Modeling the Origin of Anthropogenic Black Carbon and Its Climatic Effect Over the Tibetan Plateau and Surrounding Regions

Abstract

Black carbon (BC) in snow/ice induces enhanced snow and glacier melting. As over 60% of atmospheric BC is emitted from anthropogenic sources, which directly impacts the distribution and concentration of BC in snow/ice, it is essential to assess the origin of anthropogenic BC transported to the Tibetan Plateau (TP) where there are few direct emissions attributable to local human activities. In this study, we used a regional climate‐atmospheric chemistry model and a set of BC scenarios for quantitative evaluation of the impact of anthropogenic BC from various sources and its climate effects over the TP in 2013. The results showed that the model performed well in terms of climatology, aerosol optical properties, and near‐surface concentrations, which indicates that this modeling framework is appropriate to characterize anthropogenic BC source‐receptor relationships over the TP. The simulated surface concentration associated with the anthropogenic sources showed seasonal differences. In the monsoon season, the contribution of anthropogenic BC was less than in the nonmonsoon season. In the nonmonsoon season, westerly winds prevailed and transported BC from central Asia and north India to the western TP. In the monsoon season, BC aerosol was transported to the middle‐upper troposphere over the Indo‐Gangetic Plain and crossed the Himalayas via southwesterly winds. The majority of anthropogenic BC over the TP was transported from South Asia, which contributed to 40%–80% (mean of 61.3%) of surface BC in the nonmonsoon season, and 10%–50% (mean of 19.4%) in the monsoon season. For the northeastern TP, anthropogenic BC from eastern China accounted for less than 10% of the total in the nonmonsoon season but can be up to 50% in the monsoon season. Averaged over the TP, the eastern China anthropogenic sources accounted for 6.2% and 8.4% of surface BC in the nonmonsoon and monsoon seasons, respectively. The anthropogenic BC induced negative radiative forcing and cooling effects at the near surface over the TP.

1 Introduction

Black carbon (BC) is a carbonaceous aerosol that is mainly emitted from the incomplete combustion of carbon‐based fuels and materials. The efficient absorption of solar radiation by BC makes it the dominant insoluble light‐absorbing particulate species in the atmosphere. BC can heat the atmospheric and cool the surface through its effect on solar radiation (He, Li, Liou, Takano, et al., 2014), and furthermore modify atmospheric circulation patterns (Bond et al., 2013). Lau, Kim, and Kim (2006) and Lau et al. (2010) demonstrated that BC heating can strengthen local circulations and result in a northward shift of the monsoon rain belt and widespread warming over the Himalayas and Tibetan Plateau (TP).

In addition to absorbing and scattering light while suspended in the atmosphere, BC can reduce the amount of reflected sunlight when deposited on high albedo surfaces such as snow and ice. Climate models indicate that BC in snow/ice may play as large a role in climate change (Hansen et al., 2005; Qian et al., 2011; Zhao et al., 2014) due to a sequence of positive feedback mechanisms (Flanner et al., 2007; Hadley & Kirchstetter, 2012; Hansen & Nazarenko, 2004). Qian et al. (2011) indicated that BC in snow increases the surface air temperature by around 1.0°C on average over the TP and reduces the spring snowpack over the TP more severely than the reductions experienced due to the increased levels of CO₂ and carbonaceous particles in the atmosphere. Ji (2016) reported that BC in snow/ice increases surface temperature by 0.1–1.5°C and the snow water equivalent decreases by 5–25 mm over the TP. Moreover, the TP is surrounded by two large BC emission sources, East Asia and South Asia (Lamarque et al., 2010; Xu et al., 2009). Therefore, the snow/ice‐covered TP region is more prone to the effects of BC than other regions.

The TP is the highest and largest plateau in the world. It covers an area of 3.6 million km² at 2 000 m above the mean sea level (see Figure 1). The TP stores the largest ice mass outside the polar regions and has a vast area of seasonal and permanent snow cover. Thus, it represents a very sensitive and visible indicator of climate change, with its unique location facilitating complex interactions among the atmosphere, hydrosphere, and cryosphere (Pu, Xu, & Salomonson, 2007; Xu et al., 2009; Yao et al., 2012). Glaciers and snow melt over the TP have great potential to modify the regional hydrology in and around the region (Bolch et al., 2012; Qin, Liu, & Li, 2006; Singh & Bengtsson, 2004; Yanai, Li, & Song, 1992; Yao et al., 2012). Moreover, the long‐term trend and/or seasonal shift of the water resource provided by the TP may significantly affect agriculture, hydropower, and even national security in Asia.

Figure 1

Model domain and topographic field (m). The 264 meteorological stations and 19 AERONET sites are denoted by black dots and red dots.

The TP's climate and environmental conditions have been rapidly changing (e.g., Chen et al., 2015). For example, the surface air temperature has increased by about 1.8°C over the past 50 years (Wang et al., 2008), and several studies have reported a faster warming than the global average (Kang et al., 2010; Xu et al., 2009). As a result of this enhanced warming, the glaciers over the TP have undergone increasing widespread losses in recent decades (Bolch et al., 2012; Kang et al., 2010; Li et al., 2008; Qin, Liu, & Li, 2006) and accelerated retreats in recent years (Ding et al., 2006; Yao et al., 2007). This directly impacts runoff and potentially increases natural hazards (e.g., glacial lake outburst floods), especially for those living downstream (Qin, Liu, & Li, 2006; Yao, 2010). The rapid warming and accelerated glacier retreat have been primarily attributed to increasing levels of greenhouse gases (Duan et al., 2006), but other factors may be partly responsible such as land use changes, heating by absorbing aerosols, and the reduction of snow albedo induced by light‐absorbing impurities in snow (Flanner et al., 2007, 2009; Ji et al., 2015; Kang et al., 2000; Qian et al., 2011, 2015; Ramanathan et al., 2007; Xu et al., 2009). Owing to the lack of local industrial emissions, the light‐absorbing aerosols over the TP are mainly transported from other locations.

Previous observational and modeling studies have demonstrated a rapidly increasing trend of BC deposition on snow and ice, and BC has significantly contributed to early snowmelt and rapid glacial retreat over the TP (Kang et al., 2010; Kaspari et al., 2011; Ming et al., 2008; Qian et al., 2011, 2015; Xu et al., 2009). It has become increasingly important to understand the role of BC as anthropogenic forcing, given these. Distribution and variation of BC in snow and ice are determined by the levels and sources of atmospheric BC, with over 60% of atmospheric BC originating from anthropogenic activities (Bond et al., 2007; Lamarque et al., 2010). Therefore, it is essential to quantify the anthropogenic contributions of different geographical regions and emission sectors to atmospheric BC over the TP. However, owing to the harsh environment, limited access for fieldwork, and the sparsity of fixed instrumental stations, observational records are too poor to enable detailed, quantitative studies of the spatial and temporal trends.

Some studies have used chemical transport models (CTMs) and global climate models (GCMs) to simulate the source of aerosols, their transportation, and climatic effects over the TP (He, Li, Liou, Zhang, et al., 2014; Kopacz et al., 2010, 2011; Lau et al., 2006; Lu et al., 2012; Menon et al., 2010; Zhang et al., 2015). Compared with the statistical analysis of back trajectory approach (Lu et al., 2012) and the adjoint of the GEOS‐Chem global chemical transport model that is not source attributions but rather the source‐receptor sensitivities (Kopacz et al., 2010, 2011), the aerosol‐climate model CAM5 (Community Atmosphere Model version 5) can provide a detailed evaluation of the origin of BC emitted from various sources. However, the coarse resolution of the GCMs makes it difficult to capture the surface details of the TP (Gao et al., 2008; Ji et al., 2011, 2015; Ji & Kang, 2013). Regional climate models (RCMs) can provide the high‐resolution simulations and compensate for the shortcomings of coarser global model grids. RCMs have been developed and now include multiple modules, including atmospheric chemistry processes. The advanced regional climate‐chemistry model, Weather Research and Forecasting (WRF) model (Skamarock et al., 2005) with Chemistry (WRF‐Chem), a horizontal grid spacing of 10 km, is found to perform well with regard to the seasonal variability, source, and transport of BC in India (Kumar et al., 2015).

In this study, we used the WRF‐Chem model to produce a detailed characterization of the anthropogenic contribution to the sources, transport, and climatic effects of atmospheric BC over the TP for one complete year. The paper is organized as follows. Section 2 describes the model, data, and experimental details. Section 3 describes the validation of model performance. Section 4 and 5 evaluate the contribution of anthropogenic BC sources to the TP and its impact on radiative forcing and temperature, respectively. Finally, section 6 presents the main conclusions.

2 Model and Data

2.1 WRF‐Chem Model

The mesoscale WRF meteorological model (Skamarock et al., 2005; http://www.wrf‐model.org) has been expanded to include a chemical component (WRF‐Chem). WRF‐Chem is an online three‐dimensional, Eulerian chemical transport model that considers complex physical and chemical processes, such as the emission and deposition of pollutants, advection and diffusion, gaseous and aqueous chemical transformation, and aerosol chemistry and dynamics (Grell et al., 2005). It is capable of simulating atmospheric chemistry on a regional scale and has been successfully used in air quality studies (Fast et al., 2006; Gao et al., 2015; Wang et al., 2015; Yang et al., 2017). The WRF‐Chem version 3.6 has been applied in this study, which also included the Georgia Tech/Goddard Global Ozone Chemistry Aerosol Radiation and Transport (GOCART) (Chin et al., 2000) dust emission model coupled with the Model for Simulating Aerosol Interactions and Chemistry (MOSAIC) (Zhao et al., 2010). All major aerosol components were included, namely, SO₄²⁻, NO³⁻, NH4⁺, BC, OC, sea salt, mineral dust, and aerosol water. Aerosols were assumed to be spherical particles. The size distribution was divided into four discrete size bins defined by their lower and upper dry particle diameters. Aerosols in each bin were assumed to be internally mixed via various chemical and microphysical processes, such as emissions, nucleation, condensation, coagulation, aqueous‐phase chemistry, and water uptake by aerosols. Aerosol optical properties such as extinction, single‐scattering albedo, and the asymmetry factor were computed as functions of wavelength and three‐dimensional position. Each chemical constituent of the aerosol was associated with a complex refraction index. For each size bin, the refractive index of the aerosols was derived by volume averaging, and Mie theory was used to determine the extinction efficiency, the scattering efficiency, and the intermediate asymmetry factor. A detailed description of the computation of aerosol optical properties in WRF‐Chem can be found in Fast et al. (2006) and Barnard et al. (2010).

In the WRF‐Chem model, improvements in modeling meteorological aspects can lead to better simulations of atmospheric chemistry. Based on our previous studies over the TP, we added the Jimenez subgrid‐scale orography parameterization scheme (Jiménez & Dudhia., 2012) and used the 2013 MODIS‐based land use data to replace the default 2001 land use data (as shown in Figure S1 in the supporting information; the land use types had clearly changed from 2001 to 2013 in the TP), proven to improve WRF modeling for meteorological fields over the northeastern TP (Yang & Duan, 2016). Additionally, we assimilated AMSUA/B radiance data (NOAA‐15/16/17/18/19) using the WRF‐3DVar assimilation system to improve the model's initial field, according to previous research (Yang et al., 2015).

2.2 Experimental Design and Emissions

The simulations were performed at 25 km horizontal resolution, covering the TP and its surrounding areas with 211 and 165 grid cells in the west‐east and north‐south directions, respectively, as noted in Figure 1. The simulated domain was centered at 33°N, 88°E, and the 30 vertical sigma levels from ground level to the top pressure of 50 hPa were used for all grids. The initial meteorological fields and boundary conditions were from National Centers for Environmental Prediction (NCEP) reanalysis data with horizontal resolution of 1° × 1° at 6 h time intervals. The simulation was conducted for the entire year of 2013, and the first 6 days were used for model spin‐up. Key physical parameterization options for the WRF‐Chem modeling were the Noah land surface scheme to describe the land‐atmosphere interactions (Ek et al., 2003), the Morrison 2‐moment microphysical parameterization (Gustafson et al., 2007) with the Grell‐Devenyi (GD) cumulus scheme (Grell & Devenyi, 2002) to reproduce the cloud and precipitation processes, the Mellor‐Yamada‐Janjic (MYJ) planetary boundary layer scheme (Schaefer, 1990), and the RRTMG longwave and shortwave radiation scheme (Mlawer et al., 1997). The selection of these physical parameterizations followed Y. He et al. (2012) who demonstrated that the selections performed well for modeling meteorological elements in the TP. Based on our previous sensitivity experiments using chemical schemes, the chemistry parameterization options were the Carbon Bond Mechanism version Z (CBMZ; Zaveri & Peters, 1999) for gas‐phase chemistry and the Model for Simulating Aerosol Interactions and Chemistry (MOSAIC; Zaveri et al., 2008) aerosol reactions scheme.

As shown in Figure 2, the domains of SA, SE, SS, SW, and SN represent the entire model domain, eastern China, South Asia, central Asia, and northern China, respectively. Six sensitivity experiments detailed in Table 1 were designed in this study to assess the relative importance of anthropogenic BC from the region outside of the TP. In the control experiment CT, the original values of BC emission was unchanged. In the other five experiments SA, SE, SS, SW, and SN, the anthropogenic BC was set to zero and biogenic burning emissions were reduced to 37% of the initial values in the SA, SE, SS, SW, and SN domains, respectively (Table 1). The biogenic burning emissions decreased to 37% in the sensitivity experiments because 63% of the large‐scale biomass‐burning emissions were induced by direct or indirect human activities (Bond et al., 2013). In this study, the difference between the control and sensitivity experiments was used to estimate the impact of various anthropogenic sources on the surface BC concentration and its climate effects over the TP. To compare simulation results in different seasons, we considered the monsoon season as lasting from May to September, and the nonmonsoon season as the remainder of the year (Wu & Zhang, 1998).

Figure 2

The SE, SS, SW, and SN domains and anthropogenic BC flux distribution (ug/m²/s).

Table 1. Experimental Design

Experiment name	Setup
CT	No change to BC emissions.
SA	In the entire study domain, anthropogenic BC emissions were set to zero and biomass burning emissions were reduced to 37% of the initial values.
SE	In the “se” domain, anthropogenic BC emissions were set to zero and biomass burning emissions were reduced to 37% of the initial values.
SS	In the “ss” domain, anthropogenic BC emissions were set to zero and biomass burning emissions were reduced to 37% of the initial values.
SW	In the “sw” domain, anthropogenic BC emissions were set to zero and biomass burning emissions were reduced to 37% of the initial values.
SN	In the “sn” domain, anthropogenic BC Emissions were set to zero and biomass burning emissions were reduced to 37% of the initial values.

The anthropogenic emission inventory at a resolution of 0.5° × 0.5° was based on the Intercontinental Chemical Transport Experiment‐Phase B (INTEX‐B, Zhang et al., 2009), including CO, volatile organic compounds (VOCs), NO_X , BC, OC, SO₂, PM_2.5, and PM₁₀. The emission fluxes of the main anthropogenic aerosol components indicated low concentrations in the TP and high concentrations in South Asia (Figure S2 in the supporting information). The greenhouse gases emission data set (e.g., CO₂, CH₄, and N₂O) was derived from the Reanalysis of the TROpospheric chemical composition (RETRO, http://retro.enes.org/index.shtml) with 0.5° × 0.5° resolution. The sea salt and volcanic ash emissions in the study domain were interpolated from a global emissions data set (Freitas et al., 2011). The fire emissions inventory was based on the fire inventory from NCAR (FINN) (Wiedinmyer et al., 2010). Biogenic emissions were from the Model of Emission of Gases and Aerosol from Nature (MEGAN) (Guenther et al., 2006). Additionally, the mozbc utility and the Model for OZone and Related chemical Tracers (MOZART, http://www.acom.ucar.edu/wrf‐chem/mozart.shtml; Emmons et al., 2010) data set were used to create improved lateral chemical boundary conditions.

2.3 Observation Data

In the study area, a total of 264 national observation stations from the China Meteorological Data Network (CMDN, http://data.cma.cn) were used to verify the model simulation of 2 m temperature, 2 m relative humidity, and 10 m wind speed in 2013. Figure 1 shows the locations of the 264 national stations as black dots. For the spatial pattern, a data set from the Climate Research Unit (CRU, Mitchell & Jones, 2005; http://crudata.uea.ac.uk/cru) was used to validate the simulated surface air temperature and the modeled surface relative humidity. Modeled 500 hPa wind was evaluated using the NCEP/NCAR Reanalysis monthly mean data (https://rda.ucar.edu/datasets/ds090.2; Kalnay et al., 1996). The identification of Aerosol Optical Depth (AOD) over the study region is based on quality‐assured data from the Aerosol Robotic Network (AERONET, https://aeronet.gsfc.nasa.gov), which has been reliably and continuously producing detailed optical properties of various types of aerosols worldwide (Dubovik et al., 2002; Holben et al., 2001). Here we used the level 2.0 AOD data from 2013. Figure 1 shows the 19 AERONET sites as red dots. Additionally, the near‐surface and snow BC concentration at 17 sites from previous studies were collected for the validation of model performance.

3 Validation of Model Performance

3.1 Meteorology

The CRU data set and WRF‐Chem modeled surface air temperatures for both nonmonsoon and monsoon seasons are shown in Figure 3. The model‐simulated surface air temperature generally showed good agreement with the CRU data set. In the nonmonsoon season, the modeled temperature was as low as −16°C in the TP (Figure 3a). Cold biases were found in most regions of the TP compared to the CRU data (Figure 3c). The simulated temperature was in the range of 0–16°C during the monsoon season (Figure 3b) as warmer areas surround the TP with temperatures exceeding 28°C in South Asia and 24°C in Central Asia and the Taklamakan Desert. Compared with the CRU data (Figure 3d), the model showed lower temperature in the west of the TP and high values in South Asia.

Figure 3

Mean T2 (°C) in (a, c) nonmonsoon and (b, d) monsoon from WRF‐Chem simulated (Figures 3a and 3b) and CRU (Figures 3c and 3d).

Figure 4 depicts observed (NCEP reanalysis) and modeled average 500 hPa wind in both nonmonsoon and monsoon seasons, respectively. In the nonmonsoon season, strong westerly winds prevailed north of 15°N (Figures 4a and 4c). Compared with the NCEP reanalysis data, the model simulated the nonmonsoon dynamics well. The northwest air flow was partially divided into two branches due to the topography. One branch flowed due west, and the other was forced up by high terrain and followed a northwesterly path. Finally, the two branches converged to form westerlies at around longitude 95°E (Figure 4a). In the monsoon season, there were prevailing southwesterly winds from the Bay of Bengal to low‐latitude areas and westerly winds dominated in the higher‐latitude regions (Figures 4b and 4d). There was a difference between the observation and simulation, with two local disturbances in South Asia being weaker in the NCEP reanalysis data (Figure 4d) than in the simulation (Figure 4b). It is possible that the reanalysis did not capture such characteristics due to the coarse resolution over steep mountain regions (Himalaya), while the high‐resolution WRF‐Chem had a fine topographically induced structure. This phenomenon was discussed by Gao et al. (2008), who found that the high‐resolution model inhibited the penetration of the monsoon precipitation front from the southern slopes of the Himalayas.

Figure 4

Mean wind at 500 hPa (m/s) in (a, c) nonmonsoon and (b, d) monsoon from the WRF‐Chem simulation (Figures 4a and 4b) and NCEP reanalysis (Figures 4c and 4d).

Precipitation in the monsoon season was much greater than in the nonmonsoon season for both the simulation and the observation (Figure 5). The model could simulate the basic position of the rain band over the TP. Precipitation was low in the plateau and large areas received lower than 250 mm in the nonmonsoon season (Figure 5a). In the monsoon season, precipitation was clearly affected by the TP's terrain, with high values apparent along the windward slopes of the mountains (Figure 5c). The rain band in the Indian subcontinent reflected the monsoon's progress. In the nonmonsoon season, precipitation was low, but larger values were apparent during the monsoon season. Along the southwest coast of India, precipitation was mostly concentrated in the mountains on the windward side in the monsoon season due to blockage by the Deccan Plateau. Compared with the CRU precipitation (Figures 5b and 5d), the model simulation displayed subtler orographeric characteristics such as high precipitation along the Himalayas, Karakoram Mountains, and Tianshan Mountains. CRU data did reflect these features, and gridded observation data may be unreliable in some remote areas (Wu & Gao, 2013). Gridded precipitation was interpolated from meteorological stations that were concentrated in the valleys and experienced less precipitation than the mountainous locations, which may lead to a large underestimation in unpopulated mountainous regions. Thus, we used observed precipitation from 73 national meteorological monitoring stations over the TP to analyze the uncertainty in simulation evaluation. As shown in Figure S3, the CRU precipition was consistent with the observations and the WRF‐Chem results representd well the monthly variation of precipitation but with underestimation from May to October.

Figure 5

Cumulative precipitation (mm) in (a, c) nonmonsoon and (b, d) monsoon from WRF‐Chem simulated (Figures 5a and 5b) and CRU (Figures 5c and 5d).

The data of 264 stations were used to evaluate simulation precision for surface meteorological elements quantitatively. Figure 6 shows the mean daily variation of measured and simulated 2 m temperature (T2), 2 m relative humidity (RH2), and surface wind speed (U10) with box plots at meteorological stations in the study domain for 2013. The WRF‐Chem model showed a slightly higher T2 between January and March (mean bias (MB) was equal to 1.1°C), and a lower T2 between July and September (MB = −1.4°C) (Figure 6a). Figure 6b shows the model's representation of relative humidity annual variance, with an underestimation from January to March (MB = −5.7%) and an overestimation from July to September (MB = 8%). For the subgrid‐scale orography, the model performance in surface wind speed was not as good as the simulation of temperature and relative humidity (the Rs were equal to 0.96, 0.68, and 0.43 for T2, RH2, and U10, respectively, as shown in Table S1), and the simulation experiment produces a lower wind speed than the observation (MB = 0.2 m s⁻¹) (Figure 6c). The corresponding statistics between the observations and simulations are shown in Table S1 in the supporting information.

Figure 6

Mean daily variation of measured and simulated (a) T2, (b) RH2, and (c) U10 with box plot at meteorological stations in study domain for the year 2013.

3.2 AOD and BC Concentration

Figures 7a and 7b show the modeled AOD in the nonmonsoon and monsoon seasons accordingly. The results indicate that low AODs occurred over the TP, while high values were apparent over South Asia, the Sichuan Basin, and North China Plain. In the nonmonsoon season, AOD was less than that during the monsoon season over the TP, which may be due to more aerosols being transported to the TP from ambient regions with large concentration in the monsoon season. In eastern China and the Taklamakan desert, AOD was much larger in the nonmonsoon season than in the monsoon season, which was consistent with the distribution of the aerosol mass burden shown in Figure S2 in the supporting information. In South Asia, the model indicated low values of AOD in the southern Indian Peninsula and high values in central and eastern India. Owing to the Thar Desert generating large amounts of dust aerosols, AOD over northern Indian was greater in the monsoon season than in the nonmonsoon season. Specifically, Pandithurai et al. (2008) indicated that, in New Delhi, owing to dust storms from the Thar Desert, monthly mean AOD at 550 nm showed an increasing trend from March to June, with values of 0.55, 0.75, 1.22, and 1.18, respectively.

Figure 7

(a, b) Mean AOD and (c, d) mean surface BC concentration in nonmonsoon (Figures 7a and 7c) and monsoon (Figures 7b and 7d).

We selected 19 sites (red dots, Figure 1) from AERONET in this region and compared them with observed AOD values with simulated results from the control experiment (CT). Table 2 shows the corresponding statistical analyses of the comparisons, including the correlation coefficient (R ), mean bias (MB), and normalized mean bias (NMB). Compared with the observed AOD values at various sites, the modeled results show an underprediction (ME = −0.04, NMB = −3.5), with a mean correlation of 0.48. It can also be seen in Figure S4 that the model captures the temporal variability of observed AOD550 well throughout the year. The underestimation of AOD might be due to the assumed spherical aerosol particles in the model simulations. Previous studies (China et al., 2015; C. He et al., 2015) had shown that the optical properties of particles are more sensitive to nonspherical morphology than primary spherical structure.

Table 2. Statistical Performances of Simulated and Observed AOD550 at 19 AERONET Sites in the Year 2013

	Observed mean	Simulated mean	MB	NMB	R
Dhaka	0.76	0.57	−0.19	−25.0	0.56
Gandhi	0.63	0.54	−0.19	−14.3	0.47
Issyk	0.14	0.15	0.01	7.1	0.61
AOE_Baotou	0.15	0.21	0.04	26.7	0.45
Dalanzadgad	0.09	0.11	0.02	22.2	0.47
Dushanbe	0.24	0.32	0.08	33.3	0.41
Irkutsk	0.08	0.05	−0.03	−37.5	0.38
Jaipur	0.42	0.52	0.10	23.8	0.59
Jomson	0.10	0.11	0.01	10.0	0.61
Kanpur	0.62	0.56	−0.06	−9.7	0.48
Kathmandu	0.39	0.27	−0.12	−30.8	0.64
Lahore	0.65	0.57	−0.08	−12.3	0.49
Lumbini	0.55	0.26	−0.29	−52.7	0.36
Pokhare	0.43	0.32	−0.11	−25.6	0.21
Qoms	0.03	0.04	0.01	33.3	0.40
Iaocr	0.28	0.19	−0.09	−32.1	0.57
Mt WLG	0.23	0.29	0.06	26.1	0.57
Nam co	0.09	0.04	−0.03	−33.3	0.53
Evk2 CNK	0.04	0.05	0.01	25.0	0.46
All sites	0.31	0.27	−0.04	−3.5	0.48

Figures 7c and 7d show the modeled mean near‐surface BC concentration for the nonmonsoon and monsoon seasons, respectively. The spatial distribution showed high values along the outside margin of the TP and low flux in the inland regions. The two highest BC concentrations were in the North China Plain and Sichuan Basin (which is located to the east of the TP) due to the region's high population and megacities. In the nonmonsoon season (Figure 7c), surface BC aerosol was much larger east of 95°E than in the monsoon season (Figure 7d). This was possibly due to precipitation scavenging BC in the nonmonsoon season being less than during the monsoon season. As shown in Figure 5, the simulated precipitation east of 95°E was greater in the monsoon season and surface BC was less concentrated to the south of the Himalayas than in the nonmonsoon season because of the terrain blocking effect. This phenomenon was also reported in Ji et al. (2015) who used RegCM model. In the southern TP during monsoon season, large amounts of BC aerosol could potentially be transported from South Asia by southerly winds (the wind field shown in Figure 4b). Compared with BC concentration in the MOZART data set (Figure S5), the WRF‐Chem‐simulated surface BC concentration showed good agreement, although underestimation in the north of India.

We summarized the observations of BC concentration in atmosphere and snow at 17 sites to compare with model simulations (Table 3). Results confirm that BC is very low in the TP region. In Lhasa, Qomalangma, Namco, and Hulugou, the BC surface concentration was lower than 1 μg m⁻³. High pollution levels appeared in Sinhagad and Dhaka, with the surface BC concentration beyond 10 μg m⁻³. The simulation reproduced the spatial variation in surface BC concentrations, but there was an underestimation at these sites. Additionally, we used the observations of BC concentration in snow to evaluate the simulated results of atmospheric BC deposition. BC concentrations in snow were always higher at the periphery of the TP (such as in Laohugou (0.85 μg m⁻³) and Tienshan (0.4 μg m⁻³)), where the BC might be enriched by the proximity to sources (Ming et al., 2009). In inland of the TP, snow BC concentrations in Zhadang glacier, Southeast TP, and East Rongbuk were less than 0.1 μg m⁻³. The model also represented well the deposition pattern of atmospheric BC: high BC concentration in snow at Laohugou and Tianshan and low concentration in inland of the TP.

Table 3. Observed and Simulated Surface and Snow BC Concentration (μg m⁻³)

Name	Location	m asl (m)	Sample type	Time (LT)	Observation	Simulation
Lhasa	29.65°N, 91.03°E	3,640	Atmosphere	2013,3–12	0.46 ± 0.33	0.27	Li et al. (2016)
Qomalangma	28.36°N 86.95°E	4,276	Atmosphere	2009,8 to 2010,7	0.25	0.19	Cong et al. (2015)
Namco	30.77°N 90.98°E	4730	Atmosphere	2012,1–12	0.19	0.10	Wan et al. (2015)
Hulugou	38.23°N 99.48°E	3890	Atmosphere	2013,1–11	0.76 ± 0.49	0.57	Li et al. (2016)
Ranwu	29.32°N 96.96°E	4600	Atmosphere	2013,1–6	0.24 ± 0.19	0.09	Wang et al. (2016)
Beiluhe	34.85°N	4600	Atmosphere	2013,1–6	0.49 ± 0.2	0.24	Wang et al. (2016)
	92.94°E
Qinghai Lake	36.97°N 99.90°E	3300	Atmosphere	2012,1–12	0.84 ± 0.46	0.61	Zhao et al. (2015)
Lulang	29.46°N 94.44°E	3300	Atmosphere	2008,7 to 2009,8	0.5 ± 0.52	0.35	Zhao et al. (2017)
Hanle	32.78°N 78.96°E	4520	Atmosphere	2009,8 to 2010,7	0.07 ± 0.06	0.11	Babu et al. (2011)
Sinhagad	18.35°N	1450	Atmosphere	2010,1–12	3.8 ± 2.6	1.2	Safai et al. (2013)
Pune	73.75°E
Dhaka	23.76°N 90.39°E	7	Atmosphere	2010,3 to 2011,2	22.8 ± 6.8	4.8	Begum et al. (2012)
Southeast TP	29.35°N 97.02°E	5138	Snowpit	2015,6.10–16	0.02	0.27	Zhang et al. (2017)
Tienshan	43.1°N 86.82°E	4130	Surface Snow	2004,7–2005,7	0.4	0.15	Xu et al. (2012)
Laohugou	39.17°N	4260‐	Fresh Snow	2013,7.27–29	0.85	0.54	Li et al. (2016)
Glacier	96.17°E	5481		2013,8.4–6
Zhadang	30.48°N	5507‐	Fresh Snow	2013,7.15–16	0.05	0.17	Qu et al. (2014)
Glacier	90.65°E	5795		2013,8.24–26
East Rongbuk	28.04°N	6300	Snowpit	2004.10	0.02	0.28	Ming et al. (2009)
Glacier	86.95°E
Mera Glacier	27.72°N 86.88°E	5400	Snow/Ice	2009,2–5	0.18	0.43	Kaspari et al. (2014)

4 The Origin of Anthropogenic BC in the TP and its Transportation

In this study, we subtracted the surface BC concentration of the sensitivity experiments from those in the control experiment. The differences are supposed to reflect the various anthropogenic sources impacting BC concentration over other regions. By dividing the difference with the BC concentration of the control experiment, we determined the relative contribution rate.

Figures 8a and 8c show the mean BC differences induced by anthropogenic BC in the nonmonsoon and monsoon seasons with relative percentages shown in Figures 8b and 8d. In the nonmonsoon season, the contribution of anthropogenic emissions to total BC over the TP was significantly greater than in the monsoon season. The spatial distribution of the relative percentage of anthropogenic BC showed a gradually decreasing pattern from the southern to northern TP in the nonmonsoon season. In the monsoon season, the BC contribution from anthropogenic sources was less than in the surrounding regions, with the proportion being less than 60% over the TP and a maximum of 90% in South Asia, central Asia, and east China.

Figure 8

Variation in surface BC concentration (ug m⁻³) in (a) nonmonsoon and (c) monsoon due to anthropogenic emission and corresponding contribution percentage (%) in (b) nonmonsoon and (d) monsoon.

As shown in Figure 9, the anthropogenic BC over the TP mainly originated from South Asia in both nonmonsoon and monsoon seasons. In the southern TP, the anthropogenic sources from South Asia contributed to about 80% of the surface BC concentration in the nonmonsoon season (Figure 9g). Meanwhile, the values were only 50% in the monsoon season, which was partly attributed to precipitation scavenging in the monsoon season. In the nonmonsoon season, the contribution of anthropogenic BC originating from eastern China was less than 10% over the TP (Figure 9e) and South Asia was the dominant anthropogenic source of BC transported to the southeastern TP (Figure 9g). However, during monsoon season, anthropogenic source in eastern China resulted in surface BC concentration increasing by 0.21 μg m⁻³ (Figure 9b), which accounted for 10%–50% of the surface BC over the eastern TP (Figure 9d); for example, at locations in the northeastern TP such as the Qilian Mountains. Anthropogenic BC from central Asia influenced the surface BC over the western TP throughout the year with a proportional contribution of up to 30% (Figure 9l). In both seasons, anthropogenic sources in Northern China had little effect on surface BC over the TP (Figures 9o and 9p), with mean proportional contributions of around 1% (Figure S6). It is noteworthy that the definition of the SA, SS, SW, and SN regions was incompletely following the boundary of the TP, which might cause some deviations in calculating BC contribution from various anthropogenic sources. Due to the small BC concentration in the margin of the TP, we found that the deviations were relatively small by the comparison between Figures 8 and 9. For example, in the southwest slope of the Himalayas, divided into the TP in this study, the similar spatial pattern of anthropogenic BC contribution in Figures 8d and 9h indicated that anthropogenic BC in this region influenced little the simulated results.

Figure 9

Variation in surface BC concentration (ug m⁻³) in nonmonsoon and monsoon respectively due to different anthropogenic BC emissions around the TP ((a–d) Eastern China; (e–h) South Asia; (i–l) Central Asia; (m–p) Northern China).

The mean proportional contributions of adjacent anthropogenic sources to the surface BC concentration over the TP are shown in Figure S6. Averaged over the TP, the anthropogenic BC from South Asia accounted for 61.3% and 19.4% in the nonmonsoon and monsoon seasons, respectively. The mean contribution of anthropogenic BC originating from eastern China was much less than that from South Asia, and in the nonmonsoon season it was 2.3% lower than that in the monsoon season. The main reason is that the high‐BC areas (eastern and central China) are mostly downwind of the TP and only the South Asian monsoon can effectively bring BC emissions from China to the TP (Lu et al., 2012). As shown in Figure S7, these conclusions of our study presented good agreement with previous researches. However, the contribution proportions from our study and Zhang et al. (2015) were much less than that from Lu et al. (2012). It might be because the statistical analysis of back trajectory approach has limitations in determining contributions from distant sources to BC in the middle and upper troposphere.

The exogenous transmission of BC over the TP is mainly influenced by air flow. Figure 10 shows the variability in the anthropogenic BC concentration with wind magnitude and direction at different altitudes in the troposphere (850, 500, and 200 hPa). At 850 hPa, the northwest wind is divided into two branches in the nonmonsoon season (Figure 10a). One branch flowed east and converged with the south wind from the Bay of Bengal, which may have brought anthropogenic BC from northeast India to southwest China. As shown in Figure 9e, the south Asia anthropogenic BC influenced the surface BC in southwest China. During the monsoon season (Figure 10b), there were prevailing southwesterly winds from the Bay of Bengal to low‐latitude areas, and a cyclone was present in north India at 850 hPa. The South Asian monsoon brings abundant precipitation, which removes BC effectively from the atmosphere. For these reasons, the BC concentration was less in southwest China in the nonmonsoon season than in the monsoon season. At 500 hPa, the map of anthropogenic BC concentration indicated a larger affected area in the monsoon season (Figure 10c) than that in the nonmonsoon season (Figure 10d). Larger anthropogenic BC concentrations occurred over the southern TP. This was mainly because the intense updraft in the cyclone's center in the low atmosphere moved the surface BC upward in North India (Figure S8 shows the upward wind prevailing in the south slope of the TP), and then BC aerosols were transported to most parts of the TP by southwesterly winds. At 200 hPa, there was a prevailing westerly wind with very small anthropogenic BC aerosols in the nonmonsoon (Figure 10e) while anthropogenic BC aerosols were constantly transported by the updraft at the center of the cyclone in the midatmosphere, and the anthropogenic BC concentration was much higher in the upper atmosphere (Figure 10d).

Figure 10

Mean wind field and BC concentration at different pressures in (a, c, and e) nonmonsoon and (b, d, and f) monsoon.

The anthropogenic contribution to surface BC over the TP was less in the monsoon season than in the nonmonsoon season, although the anthropogenic BC concentration in the middle and upper atmosphere was greater in the monsoon season than the nonmoon season. This is mainly because the wet deposition of BC was much greater in the monsoon season than in the nonmonsoon season over the TP (Ji et al., 2015), which had little effect on the surface BC.

5 Anthropogenic BC Impacts on Radiative Forcing and Temperature

BC is an absorbing aerosol that can heat the atmosphere by reducing albedo and reflected solar radiation, leading to positive radiative forcing (RF) in the atmosphere and negative values at the near surface (Lau et al., 2006). In this study, the differences of the SA simulation minus the CT were considered to be the effects induced by the anthropogenic BC from source regions.

As shown in Figures 11a and 11c, for both the nonmonsoon and the monsoon seasons, atmospheric anthropogenic BC aerosol induced negative radiative forcing at the near surface in most regions except in part of the northern TP, similar to Ji et al.'s (2015) study. The atmospheric forcings of anthropogenic BC aerosol were in the range of −2 to 2 W m⁻² over the TP. The largest radiative forcing occurred in the Sichuan Basin, eastern China, with absolute values exceeding 12 W m⁻² in the monsoon season.

Figure 11

Variation in (a, c) surface radiative forcing (W m⁻²), (b, d) surface temperature (°C) due to anthropogenic BC in nonmonsoon (Figures 11a and 11b) and monsoon (Figures 11c and 11d).

Owing to the negative radiative forcing, surface temperature decreased in most regions during the nonmonsoon and the monsoon seasons (Figures 11b and 11d). In the non‐monsoon season, the greatest decrease (up to 0.24°C) in surface temperature occurred in part of eastern China, although a similar decrease occurred in the Sichuan Basin during monsoon season. The change in surface air temperature induced by atmospheric anthropogenic BC aerosol was very small over the TP with values in the range of −0.08 to 0.08°C. Moreover, the temperature distribution had a warming trend over the northern TP in both the nonmonsoon and the monsoon season which was consistent with the spatial patterns of surface radiative forcing.

Further, we analyzed the impact of anthropogenic BC aerosol on temperature at different altitudes in the troposphere (850, 500, and 200 hPa). As shown in Figure 12, anthropogenic BC aerosol produced a warming effect at each altitude because BC aerosol can absorb solar radiation and heat the atmosphere. At 850 hPa (Figures 12a and 12b), regions with a significant increase in temperature were consistent with the areas that have high BC concentrations in Figures 10a and 10b. For example, around 25°N, 90°E in Figure 10, the anthropogenic BC concentration in the nonmonsoon season was greater than that in monsoon season; therefore, during the nonmonsoon season the warming effect was stronger. At 500 and 200 hPa there was a strong corresponding relationship between temperature variation and anthropogenic BC concentration. In these two layers, the warming effect during the monsoon season was much more significant than that during the nonmonsoon season due to the intense updraft flux in the low‐ and middle‐troposphere atmospheric layers, which moved BC aerosols upward and resulted in larger anthropogenic BC concentrations in the middle‐ and upper‐troposphere atmospheric layers.

Figure 12

Variation in different pressures (200 hPa, 500 hPa, and 850 hPa) due to anthropogenic BC aerosol in (a, c, and e) nonmonsoon and (b, d, and f) monsoon.

6 Conclusions

In this study, the WRF‐Chem model was used to simulate the atmospheric BC concentration and the resulting climatic effects over the TP for the entire year of 2013. First, we evaluated the model's performance against available observations which represented the first attempt to evaluate WRF‐Chem for modeling seasonal atmospheric BC over the TP. Then, we quantitatively assessed the anthropogenic emissions impact on the distribution, transportation, and climatic effects of atmospheric BC over the TP with a set of sensitivity experiments.

It was found that WRF‐Chem can capture the main spatial and temporal features of meteorological elements over the TP. The climate‐chemistry model also performed reasonably well in modeling the AOD and near‐surface BC concentration. The primary shortcoming of the simulation, however, was the underestimation of the AOD (MB = −0.04) and BC concentration (MB = −1.2) compared to the in situ observations. Model errors arose partly because the model grid was a point source at 25 km, which represents a regional average, while the observation site was strongly influenced by local effects such as weather conditions and terrain.

The simulated surface BC concentration from anthropogenic sources showed seasonal differences. In the monsoon season, the contribution of anthropogenic BC was less than that in the nonmonsoon season. The majority of anthropogenic BC over the TP was transported from South Asia which contributed to 40%–80% (mean of 61.3%)of surface BC in the nonmonsoon season, and 10%–50% (mean of 19.4%) in the monsoon season. In the nonmonsoon season, the proportional contribution of eastern China as a source of BC on the TP was less than 10%, with a mean of 6.2%, whereas the eastern China anthropogenic source accounted for an increase in surface BC concentration over the eastern TP up to 0.21 ug m⁻³, that is, 10%–50% in surface BC. The central Asian anthropogenic BC mainly influenced the surface BC over the western TP, while the northern China anthropogenic BC had little impact on the TP surface BC, with a mean proportional contribution of around 1%.

The exogenous transmission of BC over the TP is mainly influenced by large‐scale atmospheric circulation. In the nonmonsoon season, westerly winds prevailed and transported BC from central Asia and north India to the western TP. In the monsoon season, anthropogenic BC aerosol was transported to the middle to upper troposphere over the Indo‐Gangetic Plain and crossed the Himalayas via southwesterly winds. Anthropogenic BC caused negative radiative forcing at the near surface in most regions, except in part of the northern TP. Owing to this negative radiative forcing, surface temperature decreased in most regions in both the nonmonsoon and monsoon seasons. Anthropogenic BC aerosol produced a heating effect at different altitudes in the atmosphere because of its ability to absorb solar radiation. Close correspondence between temperature variations and anthropogenic BC concentrations was identified. At 500 and 200 hPa, the warming effect in the monsoon was more significant than in the nonmonsoon season because surface BC was transported to the middle to upper troposphere.

Acknowledgments

This study was supported by the National Natural Science Foundation of China (41421061, 41630754, 91644225, and 41301061), the Key Research Program of the Chinese Academy of Sciences (KJZD‐EW‐G03–04) and the Open Program (SKLCS‐OP‐2017‐02) from State Key Laboratory of Cryospheric Science, Northwest Institute of Eco‐Environment and Resources, Chinese Academy of Sciences. The authors thank the anonymous reviewers for their constructive comments and suggestions, which helped to improve the quality of this paper. We acknowledge the China Meteorological Data Network for providing the meteorological data. AOD data are acquired from http://aeronet.gsfc.nasa.gov/cgi‐bin/type_piece_of_map_aod_v3/. Thanks to Z. Cong ([e-mail]) for offering the AOD data at Namco. The WRF‐Chem simulated results in this study are available by contacting the corresponding author ([e-mail]).

Erratum

In the originally published version of this article Figure 8 was published incorrectly. This error has since been corrected, and this version may be considered authoritative version of record.