiomass increased to 0.16 K day -1 when the BC mass concentration was heavily influenced by the polluted air mass from Southeast Asia.

Conclusions
This study explored the source contribution and the direct radiative forcing of the black carbon (BC) 25 aerosol from fossil fuel combustion and biomass burning in the southeastern margin of Tibetan Plateau.
The observed mean BC concentration was 0.7 ± 0.5 μg m -3 during the campaign. Based on the optical source apportionment using multi-wavelength absorption and chemical species, the obtained absorption Ångström exponents (AAEs) and BC mass absorption cross section (MAC) at wavelength of 880 nm were 0.9 and 12.3 m 2 g -1 for fossil fuel source and 1.7 and 10.4 m 2 g -1 for biomass burning, respectively.
Using these source-specific AAEs and BC MACs, the improved aethalometer model estimated 43% of BC from fossil fuel source and 57% from biomass burning (BCfossil and BCbiomass, respectively). The 5 diurnal cycle in BCbiomass was driven by BC regional transport and the variation of planetary boundary layer (PBL) height. In contrast, BCfossil showed a roughly reverse diurnal cycle relative to BCbiomass, which was mainly due to traffic emissions in surrounding areas and the variation of PBL height.
Three groups of air masses were identified based on their transport pathways, with the highest BC loading from the direction of northeastern India, followed by Burma and with the lowest from mainland China. 10 The potential sources contribution function (PSCF) model showed that the northeastern India and northern Burma were the most possible contribution areas to BCbiomass at the sampling site, and the Weather Research and Forecasting model coupled with chemistry (WRF-Chem) model indicates that biomass burning from the Southeast Asia can contribute 40% of BC loading at the sampling site. The southeast of Gaomeigu was the most likely contribution region for the high BCfossil. 15 The Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART) model estimated a cooling effect of -3.0 ± 1.5 W m -2 at the Earth's surface and a warming effect of +1.6 ± 0.8 W m -2 at the top of the atmosphere caused by BC aerosol. The atmospheric forcing of BC was +4.6 ± 2.4 W m -2 , comprising 73% of that caused by PM2.5. The average BCbiomass DRF was higher than that caused by BCfossil. The average heating rate caused by BC was 0.13 ± 0.07 K day -1 , with 0.07 ± 0.05 K day -1 from BCbiomass and 20 0.06 ± 0.02 K day -1 from BCfossil. The heating rate of BCbiomass increased can increase to 0.16 K day -1 , when BC mass concentration was heavily influenced by the emission of biomass burning from Southeast Asia.
Data availability. All data described in this study are available upon request from the corresponding 25 authors.
Supplement. The supplement related to this article is available online.