Distinct effects of winter monsoon and westerly circulation on dust aerosol transport over East Asia

The transport of dust aerosol in East Asia is affected by the East Asian winter monsoon (EAWM) and westerly circulation both for modern and geological periods. There are obvious seasonal changes in the intensity and range of EAWM and westerly jet; however, their impacts and relative contributions to East Asian dust transmission are still unclear. In this study, we use Regional Climate Model 4 (RegCM4) to simulate the changes in the East Asian dust cycle under present conditions, assessing the effects of EAWM and westerly jet on dust transport. The results show that the dust at the upper level is mainly transported by the westerly circulation, while that of the lower layer is mainly transported by the EAWM. In March, the westerly jet is located on the south side of the Tibet Plateau, and the high-level dust aerosol is transmitted eastward to the northern Pacific. Low-level dust is transmitted to the southeastern China with the influence of EAWM. With the northward shift of the westerly jet, the control range of the westerly winds increases in May and their correlations are weakened. In contrary, the impact of EAWM on the lower layer dust is enhanced. Due to the strengthened interaction between the westerly winds and the EAWM, they can both affect the middle-level dust transmission. The effect of EAWM is sensitive to the dust particle sizes. Under the action of EAWM, fine-grained dust is transmitted far away, while coarse-grained dust is limited to the vicinity of the source area. Once the dust is carried to the westerly layer, the influence of westerly winds on the transmission of different particle sizes dust is similar.


Introduction
Mineral dust, mainly from desert areas, is an important component of tropospheric aerosols (Ridgwell 2002) and has a significant impact on air quality and human health. Dust aerosol changes the climate by scattering and absorbing solar radiation (Song et al. 2014;Wang et al. 2014) or by changing the number of cloud droplets and the lifetime of the cloud, influencing the short-wave radiation and the amount of precipitation (Twomey 1977;Albrecht 1989;Li et al., 2011a;Liu et al. 2011;Ding et al. 2013;Yang et al. 2013;Ji et al. 2011). Dust connects different earth spheres through emissions, translations, and depositions (IPCC 2013). The long distance transport of dust aerosol affects the deposition of terrestrial loess and oceanic chemical cycles, and iron in dust has an essential impact on the exchange of ocean-atmospheric CO 2 (Cassar et al. 2008). Therefore, researches on dust aerosol become a global hotspot in past decades.
East Asia dust is an important component of global dust cycle, which accounts for about 11% of the total global emission and 6% of the total load in the atmosphere (Tanaka and Chiba 2006;Shao et al. 2011;Hu et al. 2019). Investigations of East Asia dust deposition indicate that 30% of the dust in the atmosphere settles in the desert, 20% in the mainland China, and 50% transports to the Pacific and other places across long distances (Zhang et al. 1997). Observations, numerical simulations, and satellite imagery shows that dust in East Asia can be transmitted to the North Pacific Ocean and western coast of the USA (Holzer et al. 2003;Tanaka and Chiba 2006). The Taklimakan Desert and the Gobi Desert are two major dust source areas in East Asia (Sun et al. 2001;Chen et al. 2014;Abulaiti et al. 2014; Lee and Kim 2012;Li and Zhang 2012), which account for 70% of total Asian emissions . The dust from Mongolia and China are the primary sources of dust deposits in the Loess Plateau and southeastern China (Sun 2002;Shao et al. 2003;Tan et al. 2012). The dust of Taklimakan Desert can be carried to higher than 5000 m and then transported over long distances. They are an essential source of eolian sediments in the North Pacific (Shi and Liu 2011).
East Asian monsoon and westerly circulations have an impact on the eolian dust transmission, respectively. The lowlevel dust concentration can be affected by the EAWM and the strong winter monsoon transports dust away from the source area (Lou et al. 2016;Zhang et al. 2010;Yan et al. 2011;Zhu et al. 2012;Wang et al. 2018;Li et al. 2016a, b). Hao et al. (2006) found that there is always higher AOD over the South China Sea in winter with strong winter monsoon, which is conducive to transporting aerosols from the land. Both observations and numerical simulations suggest that the seasonal variation of aerosol concentrations over eastern China has a strong negative correlation with the summer monsoon (Yumimoto and Takemura 2015;Niu et al. 2010;Qin et al. 1997). While the monsoon circulation controls the regional scale transport of Asian dust, the global scale transport is mainly realized by the westerly circulation (Guo et al. 2017;Uno et al. 2009). The strength of the westerly winds and the position of large-scale meteorological features were related to the dust concentrations over eastern Asia (Wu et al. 2017;Gao et al. 1992). The mid-latitude westerly winds are considered an essential factor in North American dust from northwestern China . By simulating the transmission path of dust during a dust storm, Guo et al. (2017) proved that the dust on the upper troposphere is mainly transmitted along with the westerly circulation.
Marine and loess deposits, as the vehicles of past climate, can reverse inference the changes in circulation strength during geological periods (Yang et al. 2015;Wen et al. 2017;Sun et al. 2008;An et al. 2001;Porter and An 1995;An et al. 1991;Dong et al. 2017;Hovan et al. 1989). Meteorological observation on modern dust emissions and decomposition results of loess grain size data have suggested that two significant dynamics, including near surface northwesterly winter monsoon and high-altitude westerly jet, are responsible for the dust transportation and deposition processes in East Asia (Sun 2004;Prins et al. 2007;Lin et al. 2016). Asian monsoon velocity changes may affect the loess grain size. When the Younger Dryas event occurs, the East Asian winter monsoon strengthened, causing the coarse dust, is carried and accumulated on the Loess Plateau (Yang and Ding 2014). During glacial intervals, the surface winter speeds are accelerated, and the particle size of the loess becomes coarse (Sun et al. 2008;Sun et al. 2012). During the cold period, strong westerly circulation brings the dust sediments to the Japan Sea (Dong et al. 2017). The intensity of the westerly circulation cannot only change the particle size of the sediment but also change the dust transport path (Nagashima et al. 2007). Therefore, to fully understand the effects of winter monsoon and westerly circulation on dust aerosol transport can provide theoretical support for paleoclimate dynamics.
In this study, we use a regional climate model experiment to evaluate the effects of EAWM and westerly winds on dust transport over East Asia with a special focus on their seasonal changes. We compare the contribution of monsoon and westerly winds to dust transport in different months (March, May, and July) as an analogy of warm and cold periods. The model and experiments are introduced in Section 2. The results, including the effects of winter monsoon and westerly circulation on dust aerosol transport over East Asia, are shown in Section 3. The discussion and conclusion are presented in Sections 4 and 5, respectively.

Model and experiments
The model used in this study is the Reginal Climate Modal version 4.1 (RegCM4.1), which is developed and supported by the National Center for Atmosphere Research (NCAR) and the International Center for Theoretical Physics. RegCM4.1 has primitive equation, hydrostatic, grid point limited area model, with σ-pressure vertical coordinate, and its dynamic framework is based on the hydrostatic core of the mesoscale model MM5. The Biosphere Atmosphere Transfer Scheme (BATS1e) is the land surface process schemes (Yang et al. 1997;Oleson et al. 2008), and the dust cycle diagnose can only use BATS1e (Sun et al. 2017). The coupling of the dust module is based on the dust emission mode (Alfaro and Gomes 2001).
The dust module of RegCM4.1 takes into account the dust emission, transmission, and dry and wet deposition of different particle sizes, which divides the dust into four particle sizes, i.e., the fine (0.01-1.0 μm), accumulation (1.0-2.5 μm), coarse (2.5-5.0 μm), and the giant (5.0-20.0 μm) modes. The description of the sanding process in the dust module is critical, which depends on the wind, soil properties, and particle size of the dust. The estimation of dust emission is mainly divided into the following three steps (Zakey et al. 2006). Firstly, it is determined whether the soil in the grid can be subject to wind erosion, which is depending on soil texture, soil type, particle size, and composition. Secondly, the friction velocity threshold of the wind erosion is calculated. When the friction velocity exceeds the threshold, the dust will leave the ground, and the dust emission can be regarded as a function of friction velocity (Marticorena and Bergametti 1995). Finally, the horizontal and vertical emissions of dust are calculated. The horizontal mass flux can be considered a function of the dust particle, and the vertical mass flux depends on the kinetic energy during dust emission.
In the experiment, changes in the East Asian dust cycle during 1988-2009 are simulated. To eliminate the influence of other aerosols, only the dust aerosol is included. The initial and boundary conditions are derived from NCAR/NCEP reanalysis data. The NOAA sea surface temperature data is chosen. The spatial resolution is set to 40 km, and the center of the domain is 32°N and 105°E, with 160 grids in the north-south direction and 240 grids in the east-west direction. The model is based on the standard σ coordinate system, which simulates 18 layers with the model top at 10 hPa. The model was integrated from January 1, 1988, to December 31, 2009, with the first 2 years as the spin-up time. The results of the last 20 years are analyzed.
In order to evaluate the model performance, we compare the simulated results with the ERA-Interim reanalysis from 1990 to 2009, which the spatial resolution is 2.5°× 2.5°. The westerly index over East Asia is defined by the averaged 200 hPa westerly wind velocity for 35-50°N and 80-110°E . The EAWM index is the averaged 850 hPa northerly wind speed for 35-45°N and 100-120°E, and the East Asia Summer Monsoon (EASM) index is the averaged 850 hPa southerly wind speed for 25-35°N and 100-120°E.To consider the seasonal changes of westerly jet and East Asian monsoon, the modeling results for dust transport process in March, May, and July are analyzed.

Results
The model performance of wind field is first validated by ERA-Interim reanalysis data (Fig. 1). The high-level westerly winds in March, May, and July are gradually weakened, and the westerly jet moves northwards. In March, the location of the westerly jet is at 30°N, and the maximum value is over the east of the Tibet Plateau (Fig. 1d). In May, the jet moves northwards from the southern part of the Tibet Plateau (Fig.  1e). The velocity decreases, and the position slightly moves to the north. In July, the jet moves to 45°N, and the speed is significantly weakened (Fig. 1f). In the low-level atmosphere, China and Mongolia are controlled by the EAWM in March, and the main wind direction is the northwesterly (Fig. 1j). In May, the northern China was still controlled by the winter monsoon, while the EASM develops over southern China (Fig. 1k). In July, the summer monsoon, prevailing by the southerly winds, controlled the whole East Asia (Fig. 1l). In the simulations, the seasonal changes in the westerly jet and monsoon are in agreement with the reanalysis data ( Fig. 1ac,g-i).
The validity of our simulation was examined by comparing the simulated aerosol optical depth (AOD) over East Asia with CALIPSO, as shown in Fig. 2. From March to May (Fig. 2 a  and b), the AOD over the Taklimakan Desert, the Gobi Desert, and the Loess Plateau gradually increased, while it decreased in July (Fig. 2c). In Central Asia, the AOD gradually increased from March to July, and AOD peaked in July, which is consistent with satellite data. In general, the model slightly overestimates the AOD over East Asia in May and July.
The spatial pattern of correlation coefficients between the westerly index and the 200 hPa westerly winds velocity is shown (Fig. 3). In March, positive correlation coefficients are higher than 0.5 in the north of Tibetan Plateau and northern China, which indicates that the westerly jet controls the wind speeds over this region (Fig. 3a). The range of the positive correlation area decreases in May and moves slightly to the north (Fig. 3b). In July, the positive correlation area is weakened and constrained to a shallow belt (Fig. 3c). For the low-level winds, the correlations between the EAWM index and 850 hPa meridional winds show that the East Asia is mainly affected by the winter monsoon in March and the correlation coefficients exceed 0.7 (Fig. 3d). The positive correlation coefficient in May is higher than 0.4, and the range becomes smaller than March (Fig. 3e). In July, the southern East Asia is mainly affected by the summer monsoon (Fig. 3f).
The winds are correlated with the dust AOD to examine the relationship between dust transport and westerly wind intensity. Figure 4 shows that in different months, the correlation coefficient distributions are different. In March (Fig. 4a), the positive correlation center is located over the Tibetan Plateau and east of source areas. It shows a band-like distribution from northern China to northwestern Pacific, which indicates that the intensity of westerly jet facilitates the westward and longdistance transport of dust. In May, when the westerly jet moves southwards, the positive correlation areas expand, but the correlation coefficients are smaller than March (Fig. 6b). In July (Fig. 4c), the jet retreats to the north side of the Tibetan Plateau, and the dust AOD does not significantly correlate with the westerly index. Since the prevailing wind direction of EAWM is northerly and northwesterly, the correlations between winter monsoon index and dust AOD are found over eastern China (Fig. 4d,e), which is distinctly different from those for westerly index. In May, the correlations are most significant with largest areas over eastern China when the summer monsoon starts to develop. When the summer monsoon matures in July, the southerly winds suppress the southward transmission of dust, and thus they mainly show negative correlations (Fig. 4f).
The correlation coefficient maps of the westerly index and monsoon index with the dust deposition fluxes are also shown (Fig. 5). Similar with the dust AOD, the correlations with deposition fluxes present that the westerlies dominate the westward transport and deposition of dust and the winter monsoon controls the northwestward ones. In May, the controlling areas of westerly index and monsoon index on dust transport are both larger than March although their intensity and correlations with wind speeds are smaller.
We express the interannual variation of the 1990-2009 index by a line chart and select the strong and weak years of the westerly and winter monsoon. Through composite analyses, the influence of westerly and monsoon on dust transmission is shown in Fig. 6. The AOD and dust concentration difference between the strong and weak westerly wind years in March shows (Fig. 6e,i) that there is a high-value area in the range of 30-45°N. And it presents a band pattern and extends from the source area to the Japan Sea, similar with Fig. 4a. In May, the area with a large difference in dust AOD and concentration (Fig. 6f,j) between strong and weak years is also consistent with the relevant high area in Fig. 4b. The EAWM   (Fig. 6g,k), the impact of the EAWM on dust is mainly confined to the East Asian continent. In May, the AOD and dust concentration difference between the strong and weak monsoon years shows that the dust can be carried to farther areas, such as the Japan Sea and the North Pacific (Fig. 6h,l).
To examine the relationships of dust deposition over specific regions, we select four representative regions: Japan Sea, Tibetan Plateau, Southern China, and Loess Plateau. The interannual variations of deposition fluxes and circulation index are calculated, as shown in Fig. 7. Over the Japan Sea which is located in the downwind region of the westerlies, the dust depositions are mainly controlled by the westerlies in March and May, no matter where the westerly jet is (Fig. 7a,b). The transport of dust over Tibetan Plateau is also obviously affected by the westerlies although the winter monsoon index might correlate to the westerly jet (Fig. 7c,d). The Tibetan Plateau is a dust source area and a channel for uplifting dust to upper troposphere; thus, the strong westerly winds will lead to increasing dust emission and settlement, especially in March (Liu et al. 2020;Xu et al. 2018). In contrast, the dust depositions over southern China and Loess Plateau have larger correlations with the winter monsoon intensity in March, and the contributions of westerlies increase in May when the jet shifts northwards (Fig. 7e,f,g,h).
The correlations of the circulation index and dry depositions of different dust bins are calculated to examine the sensitivity of dust sizes to the winds (Fig. 8). The positive correlation patterns of four particle sizes with the westerly index are similar (Fig. 8a-d); we believe that after the dust is lifted to the westerly layer, the strong westerly wind can carry the different sizes dust to the east. The positive correlation maxima for all the bins are located over Tibetan Plateau and northwestern Pacific. When the dust is carried into the westerlies, the particles in all sizes are transported eastward to northwestern Pacific. However, the correlation coefficients of EAWM show (c,f). The dotted areas indicate statistical significance with 95% (p value <0.05) from a two-tailed Student's t test Fig. 4 Distributions of correlation coefficients between westerly index and dust AOD and between monsoon index and dust AOD in March (a,d), May (b,e), and July (c,f). The dotted areas indicate statistical significance with 95% (p value <0.05) from a two-tailed Student's t test that the sizes are sensitive to the winter monsoon. When the size of dust particle is less than 1 μm (Fig. 8e), the EAWM can transport it to southern Japan and the South China Sea. As the particle size increases, the range of dust deposition gradually decreases, and most of the 5-20-μm dust settles in the Loess Plateau (Fig. 8h). Such patterns qualitatively indicate that the smaller the dust particle is, the larger areas the winter monsoon controls.
To examine the vertical distribution of eastward and southward transport of dust under the impact of westerlies and winter monsoon, respectively, Fig. 9 presents the correlation coefficient distributions of dust mixing ratio and circulation index in different months. In March (Fig. 9a), the dust of 80-90°E is quickly raised to 300 hPa and transmitted eastwards under the action of the westerlies wind. The controlling areas of westerlies occupy nearly the whole mid-to-high troposphere, from approximately 700 to 200 hPa. However, the correlation between the westerlies and the dust concentration is small and not significant over the near surface. In May (Fig.  9b), the dust transport associated with the westerly is suppressed due to the weakening of the westerly winds. The controlling areas are shifted downwards, which indicates that the westerlies affect the near surface transport at this time. For the southward transport by the winter monsoon, in March (Fig.  9c), the positive correlations between dust and westerly index are mainly concentrated below 850 hPa, showing that the winter monsoon merely controls the low troposphere since it is a shallow system. There are two related high-value areas at 38-41°N and 42-44°N, which correspond to the Taklimakan Desert and the Northern Fig. 6 The interannual variation of the westerly index and EAWM index in March (a,b) and May (c,d); the dotted line represents the standard deviation; the red and blue dot represents the weak and strong year of westerly or monsoon. Horizontal distributions of differences dust AOD (d-h) and tracers mixing ratios (i-l) between most energetic and weak westerly or monsoon years (strongest-weakest) in March and May The winter monsoon weakened in May, and the summer monsoon gradually increased (Fig. 9d). The dust is still affected by the winter monsoon and can continue to be transmitted upwards to 700 hPa. Thus, the influence of the westerly winds extends downwards, and winter monsoon rises to the upper level in May compared to a colder time of March.
Spatially, the westerly winds at 300 hPa and 550 hPa facilitate a strip of eastward dust transmission (Fig. 10a,b), consistent with the vertical distribution (Fig. 9a). At the lower levels (Fig. 10c), the strip shapes become not significant and disappear, which indicates that the influence of westerly winds is weakened. Unlike March, the correlation coefficient distribution in May shows strip patterns in the mid to low troposphere (Fig. 10e,f). For the monsoon, the EAWM cannot affect the dust transport above the mid-level in March and May (Fig. 10g,h,j,k). On the 850 hPa level (Fig. 10i), the positive correlation coefficient between EAWM and dust concentration is significant over eastern China. In May, the correlation coefficients decrease compared to March (Fig. 10l).
We selected two regions, the Japan Sea and South China, to calculate the percentages of dust content of different levels. Fig. 8 Distributions of correlation coefficients between westerly index and dust deposition of four particle sizes (a-d) and between monsoon index and dust deposition of four particle sizes (e-h) in March. The dotted areas indicate statistical significance with 95% (p value <0.05) from a two-tailed Student's t test Fig. 9 Pressure-longitude cross sections over 35°N for the correlation coefficient distributions between westerly index and tracers mixing ratios (left column) and averaged over 100-120°E for the correlation coefficient distributions between WAWM index and tracers mixing ratios (right column). The dotted areas indicate statistical significance with 95% (p value <0.05) from a two-tailed Student's t test The Japan Sea is located in the downwind region of the westerly winds, while South China is mainly affected by the EAWM. In Fig. 11, the dust contents in Japan are largest in the mid-troposphere, which reach approximately 18% of the whole layer. The ratios are reduced obviously to both the low and high troposphere. The peak of vertical distribution of dust over South China is located at low troposphere, lower than that over Japan Sea, which supports that the dust transport levels by the westerlies are relatively higher than those by the winter monsoon.

Discussion
Our analysis based on modeling data shows that the dust is raised to the upper layer and effected by the westerly jet for long distance transportation. In March, the westerly winds are strong, and the dust tends to be lifted to mid-to-high troposphere under the influence of westerly jet and delivered eastwards to the Pacific Ocean in a strip. Once the dust is lifted to the mid-upper levels Fig. 10 Distributions of correlation coefficients between westerly index and tracers mixing ratios (a-f) and between monsoon index and tracers mixing ratios (g-l) in March (first and third row) and May (second and fourth row). The dotted areas indicate statistical significance with 95% (p value <0.05) from a two-tailed Student's t test Fig. 11 Percentage of vertical dust content. The solid line is the dust content over the 35-45°N and 135-140°E area, and the dotted line is the dust content over the 25-30°N and 110-120°E area and carried into the strong westerly jet, it would be easy to be carried by a long distance. Uno et al. (2009) used the LIDAR data analysis to show that the dust was lofted to the upper troposphere around 8-10 km above the Earth's surface and transported around the globe. Further researches showed that dust aerosols originated from the Taklimakan and Gobi Desert; they were mainly located in the latitudes range from 1 to 7 km over the source region and then ascended to 2-9 km over the Pacific Ocean (Guo et al. 2017;Chen et al. 2017), and finally reached North America Holzer et al. 2003). The East Asian monsoon has an impact on the local transmission of dust over this region (Gong et al. 2006;Guo et al. 2014;Li et al. 2015;Zhao et al. 2013). The transmission of dust by the winter monsoon is mostly over the tropospheric bottom layer, which is in agreement with a previous study (Lou et al. 2016).
The movement of the westerly jet has a significant relationship with the dust transport path and sediment area in East Asia. Dong et al. (2017) proposed that the influence scope of westerly wind is broad when the westerly jet is located on the southern side of the Tibetan Plateau. Comparatively, our results indicate that the influent area is narrower when the westerly jet stay in the south of the Tibetan Plateau; and at this period, the deposition of the Japan Sea is mainly carried by the westerly winds, which could be supported by numerous observations (Wan et al. 2012;Shen et al. 2017;Dong et al. 2017;Hovan et al. 1989). Geological sediments show that strong westerly winds are closely related to higher dust flux distribution and accumulation rate; dust deposits in some cold periods could prove this conclusion, such as Last Glacial Maximum, Heinrich Event, and the Younger Dryas Event. Since dust is transported to the Japan Sea mainly via the westerly jet, the gradual increases of dust grain size during interstadials were mainly reflected by the increasing wind speed of the westerly jet. The estimated mass accumulation of aerosol dust exhibited two distinct peaks at 17.6 ka BP and 11.4 ka BP, which are coincident with Heinrich Event 1 and the Younger Dryas, respectively (Lu et al. 2013;Sun 2004;Irino and Tada 2003;Schiemann et al. 2009;Mahowald et al. 1999;Yokoyama et al. 2006). Our simulation results show that the dry deposition in the Loess Plateau has high correlation coefficient with the winter monsoon index. Paleoclimate studies use the dust particle size as a proxy of the winter monsoon intensity; they believe that the strong winter monsoon can transmit more dust to East Asian continent Xu et al. 2012;Sun et al. 2012;Sun et al. 2010;Jiang and Ding 2010;Sun 2004). Further analysis show that the strong EAWM can transport more dust with larger particle sizes to the southeast of source area. When the winter monsoon intensity increases, the particle size of the dust deposition become larger (Sun et al. 2010;Zhang et al. 2002), the dust flux and accumulation rates become higher (Kang et al. 2015;An et al. 1991), which was also identified by a modeling study (Shi and Liu 2011).

Conclusions
East Asia is typically affected by the combined westerly circulation and East Asia monsoon. From colder to warmer seasons, the westerly jet gradually moves northwards, which induces different patterns of atmospheric circulation and dust transport in the troposphere. In this study, a remarkable seasonal feature of dust transport is highlighted by the westerly winds and winter monsoon, respectively. In March, the westerly winds transport dust eastward until it reaches to the Pacific Ocean. The winter monsoon mainly transmits dust to the southeast. Westerly circulation controls the high-level dust transmission, and the monsoon controls the low-level transmission. The influence of westerly winds on the transmission of different particle sizes dust is similar, while the monsoon can transport fine-grained dust to a distant area, and the coarse-grained dust can only be settled in the near-source area. The relationship between the dust transmission and the atmospheric circulation is very complicated, which is worthy of further researches.