A Possible Linkage between Dust Frequency and the Siberian High in March over Northeast Asia

: Spring dust frequency in northeast Asia has been investigated using various approaches to understand the mechanisms of dust emission and transport. However, little attention has been paid to the linkage between dust activity and the Siberian High (SH), particularly when the SH pressure system is highly variable. In this study, we characterize the possible physical mechanisms of dust emission and transport associated with the Siberian High Intensity (SHI) and Siberian High Position Index (SHPI) in March using 18 years of ground-based observations and reanalysis data. We found that when the SHI was strong and the SH’s center was farther east (“Strong–East period”), surface and atmospheric temperatures were cooler than when the SHI was weak and the SH’s center was farther west (“Weak-West period”), due to anomalous anticyclonic pressure and strong easterlies. As a result, a reduction in the meridional temperature gradient in the lower atmosphere suppressed dust emission and transport, due to stagnant atmospheric conditions. This anomalous anticyclonic pressure in the Strong-East case seems to reduce the development of extratropical cyclones (ETC) in northeast Asia, leading to a less effective dust transport. A case study with composite analysis also showed a similar physical mechanism: stagnant air accompanying weakened westerlies in the Strong-East period suppressed dust transport to South Korea. Our ﬁndings reveal that the intensity and position of the SH can be utilized to identify spring transboundary air pollutants in northeast Asia.


Introduction
Yellow sand, or Hwang-sa in Korean, is a frequent meteorological phenomenon in which sand dust is suspended in the air and gradually falls to the surface [1]. Although an artificial factor (e.g., urbanization) may affect dust emission in northeast Asia [2,3], the yellow sand is mainly related to atmospheric circulation changes. Dust events that occur in northeast Asia generally originate from arid or semi-arid regions such as northern China, the Gobi desert in Inner-Mongolia, the Loess Plateau, and the Taklamakan desert [4][5][6][7]. Dust storms move to eastern China, the Korean Peninsula, and Japan with strong northwesterly winds induced by atmospheric baroclinic instability in the lower atmosphere [1,[8][9][10][11]. Dust aerosols contain both particulate matters (PM) with an aerodynamic diameter smaller than 10 µm (PM 10 ) and smaller than 2.5 µm (PM 2.5 ) [12,13]. PM 10 and PM 2.5 also contain other types of anthropogenic aerosols (e.g., black carbon); however, in springtime, high PM 10 and PM 2.5 concentrations are mostly observed due to yellow sand or dust storms [14][15][16][17][18]. The frequency of dust storms that originate in Mongolia varies seasonally: spring (61%), more frequent in this season. In addition, the SH still influences surface conditions and spring climate in East Asia.
The SH system can be characterized by the intensity (i.e., pressure) and position of the center of the SH: the SHI [55,65] and the SHPI [63] (see details in Section 2). In particular, the SHPI, proposed by Jia et al. [63], is a useful index to explain why the winter haze observed in January 2013 in north China was more severe than in south China. Thus, we hypothesize that the dust frequencies in dust source regions and the receptor area (e.g., downwind country like South Korea) are substantially affected by the SHI and SHPI characteristics. Our ultimate goal is to demonstrate the possible physical effects of the SH on dust frequency for a better understanding of dust emissions and transport mechanisms in springtime. Therefore, we present changes in dust frequency and cyclone frequency associated with synoptic meteorological patterns using both SHI and SHPI.

Study Domain
Our study area covers two dust source regions, R1 and R2, and a representative downwind located country, South Korea, in northeast Asia (Figures 1 and 2). Dust source regions were classified in terms of dust generation mechanisms and long-term trends from the several previous studies [8,66,67]. The region R1 (95 • -110 • E, 35 • -50 • N) represents arid areas of northern China and Mongolia, including the Gobi, Bardinjaran, Tanger, and Ulanbu deserts. The region R2 (110 • -125 • E, 40 • -50 • N) covers the Othindag and Hunshudak deserts in the middle of Inner Mongolia and northeastern China. Dust aerosols lifted up in R2 are often transported to the Korean Peninsula and Japan by strong winds and results in severe dust events and poor visibility [53]. Although the largest dust source region is located in northwestern China next to the western of R1 [46,67], we selected R1 and R2 to identify dust transport and incursion into South Korea in relation to the SH pressure system. Further details of the dust source regions are described in previous studies [46,53,68].
To explain the linkage between cyclones and dust activity associated with the SH variations, we analyzed the characteristics of the extratropical cyclones (ETC) in the dust source regions and the ETC domain (125 • -140 • E, 35 • -50 • N), i.e., the area over which ETCs pass, as defined by Lee et al. [49] (red box in Figure 1). We compared differences in the ETC frequency and intense ETC frequency over the dust source regions and the ETC domain for the Strong-East and Weak-West periods).

Dataset
To explain dust frequency in relation to the characteristics of the SH, we used two SH indices: the SHI as a measure of the intensity (or strength) of the SH and the SHPI as an indicator of the location or position of the SH's center [63,65]. The domains for calculating the SHI and SHPI are shown in Figure 1. Originally, the SHI was defined as the strength of the SH pressure system over 80°-120° E and 40°-65° N during the wintertime (December-January-February) [54,55,70]. In our study, we used the average mean sea level pressure (MSLP) in March. The SHPI represents the position of the SH in an eastwest direction over 60°-145° E, 30°-60° N, an area including not only dust source regions but also downwind regions, including the Korean Peninsula and Japan ( Figure 1). We calculated the SHPI for March over the same domain proposed by Jia et al. [63]. A higher SHPI value implies that the center of the SH is shifted to eastward of its climatological position (average of 18 years).
In this study, we defined dust frequency as an indicator of dust emission in the source region and transport to downwind region. Here, dust transport means dust incursion into

Dataset
To explain dust frequency in relation to the characteristics of the SH, we used two SH indices: the SHI as a measure of the intensity (or strength) of the SH and the SHPI as an indicator of the location or position of the SH's center [63,65]. The domains for calculating the SHI and SHPI are shown in Figure 1. Originally, the SHI was defined as the strength of the SH pressure system over 80 • -120 • E and 40 • -65 • N during the wintertime (December-January-February) [54,55,70]. In our study, we used the average mean sea level pressure (MSLP) in March. The SHPI represents the position of the SH in an east-west direction over 60 • -145 • E, 30 • -60 • N, an area including not only dust source regions but also downwind regions, including the Korean Peninsula and Japan ( Figure 1). We calculated the SHPI for March over the same domain proposed by Jia et al. [63]. A higher SHPI value implies that the center of the SH is shifted to eastward of its climatological position (average of 18 years).
In this study, we defined dust frequency as an indicator of dust emission in the source region and transport to downwind region. Here, dust transport means dust incursion into South Korea. To calculate dust frequency, we first selected the good quality of the surface synoptic observations (SYNOP) measured in the dust source regions and South Korea (Figure 2a,b). The SYNOP dataset was obtained from the external application programming interface service webpage of the Korea Meteorological Administration (KMA) [71]. The quality criteria required that 3-hourly weather observations were made at least four times a day and that more than 330 days of daily data were available at the station. The same standard was also used by Lee et al. [53]. Next, data at each station were categorized into two values: 1 if dust was observed at least once per day from the eight 3-hourly interval daily observation data and 0 if no dust event was observed. To select the day having the dust events (i.e., dust storms and blowing sand), we applied the same dust weather codes used in previous studies [40,72]. By selecting only category 1 data, we obtained the total dust recorded by SYNOP on each day for each of R1, R2, and South Korea. Then we summed these number of stations during March from 2000 to 2017 for each of R1, R2, and South Korea. Thus, we can define the dust occurrence frequency (DOF) as the average number of dust-observing weather stations (SYNOP) in the two dust source regions by year as follows: where Ns is the total number of the SYNOP data and DD is the total number of dust days at the ith station in March over all years. We defined a different dust frequency for South Korea as the dust event frequency (DEF), because dust days in South Korea are events or episodes. South Korea is located in downwind side and it is a dust receptor rather than a dust emission area. Therefore, the DEF indicates dust incursions from the source regions (i.e., R1 and R2). The calculation procedure for the DEF was the same as for the DOF. The total number of the SYNOP data (Ns) were 106 in dust source regions (51 for R1 and 55 for R2) and 36 for South Korea (Figure 2a,b).
In this study, we used the SYNOP for calculating the DOF and DEF. One can also use the surface PM 10 measurements; however, high PM 10 days do not always reflect the dust emission days or dust episodes, because they were affected by different synoptic weather patterns [73]. During the yellow sand events, a well-developed cyclone passed through the dust source regions (Mongolian deserts and northern China) and moved eastward over time. On the contrary, high-PM 10 was modulated by relatively weaker cyclonic anomalies compared to the yellow sand events. Therefore, the DOF and DEF obtained from the SYNOP would give more direct information of dust activity than the surface PM 10 measurements.
To investigate the changes in atmospheric circulation in northeast Asia in response to the SH variations, we used the ERA-Interim archive Version 2.0 reanalysis dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) [74]. Meteorological variables used for the analysis were the mean sea level pressure (MSLP), geopotential height (Z), temperature (T), zonal wind (u), and meridional wind (v) with a horizontal resolution of 0.75 • × 0.75 • on a longitude-latitude grid. To understand the features of dust transport to South Korea, we calculated the average of the MSLP and 850-hPa wind components at every 00 h (Coordinated Universal Time, UTC) for selected 12 dust days. The 12 high dust emission days for the Strong-East and Weak-West periods are listed in Table S4.
To examine the effects of snow cover on spring dust emission, we used monthly Snow Cover Extent (SCE) data version 1 from the National Oceanic and Atmosphere Administration (NOAA) Climate Data Record (CDR) [75]. Since June 1999, the satellite-derived weekly SCE product has been available at monthly resolution with National Meteorological Center (NMC) grid coordinates (88 × 88 cells) over the Northern Hemisphere on a polar stereographic projection. The SCE data for each grid cell are stored as values from 0 to 1, presenting the snow cover fraction. In our study, we converted the SCE to a percentage by multiplying by 100.
We used daily ETC data computed from the National Centers for Environmental Prediction (NCEP) and National Center for Atmospheric Research (NCAR) Reanalysis dataset [76]. The spatial resolution of the ETC data is of 2.5 • × 2.5 • , which is lower than the ERA-Interim reanalysis data. To calculate the ETC frequency and intense ETC frequency, we first selected data where the central pressure was lower than 1010-hPa and the Laplacian of the pressure was greater than 10 mPa/km −2 , respectively, and then calculated monthly data.

Data Analysis
We performed a composite difference analysis to understand the changes in dust frequency, and the meteorological patterns associated with different SH characteristics. Interannual variations of the SHI and SHPI were high, and showed a strong positive correlation coefficient (r = +0.61, p-value < 0.01) during the study period ( Figure 3a). In contrast to March, the correlation relationship between the SHI and SHPI in winter was weak (r = −0.29, p-value = 0.2519 for January, r = +0.26, p-value = 0.3112 for February) [63].
Based on the good correlation between the SHI and SHPI (Figure 3a), we used both indices to explain differences in dust frequency (i.e., DOF and DEF) and cyclone frequency (i.e., ETC frequency and intense ETC frequency) (Table S2). Here, "Strong" and "Weak" indicate the power of the SHI, whereas "East" and "West" denote the eastward and westward location of the SH's center. We selected the top three years having both high SHI and high SHPI values (2005, 2010, and 2012) ( Figure 3, Table S1). The average of datasets for three years is called the "Strong-East" period. To make composite difference, we selected the bottom three years having both low SHI and low SHPI values (2004, 2006, and 2013). The average of datasets for three years is called the "Weak-West" period ( Figure 3, Table S1). For the analysis, DOF, DEF, meteorological variables (i.e., mean sea level pressure: MSLP, 850-hPa geopotential height: Z 850 , 850-hPa horizontal wind vectors: WV 850 , 850-hPa horizontal wind speed: WS 850 , 2-m temperature: T sfc ), the ETC frequency, and intense ETC frequency were averaged for the selected three composite years. We checked the differences of meteorological patterns with the five samples (i.e., the top(bottom) five years having both high(low) SHI and high(low) SHPI values) and performed the Student's t-test. However, the difference of those patterns was not substantial as much as for the three samples; we decided to show the composite differences with three samples.

Results and Discussion
We examine the general synoptic meteorological characteristics of the SHI and SHPI. Then, we demonstrate a possible physical mechanism to explain differences in the DOF, DEF, ETC frequency, and intense ETC frequency associated with synoptic meteorological characteristics over the dust source regions and South Korea. Lastly, we show a case study to investigate the effects of the intensity and location of the SH on dust incursion into South Korea. Note that in this section, "dust source region" refers to R1 and R2.

Synoptic Meteorological Characteristics of the SHI and SHPI
To understand the climatological pressure and wind patterns in March, we present the 18-year average of MSLP and wind fields (WV850, WS850) and for the composite periods (i.e., Strong-East and Weak-West) (Figure 4). The averaged MSLP and wind patterns over  Tables S1 and S2. To better understand the dust incursion process into South Korea, we also performed a case study by selecting twelve days when high dust events occurred over the combined R1 and R2 region. We analyzed the composite difference in meteorological patterns and counts of dust-observing SYNOP for three consecutive days (Day + 1, Day + 2, Day + 3) after the start of dust emission day (Day + 0) between the Strong-East and Weak-West periods. We selected twelve days of high dust days for each Strong-East and Weak-West period (Table S4). Then we averaged twelve cases of meteorological variables and counts of SYNOP for each Day + n.
We performed the data analysis in March, 2000-2017, for the two reasons: (1) March is the month when the SH pressure weakens and fluctuations in location and intensity are greater than in other seasons in northeast Asia. (2) March is the beginning of spring dust events in South Korea, and the importance of dust prediction increases. In general, the impact of spring dust is more frequent and intense in April than in March, particularly in the downwind region [1,77]. However, we excluded data in April from the analysis to avoid discontinuity of the SHPI values during this month.

Results and Discussion
We examine the general synoptic meteorological characteristics of the SHI and SHPI. Then, we demonstrate a possible physical mechanism to explain differences in the DOF, DEF, ETC frequency, and intense ETC frequency associated with synoptic meteorological characteristics over the dust source regions and South Korea. Lastly, we show a case study to investigate the effects of the intensity and location of the SH on dust incursion into South Korea. Note that in this section, "dust source region" refers to R1 and R2.

Synoptic Meteorological Characteristics of the SHI and SHPI
To understand the climatological pressure and wind patterns in March, we present the 18-year average of MSLP and wind fields (WV 850 , WS 850 ) and for the composite periods (i.e., Strong-East and Weak-West) (Figure 4). The averaged MSLP and wind patterns over the 18-year period showed a strong anticyclone and strong westerlies over the Asian continent (60 • -120 • E). Strong northwesterly winds flowed from region R2 to the Korean Peninsula (Figure 4a). The area with pressure exceeding 1022 hPa in the study domain showed a distinct difference between the Strong-East period and West-Weak period (Figure 4b,c). The overall features of the MSLP and 850-hPa wind patterns for the Weak-West period occupied a smaller area or had weaker intensities than for the Strong-East period. For instance, compared to the 18-year meteorological patterns, the area affected by high MSLP (>1022 hPa) for the Strong-East period occupied over the full width of the SHPI calculation domain (Figure 4b). In contrast, in the Weak-West period, it was narrower in both latitude and longitudinal directions, and only the western edge of R1 was covered by strong high pressure. Compared to the Weak-West period, wind vectors flowing over the eastern edge of the high pressure (1022 hPa isobar) during the Strong-East period were stronger and moved northward (Figure 4b,c).
There are apparent differences in SHI and SHPI between the Strong-East and Weak-West periods. In the Strong-East period, the maximum value of the SHI was greater than 1027 hPa, locating within the range 45 • -52.5 • N and 85 • -98 • E. The averaged SHPI was 99.1 • E, shifted approximately 10 • eastward of the 18-year climatological SHPI (90.46 • E). The SHI was higher than 1020 hPa over a wide area of southern China where the dust source and downwind regions are situated, and this pattern extended farther north and east in the 18-year climatology. In the Weak-West period, the average SHPI was 79.6 • E, approximately 10 • westward compared to Figure 4a. Areas affected by a SHI greater than 1020 hPa occupied a smaller region than the SHI calculation domain. In the Weak-West period, the SHI in the dust source region decreased, but westerly winds blew over a greater area compared to the climatology and the Strong-East period.
Overall, the synoptic meteorological patterns for the Strong-East (Weak-West) cases were a mix of the features of the Strong (Weak) SHI (i.e., SHI-only) and East (West) SHPI (i.e., SHPI-only) (Table S2, Figure S1). Synoptic meteorological patterns during the study period showed similar features as those of the winter SH. Generally, there is a high pressure system characterized by the north-south pressure gradients and prevailing northwesterly winds over 40-60 • N in the central Asian continent [55,65].
(>1022 hPa) for the Strong-East period occupied over the full width of the SHPI calculation domain (Figure 4b). In contrast, in the Weak-West period, it was narrower in both latitude and longitudinal directions, and only the western edge of R1 was covered by strong high pressure. Compared to the Weak-West period, wind vectors flowing over the eastern edge of the high pressure (1022 hPa isobar) during the Strong-East period were stronger and moved northward (Figure 4b,c). There are apparent differences in SHI and SHPI between the Strong-East and Weak-West periods. In the Strong-East period, the maximum value of the SHI was greater than 1027 hPa, locating within the range 45°-52.5° N and 85°-98° E. The averaged SHPI was 99.1° E, shifted approximately 10° eastward of the 18-year climatological SHPI (90.46° E). The SHI was higher than 1020 hPa over a wide area of southern China where the dust source and downwind regions are situated, and this pattern extended farther north and east in the 18-year climatology. In the Weak-West period, the average SHPI was 79.6° E, approximately 10° westward compared to Figure 4a. Areas affected by a SHI greater than 1020 hPa occupied a smaller region than the SHI calculation domain. In the Weak-West period, the SHI in the dust source region decreased, but westerly winds blew over a greater area compared to the climatology and the Strong-East period.

Effects of the SHI and SHPI on Dust Emission in the Source Region
In the Strong-East period, both dust emission and the incursion into South Korea were reduced compared to the Weak-West period (Table 1). Although the composite differences of the DOF and DEF between the Strong-East and Weak-West periods did not appear considerable, results show a possibility that a strong and eastward shifted SH may reduce dust frequency over the dust source region and South Korea. Overall, the synoptic meteorological variables shown in Figure 5 for the SHI and SHPI combined had a weaker magnitude and showed a more central location of the SH pressure compared with the SHI-only and SHPI-only cases ( Figure S1). This may cause the small reduction in u for the Strong-East period compared to the East-SHPI period (Figure 6c, Figure S3c). Since there was a clear significant difference in MSLP between the Strong-East and Weak-West periods (Figure 4b,c), this may lead to changes in synoptic meteorological patterns. Consequently, DOFs for the East-SHPI and Strong-SHI periods may be smaller than that for the Weak-SHPI and Weak-SHI periods (Table S3).
the Korean Peninsula (>42° N). In addition, SCE remained above 30% in the source region (mostly in R2) (Figure 5d). This means that cooler surface temperatures may keep snow on the ground for a longer time period or generate good conditions for snowfall in the dust source regions. Zou [78] reported that a decrease of surface temperature in the dust source areas resulted in more snow cover on the surface, leading to much lower dust frequency in eastern China.  Figure 1. Black dotted areas (yellow dots in Figure 5d) indicate the differences are statistically significant above the 90% confidence level based on the Student's t-test.   To investigate the possible physical impact of the SH characteristics on the DOF, we analyzed composite differences of synoptic meteorological patterns. The results showed that during the Strong-East period, the significant anomalously high MSLP resulted in cold air incursions from high latitudes (>60 • N) to the dust source region locating around 45 • N (Figure 5a). A strong pressure gradient induced prevailing northwesterly winds mainly in western China. Although the magnitude of WS 850 over the source region was smaller than 2 m s −1 , the overall anomalous prevailing winds were easterlies (Figure 5a). This implies that the dust-laden airmass had less chance of being transported to South Korea due to weakened westerlies. In contrast to the strong anomalous easterly winds in R1, there were relatively weaker easterlies in R2 and stronger anomalous southeasterly wind patterns (Figure 5a,b). In the Strong-East period, wind speeds in both the lower atmosphere ( Figure 5b) and the near-surface ( Figure S2) were reduced by approximately 2.5 m s −1 , mostly in R1. Previous studies have suggested that low wind velocities near the surface and in the lower atmosphere generated a unfavorable condition for dust updrift in the dust source regions [39,40,44]. Weakened westerlies accompanied by air in the Strong-East period may generate less favorable conditions for bringing dust aerosols up to the lower atmosphere. Consequently, dust incursion into South Korea may be reduced. A similar mechanism was reported by Kusaki and Mikami [40]. They found that dust frequency and strong wind speeds (>6.5 m s −1 ) showed a strong positive correlation. Although the wind speed decrease found here was smaller than in the previous study, our result nevertheless shows that decreased DOF may be associated with changes in wind speed.
We also found that an anomalously high MSLP reduced the surface temperature by 3-4 • C in the dust source region, particularly in the south of R2 (Figure 5c). Temperatures at the surface and in the lower atmosphere in R1 and R2 were lower than in the north of the Korean Peninsula (>42 • N). In addition, SCE remained above 30% in the source region (mostly in R2) (Figure 5d). This means that cooler surface temperatures may keep snow on the ground for a longer time period or generate good conditions for snowfall in the dust source regions. Zou [78] reported that a decrease of surface temperature in the dust source areas resulted in more snow cover on the surface, leading to much lower dust frequency in eastern China.

Effects of the SH Variations on Dust Incursion into South Korea
Similarly to the DOF, the composite mean in DEF showed that dust incursion into South Korea was reduced in the Strong-East period (Table 1). Anomalous high pressure and easterly anomalies in South Korea were very weak, and a reduction in WS 850 was prominent (Figure 5a,b). Weaker wind speeds in the lower atmosphere (WS 850 ) contrasted with stronger wind speeds near the surface (WS 10 ) (Figure 5b, Figure S2). The decreases in both MSLP and westerlies were statistically significant over the northern part of the Korean Peninsula; thus, they are likely responsible for the smaller transport of dust aerosols to South Korea, as indicated by the decreased DEF. This may be related to the relatively weaker impact of wind speed reduction on the DEF compared to the DOF in the Strong-East period. The difference in the DEF for the SHI-only case was greater than the SHPI-only case (Table S3). WS 850 was reduced considerably for the SHPI-only case, particularly over the Korean Peninsula ( Figure S1). However, the decrease in WS 10 was relatively small in South Korea for both SHI-only and SHPI-only cases ( Figure S2). We also found that the DEF had low correlation coefficients (r in the range of −0.38-−0.36) with the SHI and SHPI (Figure 3). Since the DEF was defined for South Korea only, the discrepancy between the DEF and wind patterns associated with the SH variations may be better identified using the column-based dust aerosol and another type of reanalysis data [63,77].

Possible Effects of Cyclones on Dust Activity
We investigated the differences in the ETC frequency and intense ETC frequency over the dust source regions and the ETC passage domain (red box in Figure 1). In both dust source regions and the ETC passage domain, the ETC frequency and intense ETC frequency were higher for the Weak-West period than for the Strong-East period (Tables 2 and 3). During the Strong-East period, there were 14 fewer ETC cyclones and 10 fewer intense ETC cyclones in the dust source region than for the Weak-West period. This implies that during the Strong-East period, it is likely that cyclone development was suppressed compared with the West-Weak period in both the dust source regions and the ETC passage domain. Although the differences in the ETC frequency and intense ETC frequency in the ETC passage domain between the Strong-East period and Weak-West period were minor, it reflects that the ETC activity in both dust source and receptor areas can be connected.   16 21 Spring dust events observed in the Korean Peninsula were affected by the characteristics of the ETCs (e.g., frequency, intensity, direction, lifetime, etc.) [49]. This study [49] showed that the characteristics of the ETCs affecting springtime dust events in Korea are generated in the Altai-Sayan mountain region located at the border of Mongolia, Russia, and Kazakhstan. The ETCs passed through northeast Asia from the dust source regions via the northwestern Korean Peninsula. As we have shown in Figure 4 and Table 1, anomalously high pressure systems generated stagnant air in R1, R2, and South Korea, and this may be linked with the inhibition of cyclogenesis in springtime for the strong and eastward shifted SH (Tables 2 and 3).
Strong cyclones associated with cold fronts are responsible for dust weather in spring. An increase of near-surface wind speed in northern China triggers spring dust storms in arid regions, especially when the cold fronts of cyclones pass over. Thus, dust aerosols are more effectively lifted and transported to downwind countries with cyclones and westerly jets [73,79,80]. Therefore, it is likely that the less frequent dust activity for the Strong-East period is associated with the decrease in the cyclone activity (e.g., ETC frequency and intense ETC frequency).

Case Study
Dust events in South Korea are generally affected by synoptic meteorological conditions in the dust source regions 2-3 days before the dust incursion into South Korea [1,81]. Thus, we selected 12 days of high dust episodes in R1 and R2 regions to examine dust incursion into South Korea after dust emission (Table S4). Consistent with the synoptic meteorological patterns shown in Figure 4, both dust emission and transport decreased during the Strong-East period compared to the Weak-West period (Figures 7 and 8). During Day + 0 of the Weak-West period, R1 was centered between high and low pressure systems, and this feature remained until the next day (Figure 7). The size and shape of the SH systems were generally similar on Day + 1. As the strong SH system moved eastward, the area affected by high pressure stretched over the Yellow Sea on Day + 2. Strong pressure gradients nearby if 100 • E indicated a cold front that affected the Korean Peninsula mainly on Day + 1 and Day + 2, then it passed over Japan on Day + 3. At the 850-hPa, where dust transport is taking place in general, we confirmed that strong low pressure and anomalous easterly wind patterns were found over the boundary of R2 and South Korea for the Strong-East periods (Figure 8). These synoptic meteorological features reveal that anomalous high pressure caused air stagnation so that less dust aerosols were transported to South Korea, similar to the climatological synoptic composite patterns (Figures 4-6). Overall, this feature agreed fairly well with the decrease of number of the SYNOP having dust weathers in the Strong-East period (Figure 9). During the Strong-East period, we found a distinct reduction of station numbers observing dust in both the dust source regions and South Korea compared to the Weak-West period.
In addition, we found similar features of differences in MSLP, Z 850 , wind patterns for the SHI-only and SHPI-only years ( Figures S4-S7). Additionally, the counts of SYNOP that observed dust for the SHI-only and SHPI-only groups ( Figure S8) were consistent with Figure 9. This implies that a strong and farther east SH pressure system suppresses dust activity in both the source region and South Korea ( Figures S4 and S5 and Figures 7 and 8). Weakened westerlies reduce airflow to the eastern boundary of the SH, which is likely to affect dust transport from the source regions to South Korea. Park et al. [82] found a longer duration of dust events over northeast China because of the strong SH system and strong pressure gradients between the eastward moving low pressure system and the winter SH pressure. This implies that the DOF can be increased when the SH system weakens and low pressure systems containing dust aerosols in the dust source regions slowly pass over the Korean Peninsula. Our findings from the case study and cyclone frequency support a possible mechanism of dust transport to the downwind area: A stronger intensity of the SH and farther eastward of the SH may reduce a speed of airflow incursion into South Korea. These synoptic meteorological conditions may affect to the weakened developments of cyclone frequency in the ETC domain (Table 3).
Dust aerosols over northeast Asia are generally observed between 1 and 3 km [83]. This height fluctuates, which means that dust aerosols travel above the boundary layer [83,84]. This suggests that deposition or sinking of dust aerosols to the surface may be affected by the atmospheric stability or synoptic meteorological conditions during the transport process. To confirm this hypothesis, one can analyze the high-resolution of the satelliteretrieved vertical distribution of dust aerosol products (e.g., aerosol optical depth or aerosol index) for the different SH characteristics and investigate the dust incursion process into South Korea.

Summary and Conclusions
We investigated the physical mechanisms of dust emission and transport processes for the different SH conditions at the start of the dust season in northeast Asia. Using the surface observations and a reanalysis dataset, we found a strong and eastward shifted SH suppressed both dust emission in the source regions and transport to South Korea. During the Strong-East period, an anomalously high anticyclonic circulation led to a cool surface and smaller north-south thermal gradient. Therefore, the thermal wind decreased, resulting in air stagnation and consequently less effective dust emission. Stable atmospheric conditions are likely to reduce cyclone development in both dust source and the downwind area. Since the SH's core was mostly situated in the dust source area during our study period, the effects of the SHI on dust transport to the downwind area appears to be more discernible than those of the SHPI.
A case study also revealed that suppressed dust incursion into South Korea can be explained by the characteristics of the Strong-East SH. Surface temperatures decreased due to the enhanced surface pressure. Strong anomalous easterlies in the source region and downwind area imply that during the Strong-East case, westerlies in the dust source region and South Korea become weaker than during the Weak-West case. Overall, synoptic weather patterns in the Strong-East case seem to produce stable air conditions that are unfavorable conditions for cyclone development.
Our study provides observational evidence that the early spring dust emission and incursion into a downwind country is less effective when the SH system is strong and extends farther eastward. Our findings imply that dust activity in early springtime can be explained by both the intensity and position of the SH, but that the SHI is a better index than the SHPI for diagnosing dust emission and dust transport. To obtain robust results, the next step should be to compare results with other types of reanalysis datasets. We also suggest that a further investigation of the vertical distribution of dust aerosols over the downwind country for the Strong-East and Weak-West periods would be helpful to elucidate how much dust sinks to the ground and how much remains in the mid-troposphere. The result of our study provides useful information on dust emission and transport for the seasonal transition period when the SH is highly variable and the weather/climate conditions are anomalously cold with weakened westerlies. Moreover, the SHI and SHPI can be utilized to examine the characteristics of migrating air pollutants from emission source regions to the receptor regions in northeast Asia.
Supplementary Materials: The following are available online at https://www.mdpi.com/2073-4 433/12/2/176/s1: Figure S1: As in Figure 5 but for composite differences in meteorological fields for the SHI-only (Strong-SHI minus Weak-SHI, left panels) and SHPI-only (East-SHPI minus West-SHPI, right panels) periods for March, 2000-2017. Note that the left and right panels have the same numberings. Figure S2: Composite differences in wind speed at 10 m (WS 10 ) between the Strong-East and Weak-West period for March, 2000-2017. Figure S3: Left panels show latitude-height cross sections of composite differences in (a) geopotential height (Z; m), (b) temperature (T; • C), and (c) zonal wind velocity (u; m s -1 ) between Strong-SHI and Weak-SHI periods, averaged over 95-125 • E, the longitude range of R1 and R2. Similarly, right panels show composite differences between East-SHPI and West-SHPI periods. Negative values are shown by dashed lines. Note that the left and right panels have the same numberings. Figure S4: Same as Figure 7 but for the SHI-only group (Strong-SHI and Weak-SHI). Figure S5: Same as Figure 7 but for the SHPI-only group (East-SHPI and West-SHPI). Figure S6: Same as Figure 8 but for the SHI-only group (East-SHI and West-SHI). Figure  S7: Same as Figure 8 but for the SHPI-only group (East-SHPI and West-SHPI). Figure S8: Same as Figure 9 but for the SHI-only group (a and b) in the upper panels and SHPI-only group (c and d) in the lower panels. Table S1: Years ranked by values of the Siberian High Intensity (SHI) and Siberian High Position Index (SHPI) in March 2000-2017. Years shown in bold were selected for composite difference analysis. Table S2: Years selected for the composite analysis based on Table S1. Table S3: Same as Table 1 but for the (a) SHI-only and (b) SHPI-only periods. Table S4: A list of high dust emission dates (YYYY.DD.MM) in the dust source region during the Strong-East and Weak-West periods. We selected 12 days over the combined R1 and R2 region.