Vertical transport mechanisms of black carbon over East Asia in spring during the A‐FORCE aircraft campaign
Abstract
[1] Mechanisms of vertical transport of black carbon (BC) aerosols and their three‐dimensional transport pathways over East Asia in spring were examined through numerical simulations for the Aerosol Radiative Forcing in East Asia (A‐FORCE) aircraft campaign in March–April 2009 using a modified version of the Community Multiscale Air Quality (CMAQ) modeling system. The simulations reproduced the spatial distributions of mass concentration of BC and its transport efficiency observed by the A‐FORCE campaign reasonably well, including its vertical and latitudinal gradients and dependency on precipitation amount that air parcels experienced during the transport. During the A‐FORCE period, two types of pronounced upward BC mass fluxes from the planetary boundary layer (PBL) to the free troposphere were found over northeastern and inland‐southern China. Over northeastern China, cyclones with modest precipitation were the primary uplifting mechanism of BC. Over inland‐southern China, both cumulus convection and orographic uplifting along the slopes of the Tibetan Plateau played important roles in the upward transport of BC, despite its efficient wet deposition due to a large amount of precipitation supported by an abundant moisture supply by the low‐level southerlies. In addition to the midlatitude (35–45°N) eastward outflow within the PBL (21% BC removal by precipitation during transport), the uplifting of BC over northeastern and inland‐southern China and the subsequent BC transport by the midlatitude lower tropospheric (50% BC removal) and subtropical (25–35°N) midtropospheric westerlies (67% BC removal), respectively, provided the major transport pathways for BC export from continental East Asia to the Pacific.
1 Introduction
[2] Black carbon (BC) particles efficiently absorb solar radiation and lead to heating of the atmosphere. BC and other coemitted aerosols can also modify cloud properties and influence the precipitation efficiency by acting as cloud condensation nuclei (CCN) and ice nuclei. Therefore, BC has been recognized as one of the most important aerosol types affecting climate forcing [e.g., Hansen et al., 1997; Ackerman et al., 2000; Ramanathan et al., 2001; Jacobson, 2001; Menon et al., 2002; Ramanathan and Carmichael, 2008; Ramana et al., 2010; Bond et al., 2013]. BC particles are emitted into the atmosphere by the incomplete combustion of fossil fuels, biofuels, and biomass. BC particles emitted over source regions can be vertically transported from the planetary boundary layer (PBL) to the free troposphere (FT), followed by long‐range transport to receptor regions (e.g., the Arctic). When transported BC particles are deposited on snow or sea ice surfaces, they cause heating of the surface due to solar absorption, resulting in the melting of additional snow or sea ice and reducing the reflectivity of the surface. These effects have a substantial impact on regional‐to‐global‐scale climate systems [e.g., Hansen and Nazarenko, 2004; Flanner et al., 2007, 2009]. In addition to the long‐range transport of BC particles, the vertical distribution of BC itself can significantly affect the aerosol radiative effects. For example, Hansen et al. [2005] showed that direct radiative forcing depends strongly on the vertical profile of BC particles, and Haywood and Shine [1997] showed that direct radiative forcing is largely strengthened when BC particles exist above or within clouds. Consequently, an understanding of the vertical transport processes of BC from the PBL to the FT is critically important because these processes directly control the global‐ and regional‐scale spatial distributions of BC (both horizontally and vertically) and the amount of BC transported from source regions to receptor regions. However, there remain large uncertainties in the calculations of the vertical transport of BC in the current three‐dimensional (3‐D) models [e.g., Textor et al., 2006; Koffi et al., 2012], which may cause large differences in the vertical profiles of BC concentration between model simulations and observations [e.g., Koch et al., 2009; Bauer et al., 2010]. A multimodel comparison of global aerosol models by Samset et al. [2013] showed that at least 20% of the uncertainty in the model‐simulated direct radiative forcing of BC was due to diversity in the simulated vertical profile of BC.
[3] Several recent studies have discussed the vertical transport of BC particles using high‐altitude observations in the FT and the upper troposphere/lower stratosphere (UTLS) and/or model calculations in terms of the long‐range transport of BC from source regions (e.g., Asia) to receptor regions (e.g., the Arctic and North America) [e.g., Park et al., 2005; Koch and Hansen, 2005; Stohl, 2006; Koch et al., 2007, 2009; Hadley et al., 2007, 2010; Schwarz et al., 2010; Spackman et al., 2010, 2011; Brock et al., 2011; Liu et al., 2011; Matsui et al., 2011; Bourgeois and Bey, 2011]. During the NASA Arctic Research of the Composition of the Troposphere from Aircraft and Satellites (ARCTAS) mission in the spring and summer of 2008 [Jacob et al., 2010], BC particles in air parcels that originated from anthropogenic sources in East Asia experienced strong uplift associated with warm conveyor belts (WCBs) and were then transported to the Arctic troposphere, although most BC particles were removed from the atmosphere by precipitation during transport [Matsui et al., 2011]. Spackman et al. [2011] used the BC observations from the Tropical Composition, Cloud, and Climate Coupling (TC4) experiment in August 2007 [Toon et al., 2010] to show that BC particles in air parcels that originated from anthropogenic sources in Asia were likely transported to the tropical tropopause layer over Central America through cumulus convection associated with the Asian monsoon. However, these studies focused on the receptor regions and might have limitations regarding the quantitative understanding of the vertical transport mechanisms of BC because uplifting processes (e.g., cyclones (WCBs) and cumulus convection) near the source regions are considered to largely control the amount of BC transported to the FT on regional to hemispheric scales. To improve our understanding of the vertical transport of BC, further studies covering large source regions (e.g., East Asia) are needed.
[4] According to current emission inventories, East Asia is the largest source of anthropogenic pollutants, including BC [Streets et al., 2003; Bond et al., 2004; Zhang et al., 2009]. A number of previous studies have examined the vertical transport mechanisms of pollutants over East Asia, mainly focusing on carbon monoxide (CO) [e.g., Bey et al., 2001; Liu et al., 2003; Koike et al., 2003; Miyazaki et al., 2003; Liang et al., 2004; Oshima et al., 2004; Dickerson et al., 2007; Ding et al., 2009; Lin et al., 2010]. Some of these studies have focused on spring because the Asian influence on CO levels in the North Pacific troposphere maximizes during spring and minimizes during summer [e.g., Liu et al., 2003; Liang et al., 2004]. They have shown that episodic lifting from the PBL to the FT over China due to WCBs and cumulus convections and subsequent westerly transport in the FT were the principal processes responsible for the export of pollution from both anthropogenic sources and biomass burning in Asia in spring. Liang et al. [2004] have also described seasonal variations in the transport mechanisms of CO, namely that transport in the WCB of midlatitude cyclones is the dominant export mechanism throughout the year and export induced by deep convection becomes important during summer and fall. However, the quantitative understanding of the vertical transport mechanisms of BC over these regions is still limited because of lack of observations of BC in the FT over East Asia, although several previous studies have examined the transport processes of BC using surface observations [e.g., Uno et al., 2003; Verma et al., 2011].
[5] The Aerosol Radiative Forcing in East Asia (A‐FORCE) aircraft campaign was conducted over the Yellow Sea, the East China Sea, and the western Pacific from 18 March to 25 April 2009 (Figure 1) [Oshima et al., 2012]. During the A‐FORCE campaign, 120 vertical profiles of BC were obtained using a single‐particle soot photometer (SP2) at 0–9 km in altitude [Moteki and Kondo, 2010; Moteki et al., 2012]. The SP2 instrument measured the number concentration of BC particles in the size range of 75–850 nm volume equivalent diameter based on the laser‐induced incandescence technique. Total BC mass concentration was derived by integrating the measured BC particles over the observed size range, with an accuracy of approximately 10%. The CO concentrations, cloud microphysical properties, and aerosol number concentrations were also measured during this campaign [Oshima et al., 2012; Koike et al., 2012; Takegawa et al., 2013]. More detailed descriptions of the A‐FORCE aircraft measurements and their uncertainties are given by Oshima et al. [2012]. In our previous study, the wet removal of BC in Asian outflow was examined based on the analysis of the A‐FORCE observation data set [Oshima et al., 2012]. We found that the transport efficiency of BC (TEBC, namely the fraction of BC particles not removed from the atmosphere during transport; this definition is presented in section 2) in air parcels sampled above 2 km in altitude decreased primarily with increases in the amount of precipitation that the air parcels had experienced during vertical transport. The TEBC values for the sampled air parcels originating from northern China (north of 33°N) were found to be systematically greater than those from southern China (south of 33°N) due to the latitudinal difference in precipitation during springtime in East Asia. In light of our previous study, a combination of A‐FORCE data and 3‐D model calculations can provide opportunities to improve our understanding of the vertical transport mechanisms of BC over East Asia by validating the vertical profiles and wet removal processes of BC given by the model calculations. In particular, a comparison of the TEBC values can provide good constraints for the vertical profiles and removal of BC in the 3‐D models. A thorough understanding of the vertical transport mechanisms of BC will contribute to the reduction of uncertainties in the representation of BC transport and removal in current global aerosol models.
[6] The major objective of this study is to understand the vertical transport mechanisms of BC particles and their transport pathways over East Asia in spring using the results from regional‐scale 3‐D chemical transport model calculations. We utilized a modified version of the Community Multiscale Air Quality (CMAQ) model in which the aerosol wet deposition process is improved from the original version, and the model was applied to the A‐FORCE aircraft campaign (section 2). By evaluating the model results using the A‐FORCE observations (section 3), we identified the major uplifting processes for BC particles and the major transport pathways for the export of BC from East Asia (section 4). In particular, we conducted analyses of the mass flux of BC and meteorological fields that play an important role in the transport and wet removal of BC (section 4). In addition, we quantified the contributions of BC emitted from various Asian regions to the Asian outflow over the western Pacific (section 4).
2 Model Simulations
[7] The CMAQ model (version 4.6) [Byun and Ching, 1999; Binkowski and Roselle, 2003], driven by the Weather Research and Forecasting (WRF) model (version 2.2) [Skamarock et al., 2005], was used to calculate the spatial distributions of mass concentration of BC and its regional‐scale transport. The WRF‐CMAQ simulation was performed over all of East Asia between 1 February 2008 and 31 May 2009 [Kondo et al., 2011a], and we focused on the A‐FORCE period (20 March to 30 April 2009) in this study. In the WRF model simulation, we used four‐dimensional data assimilation (FDDA nudging) in which the circulation and thermal fields in the model were nudged toward the 6‐hourly National Centers for Environmental Prediction (NCEP) Final (FNL) operational analysis data, which are available globally on a regular grid with a resolution of 1° in both latitude and longitude at the 21 standard pressure levels (1000 ~ 100 hPa). The SAPRC‐99 gas phase mechanism [Carter, 2000] and aerosol module AERO3 were used in the CMAQ calculations. The horizontal grid system of the WRF‐CMAQ simulations consisted of 117 × 69 grid boxes, each of which occupied an area of 81 × 81 km2. The model employed 21 vertical layers up to the 100 hPa level, with finer intervals within the PBL (11 layers below a sigma level of 0.8). The Asian anthropogenic emission inventory with a grid resolution of 0.5° × 0.5° in latitude and longitude, developed for the year 2006 [Zhang et al., 2009], was used in the simulations. The biomass burning emission inventory developed for the year 2000 [Streets et al., 2003], with a grid resolution of 1° × 1° in latitude and longitude, was also used in the simulations. The annual BC emissions from biomass burning in Asia for the year 2009 estimated by Matsui et al. [2013a] are approximately 90% of that used in this study, suggesting that the use of the biomass burning inventory for the year 2000 would have little influence on the conclusions of this study. The total BC emissions and the entire area of the model horizontal grid boxes used for the CMAQ calculations are shown in Figure 1. The initial and boundary conditions for the BC mass concentrations in the model domain were set to zero. Hourly outputs were obtained from the WRF and CMAQ simulations.
[8] We modified the calculation method for the wet deposition of accumulation‐mode aerosols in the CMAQ model. A description of this modification and its evaluation have been presented in part by Kondo et al. [2011a] and are presented in detail in Appendix A. Here we briefly describe this modification. The original CMAQ model neither distinguishes the rainout (in‐cloud scavenging) and washout (below‐cloud scavenging) of aerosols nor takes into account these removal processes properly. The modified version of the CMAQ model can distinguish between rainout and washout processes and ignores the washout process for accumulation‐mode aerosols because particles of this size are much less efficiently removed by the collision of precipitating raindrops below clouds [Seinfeld and Pandis, 2006]. Several previous studies have shown that below‐cloud scavenging is negligible for accumulation‐mode aerosol particles when compared with in‐cloud scavenging rates for low and moderate rainfall rates (0.1–10 mm h−1) [e.g., Andronache, 2003; Henzing et al., 2006; Berthet et al., 2010], supporting our treatment of the wet deposition process. While the CMAQ model explicitly treats aerosol microphysics, such as condensation and coagulation (i.e., aging processes of BC), the CMAQ model assumes all aerosol species in each of the Aitken, accumulation, and coarse modes are internally mixed and the accumulation‐mode aerosols (including BC) are hydrophilic [Byun and Ching, 1999; Binkowski and Roselle, 2003]. Uncertainties caused by the hydrophilic assumption are discussed in Appendix B.
(1)| Simulation Name | Aerosol Wet Deposition | Source Region |
|---|---|---|
| CMAQ‐Modified (baseline) | Modified scheme | East Asia |
| CMAQ‐Originalbb
The CMAQ‐Original simulation is discussed in more detail in Appendix A. |
Original scheme | East Asia |
| CMAQ‐NoWetDep | None | East Asia |
| CMAQ‐China | Modified scheme | NC + SWC + SEC |
| CMAQ‐NorthChina | Modified scheme | NC |
| CMAQ‐SouthWestChina | Modified scheme | SWC |
| CMAQ‐Japan | Modified scheme | JP |
| CMAQ‐Korea | Modified scheme | KR |
| CMAQ‐SouthAsia | Modified scheme | SA |
| CMAQ‐SouthEastAsia | Modified scheme | SEA |
| CMAQ‐Anthropogenic | Modified scheme | Anthropogenic only for East Asia |
- a The individual source regions are shown in Figure 1. Emissions from anthropogenic sources and biomass burning are included in all simulations except the CMAQ‐Anthropogenic simulation.
- b The CMAQ‐Original simulation is discussed in more detail in Appendix A.
[10] To evaluate the individual contributions of various BC source regions in East Asia (see gray boxes in Figure 1) to the export of BC, sensitivity simulations with BC emissions from the individual source regions were performed using the modified aerosol wet deposition scheme (Table 1). We also performed a simulation that included only anthropogenic BC emissions from East Asia (i.e., excluding the biomass burning emissions) to estimate the relative contribution of biomass burning (Table 1). Although some of the model calculations (the baseline and some sensitivity simulations) used in this study were briefly presented by Kondo et al. [2011a] and Verma et al. [2011] for the entire calculation period (from February 2008 to May 2009), the evaluations of the CMAQ‐Modified (baseline) simulation and the WRF model simulation during the A‐FORCE period are described in more detail in sections 3 and 4.1, respectively.
3 Comparison With the A‐FORCE Aircraft Measurements
[11] The vertical profiles of the BC mass concentrations obtained from the CMAQ‐Modified simulation were compared with the A‐FORCE aircraft measurements for both specific flights and all flights together. The comparison utilized the 1 min average observation data obtained outside of clouds and the hourly model outputs that were interpolated to the corresponding time and altitude of the sampling point along the flight tracks. The 1 min aircraft data include both spirals and level flights. On average, the number of 1 min BC data points (including inside and outside of clouds) at each 1 km altitude range was greater than 300 during A‐FORCE (approximately 3000 total) [see Oshima et al., 2012, Table 3]. The bulk liquid water content and ice water content measured by a Cloud Aerosol and Precipitation Spectrometer (CAPS) [Koike et al., 2012] were used to identify cloud‐free conditions using the method of Oshima et al. [2012]. In this paper, the BC mass concentrations are shown at ambient temperature and pressure, even though the A‐FORCE aircraft measurements reported the aerosol concentrations at standard temperature and pressure (STP; 273.15 K and 1013.25 hPa). The altitude (km) of all vertical profiles shown in this paper is represented as above sea level.
[12] Figures 2a and 2b show the vertical profiles of the BC mass concentrations observed over the Yellow Sea during flight 8 (approximately 37°N, 126°E on 30 March 2009) and those observed over the East China Sea during flight 19 (approximately 33°N, 128°E on 23 April 2009), respectively (black lines). The BC mass concentrations calculated by the CMAQ‐Modified and CMAQ‐NoWetDep simulations along the corresponding flight tracks (red and green lines, respectively) and the ranges of those values within a horizontal grid interval (81 km) along the flight tracks are also shown for comparison. Focusing on these specific flights, Oshima et al. [2012] reported that the BC mass concentrations were greatly enhanced (i.e., modest amounts of BC were removed) in air parcels sampled at 3–6 km in altitude over the Yellow Sea during flight 8 as compared with the background BC concentrations. This enhancement was associated with upward transport due to a cyclone with modest precipitation over northern China. In contrast, no substantial enhancements in the BC concentrations (i.e., the removal of large amounts of BC) were observed in air parcels sampled at 5–6 km in altitude over the East China Sea during flight 19. Uplifting from cumulus convection with large amounts of precipitation over central China was likely the cause of this finding. In the case of flight 8 (Figure 2a), the CMAQ‐Modified simulation reproduced the key features of the BC vertical profile with respect to the enhancements of BC in both the PBL (0–2 km) and the FT (3–6 km). This particular simulation overestimated the BC mass concentrations by a factor of 2.4 on average (3–6 km). However, the observed BC values were within the horizontal variability (81 km) of the modeled BC values surrounding the flight tracks partly due to the uncertainties in precisely predicting the timing of the vertical transport of the BC plumes in the FT. The small differences in the BC mass concentrations in the FT between the CMAQ‐Modified and CMAQ‐NoWetDep simulations suggest the inefficient wet removal of BC and thus its high transport efficiency (i.e., TEBC = 0.96). In contrast, in the case of flight 19 (Figure 2b), relatively large differences in the BC concentrations in the FT between the two simulations suggest more efficient wet removal of BC and its lower transport efficiency (i.e., TEBC = 0.19). Although the CMAQ‐Modified simulation again overestimated the BC values by a factor of 3.3 on average (5–6 km), it reproduced the fundamental features of the observed profile of BC; specifically, there were enhancement and minimal enhancement of BC in the PBL (0–2 km) and the FT (5–6 km), respectively. The differences in the model‐simulated extent of BC wet removal between these two flights were consistent with those obtained from the observations [Oshima et al., 2012], which have shown the modest and large amounts of BC wet removal during flights 8 and 19, respectively.
[13] Figures 3a and 3b show the vertical profiles of the median values of the BC mass concentrations with their central 67% ranges for the aircraft observations and the CMAQ‐Modified simulations at latitude ranges of 26–33°N and 33–38°N, respectively, during the entire A‐FORCE period. The corresponding vertical profiles of the TEBC values are shown in Figures 3c and 3d, respectively. The CMAQ‐Modified simulation generally reproduced well the characteristics of the vertical profiles of the median BC mass concentrations despite the systematic overestimation in the FT (by a factor of 2.2 on average above 2 km in altitude). The overestimates of the BC mass concentrations in the FT suggest that the absolute values of BC (i.e., mass concentration and mass flux) in the FT would be overestimated by a factor of 2 in the CMAQ‐Modified simulation (see section 4). Figures 3c and 3d also show good agreement between the median TEBC values estimated from the model simulations and the observations at all altitude ranges (within 10% on average), including the latitudinal gradients. The latitudinal differences in TEBC values are discussed in more detail in section 4. The good agreement in the TEBC values gives confidence in the validity of the treatment of the wet removal processes of BC in the CMAQ‐Modified simulation because the TEBC values are less influenced by uncertainties in the emissions and transport processes in the model calculations. The correspondence between the TEBC values and the overestimates of the absolute value of the BC mass concentration suggest that the anthropogenic BC emissions over East Asia may be overestimated to some extent and/or that the amount of BC transported from the PBL to the FT may be overestimated in the WRF‐CMAQ model simulation, likely due to larger values of the vertical eddy diffusion coefficient in the offline WRF‐CMAQ model than those in the online model, such as WRF‐chem [e.g., Matsui et al., 2009]. A more detailed comparison of the model results with the observations is given by Kondo et al. (manuscript in preparation, 2013).
[14] To evaluate the aerosol wet removal processes used in the model in more detail, we examined the dependence of TEBC on the amount of precipitation that an air parcel had experienced during vertical transport from the PBL to the FT. Following Oshima et al. [2012], we estimated the “accumulated precipitation along trajectory (APT),” which was derived by integrating the WRF precipitation amount in a Lagrangian sense along each 5 day back trajectory of the sampled air parcels when and where the trajectory passed through the WRF precipitation water content (i.e., the sum of the rain, snow, and graupel contents) grid box anywhere from the PBL to the sampling point. The 5 day kinematic back trajectories of the sampled air parcels were calculated by applying the method described by Tomikawa and Sato [2005] to the NCEP FNL data. Note that the APT values of the sampled air parcels used in this study are the same as those presented by Oshima et al. [2012] because both studies used the same WRF model results. Detailed descriptions of the APT values and their uncertainties have been presented by Oshima et al. [2012]. Figure 4a shows the relationship between the median TEBC values and APT values with their central 67% ranges within each APT range (i.e., the APT values were divided into four evenly spaced intervals between 0.01 mm and 100 mm based on a constant common ratio) for the air parcels sampled above 2 km in altitude during the entire A‐FORCE period for the aircraft observations and the CMAQ‐Modified simulation. The regression lines for the individual sampled air parcels (1 min data; not shown) for the observations and the model results are also shown in Figure 4a (dashed lines). Decreasing tendencies in the TEBC values with increases in the APT values evident in Figure 4a indicate that the CMAQ‐Modified simulation reasonably reproduced the observed dependence of the BC wet removal on precipitation.
[15] We further examined the reproducibility of the dependence of TEBC of the air parcels on their sampling altitude and their origins for the CMAQ simulation. Following Oshima et al. [2012], the sampled air parcels were classified into four categories on the basis of the sampling altitude (2–4 km and 4–9 km) and the latitude of the origin (southern China (20–33°N) and northern China (33–50°N)) of the air parcels. The origin of an air parcel sampled above 2 km in altitude was defined as the geographical location where its 5 day back trajectory first crossed the top of the PBL. Figure 4b shows the relationship between the median TEBC and APT values with their central 67% ranges for the four categories during the entire A‐FORCE period; both aircraft observations and the CMAQ‐Modified simulation results are shown. In addition to the decreasing tendency in TEBC with increases in APT, the observed median values of TEBC in each category were also reproduced well in the CMAQ‐Modified simulation. In a manner consistent with the observations [Oshima et al., 2012] (see section 1), the median TEBC values simulated for air parcels sampled at 4–9 km in altitude were systematically smaller than those for air parcels sampled at 2–4 km. Moreover, the median TEBC values simulated for air parcels originating from southern China were systematically smaller than those of parcels originating from northern China for each of the altitude categories.
[16] It should be noted that aerosol‐cloud interactions in mixed and ice phase clouds could influence the wet removal of BC. Specifically, ice crystal growth causes the evaporation of liquid cloud and hence the release of BC aerosol back to the ambient air [Fan et al., 2012]. Although we applied a cloud microphysics scheme including mixed and ice phase processes in the WRF calculation, the off‐line WRF‐CMAQ calculation cannot represent the aerosol‐cloud interactions, which may cause errors in the estimation of BC concentrations in the upper troposphere. Nevertheless, the reasonable reproducibility of the vertical and latitudinal distributions of the BC mass concentrations and the TEBC values in the CMAQ‐Modified simulation, along with the dependence of the BC wet removal on precipitation, supports the use of the model results to examine the vertical transport mechanisms of BC particles and their transport pathways over East Asia, at least during the A‐FORCE period. Furthermore, as presented by Kondo et al. [2011a], the CMAQ‐Modified simulation reasonably reproduced the temporal variations in surface BC mass concentration observed at an island in the East China Sea (i.e., Cape Hedo observatory on Okinawa Island, Japan) during the period 2008–2009. The reproducibility during the A‐FORCE period is also shown in Appendix A, where comparisons with the BC measurements at two other ground observation sites in Japan are included.
4 Vertical Transport of BC and Pathways
4.1 Uplifting Processes and Precipitation
[17] A detailed description of the meteorological conditions during the A‐FORCE campaign based on the NCEP FNL data has been presented by Oshima et al. [2012]. Here we briefly describe the meteorological fields related to uplifting processes and their reproducibility in the WRF model simulation. Figure 5a shows the mean horizontal winds at the 850 hPa level and the mean vertical motion at the 700 hPa level during the A‐FORCE period in the WRF simulation. The local standard deviation of 700 hPa vertical velocity during the same period is also shown in Figure 5b. The WRF simulation reproduced the mean wind fields in the NCEP FNL data reasonably well [see Oshima et al., 2012, Figure 3]. As shown in Figure 5a, the midlatitude region (30–50°N, 100–140°E) in East Asia was under the influence of a modest monsoonal northwesterly flow, whereas the subtropical region (15–30°N, 90–120°E) was under the influence of persistent southerlies, part of which converged into a frontal zone that extended zonally at ~30°N over southern/central China and into the southern and eastern slopes of the Tibetan Plateau (90–105°E). Note that the low‐level southerly flow advected warm, moist air into these regions to sustain a large amount of precipitation [see Oshima et al., 2012, Figure 3]. In association with the low‐level wind convergence, mean vertical motion was upward around the frontal zone and along the major mountain slopes (Figure 5a). The mean ascent was also pronounced over the Indochina Peninsula and the Philippines in association with a large amount of precipitation (Figure 5c).
[18] To examine the mechanisms involved in the mean upward motion, we estimated the spatial distributions of the activities of migratory cyclone and cumulus convection based on the NCEP FNL data. As a measure of the activity of migratory cyclones and anticyclones, the poleward eddy heat flux 
at the 850 hPa level was locally evaluated (Figure 6a), where primes signify instantaneous deviations from the 5 day running means at each grid point and the overbar denotes time averaging over the A‐FORCE period. Note that regions of large eddy heat transport are called “storm tracks” [e.g., Nakamura et al., 2002], which generally correspond to regions where the lower tropospheric poleward airflow associated with WCBs tends to be pronounced during the passage of cyclones, especially over the ocean. The 
values were large over northeastern China (35–50°N, 110–130°E) and along the Pacific storm track around northern Japan, whereas these values were modest over southern/central China (20–30°N, 105–120°E), indicating that the activity of migratory cyclone was much higher over northeastern than southern/central China (Figure 6a). It should be noted that no distinct mean ascent was seen over northeastern China (Figure 5a) because vertical velocities associated with alternating cyclone and anticyclone activity tend to cancel out in the mean. Regardless, these disturbances gave rise to large fluctuations in vertical motion just above the PBL (Figure 5b), contributing to upward transport of pollutants from the PBL into the FT, as discussed below. These results suggest that transient ascent on the passage of cyclones played a major role in the upward transport of pollutants over northeastern China.
at the 850 hPa level (m s−1 K), (b) mean convective available potential energy (CAPE) (J kg−1), and (c) mean precipitable water (kg m−2) during the A‐FORCE period (20 March to 30 April 2009) obtained from the NCEP FNL data, where v′ and T′ are the instantaneous deviations from the respective 5 day running means at each grid point and the overbar denotes time averaging over the period. Regions without data correspond to regions with high‐altitude mountains.
[19] Figure 6b shows the spatial distribution of convective available potential energy (CAPE) averaged over the A‐FORCE period, which provides a measure of the possibility of the occurrence of deep convection. The presence of large CAPE values over southern/central China suggests active cumulus convection. The large standard deviation in vertical motion embedded in the mean ascent over southern/central China (Figure 5b) is indicative of episodic enhancement of convective activity. CAPE values were relatively large in the tropical region (10–20°N), including the Indochina Peninsula and the Philippines. In contrast, CAPE values were much smaller in the midlatitude region (30–50°N), suggesting less convective activity. These results indicate that the pronounced mean ascent in the FT over southern/central China, the Indochina Peninsula, and the Philippines (Figure 5a) was associated with active cumulus convection, although there may be an additional contribution from cyclones to the uplift of pollutants over southern/central China. The results presented here are consistent with those described in Oshima et al. [2012], in which the temporal variations in migratory cyclones and cumulus convections over continental China were demonstrated using other measures based on the NCEP FNL data and Multifunctional Transport Satellite (MTSAT) infrared (IR) images.
[20] We discuss here precipitation activity over East Asia during the A‐FORCE period on the basis of the WRF model simulation because ascent associated with precipitation formation is also involved in the vertical transport of air parcels. Figures 5c and 5d show the mean precipitation and the fraction of convective precipitation in the WRF simulation during the A‐FORCE period, respectively. The mean precipitation was particularly large over southern/central China, including the eastern slope of the Tibetan Plateau, the Indochina Peninsula, the Philippines, and the western Pacific (the tropical region and to the south and east of Japan). Within most of these regions, particularly in the tropical and subtropical regions, CAPE values were large (Figure 6b) with abundant moisture (Figure 6c) and they were consistent with convective precipitation accounting for more than half of the total precipitation (Figure 5d). The large amount of precipitation along the eastern slope of the Tibetan Plateau was likely due to orographic uplifting of warm, moist air parcels. In contrast, a modest amount of precipitation simulated over northeastern China (Figure 5c) was mostly nonconvective (Figure 5d), generated by large‐scale condensation associated with migratory cyclones (Figures 5b and 6a).
[21] In the following, the reproducibility of precipitation in the WRF simulation is further assessed, although Oshima et al. [2012] validated it with the Global Precipitation Climatology Project (GPCP) data [Huffman et al., 2001] for the midlatitude and subtropical regions during the A‐FORCE campaign [see Oshima et al., 2012, Figure 4]. Daily GPCP precipitation data available on a latitude‐longitude grid with a resolution of 1° based on satellite and rain gauge measurements were used for the assessment. Comparisons of the results between the WRF precipitation and the GPCP data for several BC source regions over East Asia (enclosed within the thick black lines in Figure 5c) are summarized in Table 2. The comparison shows that the mean WRF precipitation generally overestimated the mean GPCP precipitation during the A‐FORCE period (at most by approximately 60% over North China and Southeast Asia), whereas the model underestimated it by 34% over South Asia (see Table 2). The overestimates and underestimates of precipitation over these regions may lead to the respective prediction of smaller and greater BC values in the FT in the CMAQ simulations. Although the discrepancies shown in Table 2 could cause some uncertainties in the CMAQ simulations, we nevertheless conclude that the WRF simulation reproduced the regional‐scale spatial distribution of precipitation reasonably well during the A‐FORCE period. Attempts to reduce the discrepancies in precipitation should be made in the future.
| Evaluation Region | Number of Data Points | Mean GPCP (mm d−1) | Mean WRF (mm d−1) | NMB (%) | r | NRMSE |
|---|---|---|---|---|---|---|
| (A) Japan | 15,750 (630) | 3.45 | 3.68 | 6.55 | 0.603 (0.716) | 2.00 (1.43) |
| [22–47°N, 130–145°E] | ||||||
| (B) North China | 18,900 (756) | 0.778 | 1.26 | 62.2 | 0.478 (0.631) | 4.09 (2.92) |
| [35–50°N, 100–130°E] | ||||||
| (C) South China | 18,900 (756) | 3.41 | 4.07 | 19.4 | 0.432 (0.553) | 2.35 (1.62) |
| [20–35°N, 100–130°E] | ||||||
| (D) Southeast Asia | 8,400 (336) | 3.65 | 5.93 | 62.6 | 0.408 (0.599) | 2.75 (1.62) |
| [10–20°N, 90–110°E] | ||||||
| (E) South Asia | 16,800 (672) | 0.991 | 0.650 | −34.4 | 0.171 (0.382) | 4.93 (3.04) |
| [20–30°N, 70–100°E] | ||||||
| [15–20°N, 70–90°E] |
- a NMB, normalized mean bias; r, correlation coefficient; NRMSE, normalized root‐mean‐square error. The evaluation regions are shown in Figure 5c. The values are spatially and temporally averaged for each 1° × 1° grid box and 1 day, respectively, within each region. The values in parentheses are spatially averaged for each 5° × 5° grid box within each region.
4.2 Vertical Transport of BC
[22] The spatial distributions of the mean horizontal mass flux of BC integrated over the PBL (700 ~ 1000 hPa layer) and the FT (200 ~ 700 hPa layer) during the A‐FORCE period in the CMAQ‐Modified simulation are shown in Figures 7a and 7b, respectively. It should be noted that the actual level of the PBL top is generally lower than the 700 hPa level. The lowermost layer below that level, which includes the PBL, is symbolically referred to as “PBL” in this study. The horizontal convergence and divergence of the corresponding column‐integrated horizontal BC fluxes are also shown in Figure 7. The strong divergence regions of the horizontal BC fluxes within the PBL (approximately 30–40°N, 110–120°E, green and blue regions in Figure 7a) generally correspond to the large BC source regions (Figure 1) and the divergences represent the horizontal transport of BC emitted from those source regions to the surrounding regions. Below the 700 hPa level (Figure 7a), there were strong southwesterly BC fluxes over northeastern China (approximately 35–45°N, 110–130°E), carrying a large amount of BC from its sources in northern/central China (i.e., the divergence regions). The BC flux was converging over northeastern China, being confluent with the prevailing clean northwesterlies (40–50°N), as presented in section 4.1 (Figure 5a). Figure 7a also shows another strong poleward BC fluxes converging into the frontal zone over southern/central China (approximately 25–30°N, 105–115°E) and the eastern slope of the Tibetan Plateau (100–105°E). The convergence regions of the horizontal BC fluxes (yellow and red regions in Figure 7a) appeared around the divergence (i.e., BC source) regions, and the convergences represent the upward BC transport from the PBL to the FT over those regions.
[23] Figure 8a shows the spatial distribution of the upward BC mass fluxes at the 700 hPa level in the CMAQ‐Modified simulation averaged over the A‐FORCE period. Corresponding to the horizontal BC flux convergence regions within the PBL (Figure 7a), two types of pronounced mean upward BC fluxes from the PBL to the FT were found over northeastern China (approximately 35–45°N, 110–130°E) and inland‐southern China (approximately 25–35°N, 100–120°E). It should be emphasized that the regions with strong upward BC fluxes were not collocated with large BC sources (Figure 1), consistent with the geographical relationship between the horizontal BC flux divergence and convergence regions within the PBL (Figure 7a).
at the 700 hPa level during the A‐FORCE period (20 March to 30 April 2009) in the CMAQ‐Modified simulation (µg m−2 s−1), where the overbar denotes time averaging over the period. Positive fluxes are upward. The solid black lines denote the terrain elevations (m) used in the model simulation. Time averages of (b) the mean‐motion term 
and (c) the transient term 
in equation 4 (µg m−2 s−1), where a prime denotes the instantaneous deviations from the respective mean at each grid point. See the text for details.
(2)
(3)
(4)
(Figure 8c) was dominant, which is consistent with large fluctuations in vertical motion (Figure 5b) associated with strong transient eddy activity (Figure 6a). This collocation suggests the particular importance of migratory cyclones for the upward BC transport from the PBL to the FT, which is also supported by a positive correlation between temporal fluctuations in [BC] and w during the A‐FORCE period (r = 0.50 at the 700 hPa level and within 33–48°N, 110–130°E) (not shown). In contrast, the large values of the mean‐motion term 
over inland‐southern China (Figure 8b) are primarily attributed to the pronounced mean ascent (Figure 5a). As shown in section 4.1, the mean ascent over this region was mainly caused by orographic uplifting along the slopes of the Tibetan Plateau and by active cumulus convection, with a modest additional contribution from migratory cyclones. The importance of orographic uplifting of pollutants (e.g., CO) to the FT along mountain slopes in East Asia has also been described in several previous studies [e.g., Liu et al., 2003; Chen et al., 2009; Ding et al., 2009; Lin et al., 2010]. The weak negative correlation between temporal fluctuations in [BC] and w during the A‐FORCE period (r = −0.24 at the 700 hPa level and within 20–33°N, 100–120°E) (not shown) suggests that mechanisms for BC uplifting may differ between the two regions. We thus concluded that migratory cyclones played a major role in the BC uplifting from the PBL to the FT over northeastern China, whereas orographic uplifting and cumulus convection were responsible for the upward BC transport over inland‐southern China.
[25] The 42 day period (i.e., 20 March to 30 April 2009, the A‐FORCE period) used for the time averaging and defining the transient fluctuations in equations 2 and 3 may be much longer than the typical time scales of some uplifting processes. Considering the time scales of synoptic‐scale disturbances, we conducted another estimation of the mean‐motion and transient terms in equation 4 by replacing the 42 day means in equations 2 and 3 with the respective 5 day running means. This substitution gave almost the same spatial distributions for those terms shown in Figures 8b and 8c (not shown). There were no substantial increasing or decreasing trends in the [BC] and w time series during the A‐FORCE period. These results indicate the validity of our analyses, in which the mean‐motion and transient terms in equation 4 were used to identify the mechanisms of the upward BC transport.
[26] Figures 9a and 9b show the spatial distributions of the amounts of wet and dry depositions of BC, respectively, averaged over the A‐FORCE period in the CMAQ‐Modified simulation. The contribution of the wet deposition of BC to the total deposition was estimated to be 90% in East Asia (10–50°N, 80–140°E). The spatial distribution of the BC dry deposition amounts (Figure 9b) was generally similar to BC emissions (Figure 1) because the dry deposition flux of BC from the atmosphere to the surface is approximately proportional to the local BC concentration near the surface. In contrast, the spatial distribution of BC wet deposition (Figure 9a) consisted generally of a combined effect of the upward BC flux (Figure 8a) and the precipitation distribution (Figure 5c) because the wet removal occurs in association with precipitation during the vertical transport of air parcels containing BC. The amount of BC wet deposition was largest over inland‐southern China, including the eastern slopes of the Tibetan Plateau, where both the upward BC flux and precipitation were pronounced. Figure 9a also shows fairly large amounts of BC wet deposition over northeastern China, where the strong upward BC flux was collocated with modest precipitation associated with cyclones. The region of modest amounts of BC wet deposition extended farther toward the east along the Pacific storm track.
[27] To examine the intensity of the wet removal of BC from air parcels during their vertical transport over the two uplifting regions (i.e., northeastern China and inland‐southern China), the TEBC and TEBCflux values were evaluated for the mean BC mass concentrations and the mean upward BC mass fluxes in the simulations, respectively (see section 2). The mean TEBCflux (TEBC) values were 0.35 (0.57) and 0.66 (0.80) over inland‐southern China (20–33°N, 100–120°E) and northeastern China (33–48°N, 110–130°E), respectively, at the 700 hPa level during the A‐FORCE period. This result indicates that the intensity of the wet removal of BC over inland‐southern China was approximately twice (i.e., a 65% (43%) decrease by wet removal) that over northeastern China (a 34% (20%) decrease), which is nearly consistent with the geographical dependence of the amount of BC wet deposition (Figure 9a).
4.3 Spatial Distributions of BC and TEBC
[28] In this section, we describe the spatial distributions of BC and possible mechanisms involved in forming the distributions with a particular focus on the outflow region (10–50°N, 120–140°E) in East Asia. Below the 700 hPa level (Figure 7a), the major transport pathway for Asian BC particles was toward the western Pacific, carried by the midlatitude westerlies (35–45°N). In addition, some of the BC particles were transported toward the East China Sea (south of 35°N) by the northwesterlies. In the FT (200–700 hPa level; Figure 7b), there were two major branches of westerly BC fluxes, specifically, subtropical (25–35°N) and midlatitude (35–45°N) regions over the outflow regions (the convergence regions in Figure 7b). These branches reflect the upward BC transport from the PBL to the FT over inland‐southern China and northeastern China (the divergence regions in Figure 7b), respectively, and their subsequent westerly transport in the FT.
[29] Figures 10a and 10b show the spatial distributions of the mean BC mass concentrations integrated over the PBL (700–1000 hPa layer) and the FT (200–700 hPa layer), respectively, during the A‐FORCE period in the CMAQ‐Modified simulation. Below the 700 hPa level, the spatial distribution of the column‐integrated BC appears to be controlled primarily by the spatial distribution of BC emissions (Figure 1), and the region of large BC column extended eastward due to the advective effect by the westerlies (Figure 7a). Above the 700 hPa level, the BC column was greatest over inland‐southern China (Figure 10b), arising from the upward transport of BC mostly originating from emissions in southern China (55% and 21% contributions from Southwest China (SWC) and Southeast China (SEC) to the upward BC mass flux, respectively) in the vicinity of the uplifting region. Because of the pronounced advective BC fluxes in the FT by the strong westerlies (Figure 7b), the regions of large free tropospheric BC column extended far downstream from the uplifting regions over northeastern China and inland‐southern China.
[30] Figures 10c and 10d show the spatial distributions of the mean TEBC values for the PBL and FT columns during the A‐FORCE period, respectively. The mean TEBC values were generally smaller at lower latitudes for both BC columns. The spatial distribution of the mean TEBC values was generally consistent with that of mean precipitable water (Figure 6c). Specifically, regions with more atmospheric moisture correspond with smaller TEBC values. Below the 700 hPa level (Figure 10c), a smaller TEBC value region intruded over inland‐southern China, where abundant moisture was supplied by the low‐level southerlies (section 4.1), yielding a large amount of precipitation and resultant BC removal. The TEBC values in the free tropospheric column were smaller over the uplifting regions (i.e., northeastern China and inland‐southern China) due to the effective wet removal of BC from air parcels uplifted from the PBL to the FT (Figure 10d). The smaller TEBC value regions in the FT extended eastward from the respective uplifted regions, reflecting the westerly transport of BC in uplifted air parcels (Figure 7b). It should be noted that the TEBC values over inland‐southern China were smaller than those over northeastern China, as described in section 4.2. These results indicate that the latitudinal dependence of precipitation induced by the moisture supply by the low‐level southerlies and the subsequent uplifting of air parcels was responsible for the spatial distributions of BC and TEBC in East Asia.
[31] The vertical distributions of the mean BC mass concentration averaged longitudinally between 120°E and 140°E during the A‐FORCE period in the CMAQ‐Modified simulation and the corresponding mean TEBC values are shown in Figures 11a and 11b, respectively. Although the BC mass concentrations were greatest below the 700 hPa level over the midlatitude region (30–45°N), the high BC concentration regions spread to the midtroposphere (~500 hPa level) over a wide latitude range (20–45°N). In contrast, TEBC exhibits more distinct dependency on latitude and altitude. Specifically, TEBC decreased with decreasing latitude and increasing altitude (Figure 11b), reflecting the latitudinal distributions of moisture and precipitation. As shown in Figure 4b, the A‐FORCE aircraft measurements clearly indicated the latitude and altitude dependency of the TEBC values, which is consistent with the longitudinally and temporally averaged TEBC values over the outflow region throughout the A‐FORCE period in the CMAQ‐Modified simulation (e.g., Figure 11b). This good agreement indicates that the general features of wet removal of BC during springtime in East Asia were well captured in the A‐FORCE aircraft measurements.
4.4 Export of BC From East Asia
[32] Figure 12 shows a meridional cross section of the mean eastward mass flux of BC at 140°E averaged over the A‐FORCE period in the CMAQ‐Modified simulation. The eastward BC flux was strongest in the PBL and the lower FT (600–900 hPa level) at 35–45°N and the flux was also strong in the middle FT (400–700 hPa level) at 25–35°N. Contributions of the individual BC source regions (Figure 1) to the mean eastward BC flux at 140°E are summarized in Table 3. These results indicate that there were three major transport pathways for export of BC from East Asia to the western Pacific. The three major transport pathways of BC and its vertical transport mechanisms along each pathway are schematically shown in Figure 13. One pathway was PBL outflow in the midlatitude region (35–45°N) through which BC originating primarily from northern China was advected by the low‐level westerlies without uplifting out of the PBL (i.e., contributions of 76% and 8% from North China (NC) and South China (SC), respectively, to the eastward BC fluxes). The greater TEBCflux value of 0.79 (i.e., 100%/ 127% in Table 3) for the PBL outflow (35–50°N, 700–1000 hPa) indicates that this outflow was an efficient transport pathway for BC export from continental East Asia to the Pacific. The importance of the boundary layer outflow of Asian pollution (e.g., CO) has also been discussed by Liang et al. [2004]. The second pathway was through uplifting from the PBL to the FT by migratory cyclones over northeastern China and the subsequent eastward transport in the lower FT over the midlatitude region (35–45°N). In the second pathway, BC originated mostly from northern China (i.e., contributions of 58% and 20% from NC and SC, respectively), which is consistent with the strong divergence of the horizontal BC fluxes within the PBL over northern/central China (Figure 7a). Along the second pathway, approximately half of the BC mass (TEBCflux of 0.50, i.e., 100%/199% in Table 3) was removed from the atmosphere by precipitation until its arrival at 140°E. The third pathway was orographic uplifting and/or convective upward transport from the PBL to the FT over inland‐southern China followed by westerly transport in the mid‐FT over the subtropical region (25–35°N). In the third pathway, the primary origins of the exported BC were southern China and India (i.e., 43%, 30%, and 16% contributions from South Asia (SA), SC, and NC, respectively). Despite the efficient wet removal of BC (TEBCflux of 0.33, i.e., 100%/304% in Table 3) due to the large amount of precipitation (see sections 4.2 and 4.3), the eastward BC flux in the FT over the subtropical region was comparable in magnitude to that over the midlatitude region (Table 3).
| Category | Corresponding Simulation | 20–35°N | 35–50°N | ||
|---|---|---|---|---|---|
| 200–700 hPa | 700–1000 hPa | 200–700 hPa | 700–1000 hPa | ||
| East Asia | CMAQ‐Modified | 683 (100%) | 617 (100%) | 672 (100%) | 1199 (100%) |
| CH | CMAQ‐China | 315 (46.2%) | 430 (69.7%) | 521 (77.5%) | 1013 (84.4%) |
| NC | CMAQ‐NorthChina | 110 (16.2%) | 283 (46.0%) | 390 (58.0%) | 914 (76.2%) |
| SC | CMAQ‐China | 205 (30.0%) | 146 (23.7%) | 131 (19.5%) | 99.1 (8.26%) |
| −CMAQ‐NorthChina | |||||
| SWC | CMAQ‐SouthWestChina | 105 (15.4%) | 64.8 (10.5%) | 58.1 (8.65%) | 6.18 (0.515%) |
| SEC | CMAQ‐China | 99.7 (14.6%) | 81.3 (13.2%) | 73.0 (10.9%) | 92.9 (7.75%) |
| −CMAQ‐NorthChina | |||||
| −CMAQ‐SouthWestChina | |||||
| JP | CMAQ‐Japan | 1.62 (0.238%) | 37.4 (6.06%) | 2.32 (0.345%) | 50.4 (4.20%) |
| KR | CMAQ‐Korea | 2.79 (0.408%) | 23.1 (3.75%) | 6.66 (0.991%) | 76.4 (6.37%) |
| SA | CMAQ‐SouthAsia | 294 (43.1%) | 89.5 (14.5%) | 94.8 (14.1%) | 1.79 (0.150%) |
| SEA | CMAQ‐SouthEastAsia | 42.2 (6.19%) | 18.2 (2.95%) | 1.43 (0.213%) | 0.0620 (0.00517%) |
| AN | CMAQ‐Anthropogenic | 577 (84.5%) | 558 (90.4%) | 612 (91.0%) | 1115 (93.0%) |
| BB | CMAQ‐Modified | 106 (15.5%) | 59.1 (9.58%) | 60.2 (8.97%) | 84.1 (7.01%) |
| −CMAQ‐Anthropogenic | |||||
| East Asia | CMAQ‐NoWetDep | 2077 (304%) | 1292 (210%) | 1334 (199%) | 1524 (127%) |
- a The simulations are summarized in Table 1; the individual source regions are shown in Figure 1. The units of the eastward mass flux of BC are ng m−2 s−1. The values in parentheses are the relative contributions (in %) to the CMAQ‐Modified simulation. SC, South China; SEC, Southeast China; AN, anthropogenic; BB, biomass burning.
[33] A rough estimation was made for the total amount of transported BC across the 140°E meridian in the FT (between 200 and 700 hPa levels, approximately 3 ~ 12 km in altitude) and the PBL (700 ~ 1000 hPa, 0 ~ 3 km) for each of the three pathways. Given that BC values in the FT were overestimated (by a factor of 2) in the CMAQ‐modified simulation (see section 3), the total amount of the transported BC in the FT outflows (the second and third pathways) was nearly comparable in magnitude to that in the PBL outflow (the first pathway). The contribution from biomass burning was greatest (16%) in the third pathway among the four possible transporting categories shown in Table 3. However, the particular contributions may be underestimated in this study due to the lack of seasonal variations in biomass burning emissions in the simulations. Nevertheless, the relatively large contributions from biomass burning to the Asian pollution outflow in the FT over the subtropical region in spring are consistent with the findings of previous studies [e.g., Bey et al., 2001; Liu et al., 2003; Uno et al., 2003]. It should be noted that our estimation of the BC transport was made for spring 2009 because we focused on the A‐FORCE aircraft measurements and the export of Asian pollution to the Pacific is dominant in spring [e.g., Liu et al., 2003; Liang et al., 2004]. However, the spatial distribution of the eastward BC transport (e.g., Figure 12) varies greatly depending on the meteorological fields. Seasonal variations in the Asian BC outflow to the Pacific in the period 2008–2010 are discussed by Matsui et al. [2013a].
5 Summary and Conclusions
[34] In this study, a modified version of the WRF‐CMAQ model was used to examine the vertical transport mechanisms of BC particles and their 3‐D transport pathways during springtime in East Asia (20 March to 30 April 2009) through the simulations for the A‐FORCE aircraft campaign. Comparisons of the baseline (CMAQ‐Modified) model simulation with the A‐FORCE aircraft observations revealed that the vertical profiles of BC mass concentration, with respect to both the median values of all flights and the in situ values for specific flights (flights 8 and 19), were reproduced reasonably well in the model simulation, although the CMAQ‐Modified simulation generally overestimated the BC mass concentrations in the FT by a factor of 2. The reasonable reproducibility (within 10% differences on average) was also revealed in the CMAQ‐Modified simulation for the vertical distribution of the transport efficiency of BC (TEBC), along with the features of the wet removal of BC (i.e., the dependence of TEBC of air parcels on their sampling altitude and latitude, their origin, and the amount of precipitation that the air parcels had experienced during transport), during the entire A‐FORCE period.
[35] Our analyses of the model results and meteorological fields showed two types of pronounced mean upward mass fluxes of BC from the PBL to the FT over northeastern China (approximately 35–45°N, 110–130°E) and inland‐southern China (approximately 25–35°N, 100–120°E) during the A‐FORCE period. This upward BC transport was found to be consistent with the convergence of horizontal BC mass flux within the PBL over the respective regions. Over northeastern China, the major uplifting mechanism of BC was migratory cyclones with modest amounts of precipitation, which caused modest levels of wet deposition of BC (34% decrease in the upward BC mass flux due to the wet removal by precipitation). In contrast, over inland‐southern China, both cumulus convection and orographic uplifting along the slopes of the Tibetan Plateau were responsible for the upward BC transport into the FT despite the highest levels of BC wet deposition caused by the large amount of precipitation (65% decrease in the upward BC mass flux).
[36] Over the outflow region (10–50°N, 120–140°E), the BC mass concentration averaged during the A‐FORCE period was found to be the highest below the 700 hPa level in the midlatitudes (30–45°N), with strong westerly horizontal BC mass fluxes within the PBL. The mean BC mass concentration was also relatively high in the midtroposphere (~500 hPa level) over a wider latitude range (20–45°N), arising from the upward BC transport from the PBL to the FT over the major uplifted regions (i.e., northeastern and inland‐southern China) and the subsequent westerly BC transport in the FT. The mean TEBC values were found to exhibit a distinct dependency on both latitude and altitude. Specifically, the TEBC value decreased with decreasing latitude and increasing altitude. The latitudinal difference in precipitation induced by moisture supply by the low‐level southerlies and the subsequent uplifting processes were responsible for the spatial distributions of BC and TEBC.
[37] We identified three major transport pathways for BC export from East Asia to the western Pacific (across the 140°E meridian) during springtime in the A‐FORCE period from the CMAQ‐Modified simulation (Figure 13). Two of them were the midlatitude (35–45°N) pathways for BC originating primarily from northern China, one for BC remaining within the PBL by near‐surface eastward outflow and the other for BC uplifted into the FT in association with migratory cyclones over northeastern China and subsequently transported by the lower tropospheric westerlies. The third pathway was for BC uplifted from the PBL to the FT along the slopes of the Tibetan Plateau and/or by convective updrafts over inland‐southern China and subsequently transported by the midtropospheric westerlies in the subtropical region (25–35°N). The major origins of the exported BC through this pathway were likely southern China and India. The mean eastward TEBCflux values at 140°E during the A‐FORCE period were estimated to be 0.79, 0.50, and 0.33 for the first, second, and third transport pathways, respectively. Although the PBL outflow was an efficient transport pathway for BC, the two FT outflows also played significant roles in the BC export from continental East Asia to the Pacific during the A‐FORCE period in spring.
Acknowledgments
[45] This work was supported by the Ministry of Education, Culture, Sports, Science, and Technology and the Japan Society for the Promotion of Science (MEXT/JSPS) KAKENHI grant numbers 23221001 and 23710029. This work was also supported by the strategic international cooperative program of the Japan Science and Technology Agency (JST), the global environment research fund of the Japanese Ministry of the Environment (A‐0803 and A‐1101), and the GRENE Arctic Climate Change Research Project. This study was conducted as a part of the Mega‐Cities: Asia Task under the framework of the International Global Atmospheric Chemistry (IGAC) project. HN was supported in part by MEXT through Grant‐In‐Aids in Scientific Research on Innovative Areas 2205 and 2509 and by the Japanese Ministry of Environment through the Environment Research and Technology Development Fund A1201. The trajectory calculation program used in this paper was developed by Y. Tomikawa of the National Institute of Polar Research and K. Sato of the University of Tokyo, Japan. We are especially grateful to Y. Kanaya for his work with the field measurements at the Fukue observatory and for providing the BC data. We also thank A. Takami, K. Kawana, M. Irwin, and S. Ohata for their continued assistance with the field measurements at the Hedo observatory. We express our appreciation to K. Nakagomi and T. Ikeda for their cooperation with the field measurements at the Happo observatory. We thank M. Kajino and M. Deushi for providing helpful comments on this paper.
Appendix A: Modification of the Aerosol Wet Deposition Scheme and Evaluation
(A1)[39] The modified aerosol wet deposition scheme can distinguish between rainout and washout processes and ignores the washout process for accumulation‐mode aerosols. Specifically, the modified scheme ignores the wet deposition of accumulation‐mode aerosols when the cloud water mixing ratio is smaller than the prescribed threshold value, which is assumed to be the same as the CLWtv value used in the original scheme. This modification is added to the calculations for the aerosol wet deposition in the original scheme, which enables the elimination of the rainout that should be treated as washout in the original scheme (e.g., heavy rain). It should be noted that this approach should not be applied for Aitken and coarse‐mode particles because particles of those sizes are more efficiently removed by the collision of precipitating raindrops, depending on the diameters of the particles and the rain droplets and the precipitation intensity [Seinfeld and Pandis, 2006]. The original aerosol wet deposition scheme is applied for the Aitken and coarse‐mode particles.
[40] In the following, evaluations of the modified aerosol wet deposition scheme adopted in the CMAQ model (i.e., the CMAQ‐Modified simulation) using the A‐FORCE aircraft measurements and the surface BC measurements are described through comparisons with the simulation using the original CMAQ scheme (i.e., the CMAQ‐Original simulation) (see Table 1). Figure A1 shows the vertical profiles of the median values of the BC mass concentrations and the TEBC values with their central 67% ranges for the aircraft observations, the CMAQ‐Modified simulation, and the CMAQ‐Original simulation during the entire A‐FORCE period. The CMAQ‐Original simulation significantly underestimated the TEBC values at all altitude ranges by up to 50%, but the CMAQ‐Modified simulation greatly improved the calculations of TEBC (Figure A1b). The improvement in the TEBC values supports the validity of the modified scheme (i.e., the treatment of aerosol wet deposition processes) and also indicates the problems with the original scheme. In contrast, the CMAQ‐Modified simulation did not show improvements for the BC mass concentrations (Figure A1a); however, the absolute values of BC in the FT depend on the emissions and the vertical eddy diffusion coefficient used in the model simulations (see section 3).
[41] Temporal variations of BC mass concentrations were compared with the surface BC measurements at the Hedo (26.9°N, 128.3°E, 60 m above sea level), Fukue (32.8°N, 128.7°E, 80 m), and Happo (36.7°N, 137.8°E, 1850 m) observatories in Japan during the period March–April 2009, which covers the entire A‐FORCE period. In these observations, BC mass concentrations in the fine mode (i.e., particles with aerodynamic diameters smaller than 2.5 µm) were measured using a filter‐based absorption photometer, the Continuous Soot Monitoring System (COSMOS), with an accuracy of approximately 10% [Miyazaki et al., 2008; Kondo et al., 2009, 2011b]. More detailed descriptions of the measurements are given by Kondo et al. [2011a] and Verma et al. [2011] for Hedo, Kanaya et al. [2013] for Fukue, and Liu et al. [2013] for Happo. Figure A2 shows the temporal variations of the hourly BC mass concentrations observed and those calculated by the CMAQ‐Modified and CMAQ‐Original simulations for Hedo, Fukue, and Happo for the period March–April 2009, and the comparison results are summarized in Table A1. The observed temporal variations in the BC mass concentrations at the three sites were generally well reproduced by the CMAQ‐Modified simulation, although the simulation did not reproduce the large enhancements of BC observed at Fukue (Figure A2b). Relative to the CMAQ‐Original simulation, the CMAQ‐Modified simulation yielded more accurate BC mass concentration levels (see the normalized mean bias (NMB) in Table A1), whereas the correlation coefficient (r) and the normalized root‐mean‐square error (NRMSE) were not improved (i.e., they were similar for the two simulations), partly due to errors in precisely predicting the timing of the BC transport from continental China with time resolutions of a few hours in the simulations [e.g., Kondo et al., 2011a].
| Mean BC Mass Concentration (ng m−3 STP) | NMB (%) | r | NRMSE | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Site Name | Number of Data Pointsbb
Number of observed hourly BC data points available for March–April 2009. These data points are used for the statistics. |
Observations | CMAQ‐Mod | CMAQ‐Org | CMAQ‐Mod | CMAQ‐Org | CMAQ‐Mod | CMAQ‐Org | CMAQ‐Mod | CMAQ‐Org |
| Hedocc
Hedo (26.9°N, 128.3°E, 60 m), Japan. |
1464 | 353 (317) | 328 (238) | 239 (211) | −7.22 | −32.4 | 0.620 | 0.657 | 0.716 | 0.749 |
| Fukuedd
Fukue (32.8°N, 128.7°E, 80 m), Japan. |
750 | 874 (666) | 566 (316) | 453 (319) | −35.2 | −48.2 | 0.596 | 0.727 | 0.712 | 0.736 |
| Happoee
Happo (36.7°N, 137.8°E, 1850 m), Japan. |
1044 | 213 (183) | 205 (181) | 171 (183) | −3.39 | −19.5 | 0.487 | 0.557 | 0.866 | 0.833 |
- a NMB, normalized mean bias; r, correlation coefficient; NRMSE, normalized root‐mean‐square error; CMAQ‐Mod, CMAQ‐Modified simulation; CMAQ‐Org, CMAQ‐Original simulation. The model values at the bottom layer (Hedo and Fukue) and those at a sigma level of 0.875 (Happo) are used for the statistics. The BC mass concentrations are in units of ng m−3 at STP. The values in parentheses are standard deviations.
- b Number of observed hourly BC data points available for March–April 2009. These data points are used for the statistics.
- c Hedo (26.9°N, 128.3°E, 60 m), Japan.
- d Fukue (32.8°N, 128.7°E, 80 m), Japan.
- e Happo (36.7°N, 137.8°E, 1850 m), Japan.
[42] To remove the uncertainties in the prediction of the timing of the BC transport in the simulations, the frequency distributions of the hourly BC mass concentration for the observations and the CMAQ‐Modified and CMAQ‐Original simulations for Hedo, Fukue, and Happo during the period March–April 2009 are shown in Figure A3. The shapes of the frequency distributions in the CMAQ‐Modified simulation were very similar to those observed at the three sites, although this simulation did not adequately predict the observed BC values exceeding 1.5 µg m−3 at Fukue, as shown in Figure A2b. In contrast, the CMAQ‐Original simulation substantially overestimated the frequencies of the lowest BC values (0–0.1 µg m−3) at the three sites due to the erroneous calculations of the washout process (below‐cloud scavenging) in the original scheme; however, this did not appear in the observations and the CMAQ‐Modified simulation (Figure A3). The occurrence of the artificial zero values for the BC mass concentration when using the original scheme in the CMAQ calculations can be seen in Figure A2 and in the results of previous studies for Hedo during different time periods [Kondo et al., 2011a]. It should be noted that the erroneous reductions in the BC concentrations (i.e., zero values) typically occurred at remote regions, such as islands surrounded by ocean. These errors are difficult to be recognized at the source regions because of the continuous emissions.
[43] The evaluations presented in this appendix suggest that the modified aerosol wet deposition scheme in the CMAQ model provides more reliable estimates of the spatial distributions of BC on a regional scale than the original CMAQ scheme. The modified scheme has been used in several CMAQ model calculations, which gave reasonable results in previous studies [e.g., Kondo et al., 2011a; Verma et al., 2011; Liu et al., 2013; Matsui et al., 2013a].
Appendix B: Uncertainties in the Hydrophilic Assumption of Internally Mixed BC Particles
[44] In this appendix, we discuss the possibly uncertainties caused by the assumption in the CMAQ model that BC particles included in the accumulation‐mode internally mixed aerosols are hydrophilic (see section 2). In general, freshly emitted BC particles are in a hydrophobic state (i.e., little coating materials on BC) and they become hydrophilic BC (i.e., internally mixing with water‐soluble compounds) through aging processes in the atmosphere [e.g., Weingartner et al., 1997; Sakurai et al., 2003; Riemer et al., 2004, 2009; Toner et al., 2006; Moteki et al., 2007; Oshima et al., 2009a; Liu et al., 2011]. The hydrophilic BC particles are able to act as CCN at a given supersaturation level [Seinfeld and Pandis, 2006]. Therefore, the aging processes influence the atmospheric lifetime of BC [Stier et al., 2006]. Several previous studies have shown that the conversion time scale for BC from the hydrophobic state to the hydrophilic state over the source and outflow regions in East Asia was less than 1 ~ 2 days [e.g., Park et al., 2005; Goto et al., 2012; Oshima and Koike, 2013]. As discussed by Oshima et al. [2012] in detail, the time scales of uplifting processes (e.g., migratory synoptic‐scale cyclones and convective activities) were longer than the conversion time scales. Therefore, most BC particles were likely to be CCN active within the PBL before uplifted into the FT over East Asia during the A‐FORCE period. Previous modeling studies have also shown that most BC particles had sufficient coating materials to be CCN active due to the aging processes within the PBL over the anthropogenic source regions in East Asia in spring [Oshima et al., 2009b; Matsui et al., 2013b]. Although the hydrophilic assumption for the BC particles can generally lead to an overestimation of their wet removal, uncertainties caused by this assumption would have little influence on the conclusions of this study. The good agreement of the TEBC values between the simulation and the observations also supports the validity of the simulations.
