Impacts of atmospheric transport and biomass burning on the interannual variation in black carbon aerosols over the Tibetan Plateau

Atmospheric black carbon (BC) in the Tibetan Plateau (TP) can largely impact regional and global climate. Still, studies on the interannual variation in atmospheric BC over the TP and associated 20 variation in BC sources and controlling factors are rather limited. In this study, we characterize the variations in atmospheric BC over the TP surface layer through analysis of 20-year (1995-2014) simulations from a global chemical transport model, GEOS-Chem. The results show that, of all areas in https://doi.org/10.5194/acp-2020-299 Preprint. Discussion started: 15 May 2020 c © Author(s) 2020. CC BY 4.0 License.

studies explored the mechanisms of BC transport to the TP, large uncertainties remain in the quantified fractional contributions of BC transport from different source regions to the TP (Yang et al., 2018).
More importantly, how BC transport to the TP varies interannually and what are underlying mechanisms for the variation are unclear. Therefore, it is necessary to examine how seasonal BC transport to the TP varies from year to year and whether there is a connection between the Asian monsoon and the 100 interannual variation in BC transport to the TP.
Observations and simulations showed previously that anthropogenic and fire emissions are major sources of atmospheric BC in the TP (Lu et al., 2012;Zhang et al., 2015). Zhang et al. (2015) estimated that biomass burning together with biofuel emissions can contribute to around half of the annual mean 105 BC column burden over the TP. Engling et al. (2011) reported that BC emissions from fire events in Southeast Asia in spring could probably increase the BC concentrations over a mountain site in the southeastern part of the TP. Putero et al. (2014) suggested that over half of the high BC episodes in the southern Himalayas were likely affected by the fire events in South Asia. However, these studies demonstrated the influences of biomass burning in a relative short term or during some fire events, few 110 investigated the influences in a long term over a decade. The influence of biomass burning on the interannual variation in atmospheric BC over the TP warrants an in-depth study.
In this study, we aim to assess the impacts of atmospheric transport and biomass burning on surface BC concentrations over the TP, especially on the interannual variation in BC during 1995BC during -2014 To estimate BC transport from different source regions to the TP, we adopt a numerical approach based on a global chemical transport model, GEOS-Chem (Bey et al., 2001), and a trajectory model, the Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT) (Draxler and Hess, 1998;Stein et al., 2015). In the following, the method and models are described in section 2. Section 3 https://doi.org/10.5194/acp-2020-299 Preprint. Discussion started: 15 May 2020 c Author(s) 2020. CC BY 4.0 License. et al. (2014b). The 13 sites were grouped into urban, rural, and remote sites (He et al., 2014b). The observational data are available for 2006 at 9 of the 13 sites and available at the other sites for different periods, i.e., 1999-2000, 2004-2005 and 2008-2009. BC observations at a remote site during 2015-2017 170 from another study (Chen et al., 2018) were also used.
The annual mean surface BC concentrations from GEOS-Chem and observations are compared in Table 1. The observed surface BC concentrations are below 2 µg m -3 at remote sites, about 2-5 µg m -3 at rural sites, and as high as 5 µg m -3 at urban sites (He et al., 2014b). Compared with the observations, 175 GEOS-Chem performs well at the remote sites, moderately at the rural sites, and poorly at the urban sites ( Table 1). The simulations substantially underestimate surface BC concentrations at the urban sites, likely due to the coarse horizontal resolution in the model that dilutes the intensity of local emissions in a model grid. Taking the rural and remote sites only (Figure 2a), we found a high consistency between the annual mean simulations and observations, with a significant correlation coefficient of 0.99. The 180 comparison suggests that GEOS-Chem can generally capture the spatial variation in surface BC concentrations over the TP. Moreover, the seasonality of simulated surface BC concentrations was evaluated at three sites (Figures 2b-2d). GEOS-Chem simulates low BC concentrations in summer and high BC concentrations in winter and spring at the sites. The amplitude of the seasonal variation in the simulations is weaker than that in the observations. Moorthy et al. (2013) found that simulated surface 185 BC concentrations by the Goddard Global Ozone Chemistry Aerosol Radiation and Transport (GOCART) model in winter were lower than the observed ones at a TP site and they attributed this to the biases in the atmospheric boundary layer parameterization scheme. Wintertime surface BC concentrations were also underestimated by the Community Atmosphere Model version 5 (CAM5,

Meteorological and fire data
The meteorological data used in this study are the NCEP/NCAR (National Centers for Environmental to verify the biomass burning emissions in the model. ATSR is onboard the Second European Remote-Sensing Satellite (ERS-2). The spatial resolution of the data is 1 km, and the sensor achieves a global coverage every three days. The ATSR satellite data with the period of 1997-2011 were gridded to the GFED3 grids with a resolution of 0.5 o × 0.5 o in longitude and latitude.

Transport estimation
Combining GEOS-Chem simulations and HYSPLIT (version 4, http://www.arl.noaa.gov/HYSPLIT_info.php, Draxler and Hess, 1998;Stein et al., 2015) trajectories, we estimated the contributions of different source regions in the world to surface BC in the TP during 1995-2014. HYSPLIT is an atmospheric transport and dispersion model (Fleming et al., 2012), 210 developed by the Air Resources Laboratory of the National Oceanic and Atmospheric Administration (NOAA). Meteorological inputs to HYSPLIT are the NCEP/NCAR reanalysis at a resolution of 2.5 o latitude × 2.5 o longitude. We evenly divided the TP into 70 GEOS-Chem grids. Considering that the average lifetime of atmospheric BC is about a week, we simulated 7-day backward trajectories originated from each of the 70 grids. The trajectories were initialized four times a day (00, 06, 12 and 18 215 UTC) during 1995-2014. The starting altitude for the trajectories is 100 m above ground which is within the typical planetary boundary layer in the TP (Ram et al., 2010b). We divided the world into six regions for D days (D=7 in this study). To make the estimation stable, the amount of BC transported to a TP surface grid on a day is assumed to be a mean of the BC transport along the backward trajectories originated from that grid in the past D days, i.e.,

= ∑ =1
(1) where BCd is the amount of BC that are transported to that TP surface grid along the backward 230 trajectory on a previous day d (d=1, 2, ... D).

Equation
(1) provides a way to estimate the amount of BC that is transported to the TP from any model grid outside the TP during a period of interest. For a grid , , , the total amount (Ci,j,k) of BC transported from , , to the TP is estimated by 235 where i, j, k are indices for the model grid in longitude, latitude, and altitude coordinates, respectively. n is an index for the number of trajectories. N is the total number of trajectories that have passed through the grid gi,j,k during the period of interest, for example, in a month. c is the daily BC concentrations at gi,j,k when trajectory n passing gi,j,k, and v is the volume of gi,j,k. M is the number of trajectories in a day 240 (M=4 in this study). Therefore, the total amount of BC transported to the TP (Ti,j) from the entire tropospheric column above a surface grid gi,j,0 in a source region during the period of interest is assessed by where K is the number of model layers in the troposphere. 245 Finally, the amount of BC transported from a source region to the TP surface can be summed up and the fractional contributions of different source regions to surface BC in the TP can be quantified. Over the western TP, BC concentrations are the 3 rd lowest among the five subregions, with a seasonality of high BC in winter and spring and low BC in summer and autumn (Figure 6d). The higher values in spring and winter agree with the BC measurements at sites in the western Himalayas (Nair et al., 2013). BC transport from South Asia contributes to 93% of surface BC in winter and 76% in 305 summer ( Figure 7d). Such seasonality with winter high and summer low in the fractional contribution of South Asia to surface BC over the western TP were also suggested by Zhang et al. (2015).
In the northern TP, BC concentrations are the 2 nd lowest among the five subregions, which are at contributes to 72% and 91% surface BC in this subregion. In contract, East Asia contributes to 58% 325 surface BC there in summer. To further examine influence of biomass burning in spring, we integrally analyzed data from ATSR satellite fire counts, GFED3 fire emissions, and the GEOS-Chem simulations. Both ATSR and GFED3 data show that fires occur frequently over the Indo-Gangetic Plain, central India, and Southeast Asia (Figures 1 and 9). Fire activities in Asia are well described in the GFED3 inventory that is used in the 345 GEOS-Chem simulations (Figures 9b and 9c). We found the interannual variation in BC anomalies in the TP from CTRL simulation is significantly correlated to the fire counts in the Indo-Gangetic Plain (r=0.76, p<0.05), and central India (r=0.67, p<0.05). The correlation was insignificant for Southeast Asia (r=0.19,p>0.05). In spring of 1999, extreme fire activities occurred in the Indo-Gangetic Plain and central India (Figure 9b). Driven by the favourable atmospheric circulation, the strong BC emissions 350 from the extremely active fires greatly enhanced surface BC concentrations in the TP (Figure 9d). In the CTRL simulation, positive BC anomalies appear over the entire TP, with a regional mean of 0.15 µg m -3 or 31% relative to the 1995-2014 climatology (Figure 8a). Additionally, in winter, biomass burning was extremely strong in 1998 (Figure 8d). The extremely active fires enhanced the regional mean surface BC concentrations in the TP by 0.02 µg m -3 or 5% relative to the climatology.

Influences of biomass burning on surface black carbon over the Tibetan Plateau
in summer, autumn, and winter, suggesting that meteorology plays an important role in modulating the interannual variation in surface BC in the TP. Such a role will be explored in sections 4.2 and 4.3 from the influences of the Asian monsoon on BC transport to the TP in summer and winter, respectively. East Asia to the TP. Furthermore, the EASMI also correlates negatively with the zonal wind at 850 hPa over central China (Figure 10b). When the EASM is stronger, the zonal wind in the monsoon circulation weakens over this region (Yang et al., 2014;Han et al., 2019), suggesting that westward winds may 385 occur more or with higher speed. Therefore, BC transport to the TP from central China is enhanced (Figure 10c), as a significantly positive correlation is found between the strength of the EASM and BC transport from central China to the TP surface (r=0.49, p<0.05) and between the strength of the EASM and BC transport from central China to the eastern TP surface (r=0.48, p<0.05). This is further confirmed by the differences in BC transport to the TP surface between summers with strong and weak 390 EASM (Figure 10d).
How the SASM impacts the BC transport from South Asia over the TP surface in summer is also examined ( Figure 11). Serving as a heat source in the Asian summer monsoon system, the TP promotes strong convection and modulates the meridional circulation (Xu et al., 2014). Driven by the meridional 395 circulation, BC in South Asia can be transported northward and upward to the TP. BC transport from northeastern South Asia to the TP accounts for 30% of the total BC transport from South Asia ( Figure   4b). Interannually, BC transport from northeastern South Asia is significantly correlated with the meridional wind at 500 hPa (r=0.65, p<0.05, Figure 11a), which is also closely correlated to the strength of the SASM (Figure 11b). In strong SASM years, an anomalous cyclone locates over the northern 400 South Asia at 500 hPa and correspondingly the meridional wind over the northeastern South Asia is increased (Figure 11d). This well explains why the interannual variation in BC transport from northeastern South Asia correlates positively with the strength of the SASM (r=0.55, p<0.05 for the TP, r=0.56, p<0.05 for the STP, Figure 11c). Among all source regions, the differences in BC transport from northeastern South Asia to the TP is largest between summers with strong and weak EASM (Figure 11d).

Plateau
The Asian winter monsoon is a predominant climate feature in Asia and an important modulator of the distribution and transport of air pollutants (Mao et al., 2017;Zhu et al., 2017). However, the impact of 410 the Asian winter monsoon on the interannual variation in BC transport to the TP scantly studied. In this section, we assess such impact with two climate indices. We measure the intensity of East Asian winter monsoon (EAWM) by an index defined by Jhun and Lee (2004). The EAWM index (EAWMI) represents the EAWM intensity by the meridional wind shear associated with the jet stream in the upper troposphere. It can be calculated by the difference in the regional averaged zonal wind speed at 300 hPa 415 between the areas 27.5-37.5 o N, 110-170 o E and 50-60 o N, 80-140 o E. Using the EAWMI, it is found that the EAWM is closely correlated with the interannual variation in pollution transport over East Asia (Li et al., 2016;Han et al., 2019). Furthermore, the Siberian High is a key component of the EAWM system (Wu and Wang, 2002) and its strength can be described using an index defined by Wu and Wang (2002). This Siberian High index (SHI) can be calculated from the regional mean sea level pressure over the 420 area of the Siberian High (40-60 o N, 80-120 o E). The EAWMI and SHI are highly correlated (r=0.72, p<0.05). Interannually, from 1995 to 2014, biomass burning can explain over 75% of the variation in springtime surface BC concentrations over the TP if biomass burning and meteorology are both considered in GEOS-Chem simulations. Indeed, springtime surface BC in the TP is significantly 470 correlated to the total number of fire counts over the Indo-Gangetic Plain in South Asia (r=0.76, p<0.05), according to ATSR satellite data. In the spring of 1999, the extremely strong biomass burning in South Asia largely elevated surface BC concentrations (0.15 µg m -3 or 31% relative to the climatology) over the TP. We noticed that the strong biomass burning in South Asia in the winter of 1998 also enhanced BC concentrations over the TP. 475 The interannual variation in surface BC over the TP are greatly influenced by meteorology. Specifically, the Asian monsoon system alters the long-range transport of BC to the TP by modulating the atmospheric circulation. In summer, when the EASM is stronger, the more frequent or stronger westward wind in the lower troposphere can enhance BC transport from central China to the TP. When 480 the SASM is stronger, the increased meridional wind over the northeastern South Asia in the middle troposphere can enhance BC transport from northeastern South Asia to the TP. In winter, when the EAWM is stronger, the reduced zonal wind in the lower troposphere tends to increase BC transport from central China to the TP. A stronger Siberian High can enhance the zonal wind in the middle troposphere over the TP and consequently increases BC transport from northern South Asia to the TP. 485 The findings in this study provide an enhanced understanding of the long-range transport of BC to the TP. We comprehensively assessed the BC transport from worldwide source regions to the TP. Our results reveal the source regions of surface BC over the entire TP in the four seasons, which was investigated by limited studies (Zhang et al., 2015). The influences of South Asia and East Asia on the 490 TP were noticed by previous studies. Most of them were focused on limited locations (Cao et al., 2011;Engling et al., 2011;Chen et al., 2018) or in one or few seasons (Zhao et al., 2017;Wang et al., 2018).
Here, we further quantified the influence of South Asia and East Asia over the entire TP in the four seasons, in terms of both fractional contribution and affected areas in the TP. Moreover, we identified three key areas within South Asia and East Asia and found that the contribution of BC from there to 495 surface BC in the TP is highest among South Asia and East Asia.
Biomass burning is an important source of atmospheric BC in the TP (Zhang et al., 2015). It was observed that BC emissions from biomass burning in South Asia could be transported to the TP by the atmospheric circulation (Cong et al., 2015), and resulted in high BC episodes in the southern TP 500 (Engling et al., 2011;Putero et al., 2014). Only limited numerical studies explored connections between https://doi.org/10.5194/acp-2020-299 Preprint. Discussion started: 15 May 2020 c Author(s) 2020. CC BY 4.0 License. biomass burning and surface BC in the TP over a long-term period (Mao and Liao, 2016). Here, we demonstrated that biomass burning is an important driver of the interannual variation in surface BC over the TP in spring. In particular, we found that there were extremely strong fire activities over Indo-Gangetic Plain, central India, and Southeast Asia from 1998 winter to 1999 spring that largely enhanced 505 surface BC concentrations over the entire TP. This extreme anomaly in fire activities and associated influence on BC over the TP may have not been fully documented.
We found that the Asian monsoon system can significantly modulate the interannual variation in BC transport from South Asia and East Asia to the TP. Asian monsoon can influence the atmospheric 510 circulation over the TP and its surroundings (Xu et al., 2014;Han et al., 2019). For summer, previous studies mainly focused on the transport pathway build by the SASM, which can meridionally transport BC from South Asia to the TP (Zhao et al., 2017;Kang et al., 2019). In this study, we further revealed that the EASM can modulate the westward transport of BC from central China to the TP. In winter, the Asian monsoon system also significantly influences the BC transport from northern South Asia and 515 central China to the TP. These results can shed some light on the transport mechanisms of other atmospheric species to the TP, such as water vapour.