Crop Residue Burning in Northeast China and Its Impact on PM 2.5 Concentrations in South Korea

: The impact of crop residue burning in northeastern China on South Korean PM 2.5 concentrations was assessed via weather conditions, air quality modeling (AQM), and PM 2.5 composition data during two cases exceeding 35 µ g · m − 3 in November 2015. PM 2.5 concentration simulations of Case 1 differed from observations by 3.7–17.6 µ g · m − 3 , overestimating the levels by 6–36%; however, Case 2 varied by 20.0–59.8 µ g · m − 3 from observations, with a 53–91% underestimation. Case 1 was generally well simulated, whereas the Case 2 simulation failed because the emissions of crop residue burning in northeastern China, as conﬁrmed through satellite analysis (MODIS ﬁres and thermal anomalies) and previous research, were not considered. The portion of organic/elemental carbon ratio during Case 2 was 1.6–2.3 times higher than that of Case 1. These results suggest that it is neces-sary to consider the effects of crop residue burning in northeast China to establish countermeasures to improve air quality and air quality forecasting in South Korea. These can be used as scientiﬁc evidence to plan the policies that need to be enacted in the to improve Korean air quality. the policies that are going to be enacted in the future aimed at improving Korean air quality consider crop residue burning that the air quality forecasting system may to to a lack of data regarding residue to anticipate crop residue burning in irregular here that the performance of any AQM must be improved via the of air quality


Introduction
PM 2.5 is particulate matter with an aerodynamic diameter < 2.5 µm that can harmfully affect human health by causing cardiovascular and respiratory diseases and cancer, as well as increasing early mortality [1][2][3][4][5]. The Organization for Economic Cooperation and Development (OECD) projected that the early deaths from outdoor air pollution in 2060 will triple those in 2010, and the resulting losses of gross domestic product (GDP) will be 2.6% for China and 0.6% for South Korea, representing the two most drastically impacted nations among OECD members [6]. Accordingly, to adequately address this issue, the causes of PM 2.5 must be understood, and appropriate measures must be undertaken.
Among the sources of PM 2.5 , which can be classified into natural emissions such as forest fires or yellow dust and anthropogenic emissions that originate from factories or cars, the gaseous and particulate pollutants produced by human activities are the largest contributors [7][8][9]. In particular, South Korean PM 2.5 concentrations are dependent on domestic pollutants and the influx of foreign pollutants via long-range transport [10]. Located within the westerly wind zone, pollutants produced on the windward side can be transported into South Korea [11][12][13][14]. Particulate levels also vary with meteorological conditions. If high atmospheric pressure forms without precipitation, the wind speed will be extremely weak to diffuse pollutants, thereby increasing PM 2.5 concentrations; however, in high wind speeds, significant diffusion takes place, decreasing the pollutant levels [15,16]. In particular, it is expected that the long-range transport of foreign pollutants in high wind speeds, significant diffusion takes place, decreasing the pollutant levels [15,16]. In particular, it is expected that the long-range transport of foreign pollutants creating high PM2.5 concentrations in South Korea will increase in the future [17][18][19][20]. It has also been found that pollutants produced from the northeastern region of China, located in the windward position of South Korea, can directly influence the air quality of South Korea [21,22].
The industrial and economic sectors of northeastern China are based on traditional industries such as crop cultivation, and 30% of the land area is dedicated to agriculture [23]. The burning of surplus crops has continuously occurred throughout this non-growing agricultural season in northeastern China. Furthermore, although this area reportedly possesses a larger amount of crop residues burned on site compared to other regions of China [24][25][26][27], the research surrounding the correlated impacts on PM2.5 concentrations in South Korea is insufficient. Therefore, the impact of crop burning in northeastern China on the PM2.5 concentration in South Korea was investigated in this study by analyzing meteorological data, air quality simulations, PM2.5 concentrations, and composition data during the high-concentration periods of November 2015.

Ground and Satellite Data
Among the monitoring networks operated by the Ministry of Environment of South Korea, this study utilized the data collected from 2 November to 12 November, 2015, by Intensive Air Pollution Monitoring Stations in Seoul (SMA-IAMS), Daejeon (JB-IAMS), and Jeju (JJ-IAMS), which recorded PM2.5 mass, organic carbon (OC) and elemental carbon (EC) compositions, and ion composition (SO4 2− and NO3 − ; Figure 1b). SMA-IAMS is located in an urban center with densely packed buildings, JB-IAMS is located in a residential and commercial district in the vicinity of a six-lane road, and JJ-IAMS is 600 m above sea level on the west side of Mt. Hallasan to monitor the national background PM concentrations and long-range pollutant transport [28]. To investigate pollutant source and transport trajectory, moderate-resolution imaging spectrometer (MODIS) fires and thermal anomalies data (https://worldview.earth- To investigate pollutant source and transport trajectory, moderate-resolution imaging spectrometer (MODIS) fires and thermal anomalies data (https://worldview.earth-data. nasa.gov, accessed on 6 April 2021), and the hybrid single-particle Lagrangian trajectory (HYSPLIT) of the National Oceanic and Atmospheric Administration/Air Resources Laboratory (NOAA/ARL) were utilized [29]. The MODIS data maintained a daily 1 km × 1 km Atmosphere 2021, 12, 1212 3 of 12 resolution, employing Terra (MOD14) and Aqua (MYD14) satellite data [30] to detect thermal sources, such as fires, from 4 November to 8 November, 2015. HYSPLIT analyzed the rear trajectory for the centrally located JB-IAMS site at 100, 200, and 500 m on 2 November (21:00 KST) and 10 November (03:00 KST).

Air Quality Model
To analyze the meteorological conditions and air quality changes in northeastern Asia, the Weather Research and Forecast (WRF) v.3.3 meteorological model [31], Sparse Matrix Operator Kernel Emission (SMOKE) v.3.1 emission processing model [32], and Community Modeling Air Quality (CMAQ) v.4.7.1 air quality model [33] were used. The emission inventory for the air quality simulation used the Model Inter Comparison Study for Asia (MICS-Asia) 2010 [34] for foreign regions and the Clean Air Policy Support System (CAPSS) 2010 emission list for domestic regions. For the initial and boundary conditions of the WRF, the simulated results of the unified model (UM) provided by the Korea Meteorological Administration (KMA) were used. The WRF meteorological simulation was used as the input data for the air quality model of the Meteorology Chemistry Interface Processor (MCIP) v.3.6 ( Figure 1a) [35].
The air quality modeling (AQM) area was a nested grid consisting of domain 1, including northeastern Asia, and domain 2, comprising only the Korean Peninsula ( Figure 1b). Domain 1 was 128 × 174 grids along the latitude and longitude, respectively, with a 27 km horizontal grid resolution; domain 2 was 82 × 67 grids along the latitude and longitude, respectively, with a 9 km horizontal grid resolution. The two domains were constructed equally in the vertical direction to simulate air quality up to 19 km above ground level, and the stretching grid of the lower section was used to generate 15 layers for vertical grid decomposition.
The major components of the meteorological and air quality models are listed in Table 1. The Yonsei University Scheme (YUS) [36] was used as the planetary boundary layer for the WRF model. For cloud microphysics, the WRF Single Moment 3 Class (WSM3) [37,38] was used, and the Kain-Fritsch scheme [39] was selected for the cumulus cloud option. The fifth-generation CMAQ aerosol module (AERO5) was selected for the air quality model (AQM) [40]. For the chemical mechanism, the Statewide Air Pollution Research Center v.99 (SAPRC 99) [41] was used, and for the horizontal diffusion, the YAMO scheme was used [42].

Meteorological Conditions
To assess the meteorological conditions of Case 1 and 2, the yellow dust weather list provided by the KMA was used. Figure 2 illustrates the early stages of the high PM concentrations in each case. For Case 1, a high-pressure system was formed in central China, forming favorable conditions for westerly or northwesterly airflow to South Korea along with the high-pressure air circulation (clockwise; Figure 2a). From the afternoon of 2 November, this high-pressure system slowly migrated and settled in South Korea ( Figure 2b) on 3 November (Figure 2c). Owing to this movement, the westerly to northwesterly air current was formed on the afternoon of 2 November, and by 3 November, the air was relatively stagnant due to the influence of the South Korean high-pressure system. During Case 2, South Korea was affected by a low-pressure system on 8 November (Figure 2d), and on 9 November, then it was replaced by a high-pressure system as it moved toward Japan ( Figure 2e). Thus, the westside of South Korea was under a high-pressure system, whereas the eastside experienced low-pressure conditions, leading to the formation of a north wind/air current. On 10 November, the same air pressure placement persisted, although the gradient had increased; thus, the north wind/air current remained, providing the meteorological conditions for stronger wind formation on 10 November (Figure 2f).

Meteorological Conditions
To assess the meteorological conditions of Case 1 and 2, the yellow dust weather list provided by the KMA was used. Figure 2 illustrates the early stages of the high PM concentrations in each case. For Case 1, a high-pressure system was formed in central China, forming favorable conditions for westerly or northwesterly airflow to South Korea along with the high-pressure air circulation (clockwise; Figure 2a). From the afternoon of 2 November, this high-pressure system slowly migrated and settled in South Korea (Figure 2b) on 3 November (Figure 2c). Owing to this movement, the westerly to northwesterly air current was formed on the afternoon of 2 November, and by 3 November, the air was relatively stagnant due to the influence of the South Korean high-pressure system. During Case 2, South Korea was affected by a low-pressure system on 8 November (Figure 2d), and on 9 November, then it was replaced by a high-pressure system as it moved toward Japan ( Figure 2e). Thus, the westside of South Korea was under a high-pressure system, whereas the eastside experienced low-pressure conditions, leading to the formation of a north wind/air current. On 10 November, the same air pressure placement persisted, although the gradient had increased; thus, the north wind/air current remained, providing the meteorological conditions for stronger wind formation on 10 November (Figure 2f). To study the initial air patterns during the high PM concentration events, a rear trajectory analysis was performed based on the centrally located JB-IAMS data. Figure 3a shows these back-projected results for 2 November at 21:00 KST and suggests that South To study the initial air patterns during the high PM concentration events, a rear trajectory analysis was performed based on the centrally located JB-IAMS data. Figure 3a shows these back-projected results for 2 November at 21:00 KST and suggests that South Korea was affected by the air current around Shantung Province. The northwesterly air changed to a westerly-northwesterly flow by 31 October 09:00 KST, becoming uniformly westerly as of 1 November 09:00 KST. Figure 3b shows these results for 10 November 03:00 KST and suggests that South Korea was affected by an air current from the wind-ward region. On 7 November 09:00 KST, the air current flowing via the northerly winds from Russia was affected by the easterly winds, moving to the boundary of northeastern China and North Korea. Then, by 8 November 09:00 KST, it was affected by the northerly winds, moving into South Korea.
Korea was affected by the air current around Shantung Province. The northwesterly air changed to a westerly-northwesterly flow by 31 October 09:00 KST, becoming uniformly westerly as of 1 November 09:00 KST. Figure 3b shows these results for 10 November 03:00 KST and suggests that South Korea was affected by an air current from the wind-ward region. On 7 November 09:00 KST, the air current flowing via the northerly winds from Russia was affected by the easterly winds, moving to the boundary of northeastern China and North Korea. Then, by 8 November 09:00 KST, it was affected by the northerly winds, moving into South Korea.
During the initial stages of high PM concentrations, the pressure placement and air pattern analyses suggested that South Korea was subjected to a (north) westerly airflow in Case 1 and a northerly airflow in Case 2. Their specific roles in creating the high PM concentrations observed in South Korea are described in Sections 3.2-3.4.

AQM
The PM2.5 concentrations were simulated via AQM over the research period. For the SMA-IAMS and JB-IAMS sites, the simulated concentrations of the lowest layer (z = 32 m) were used; however, for JJ-IAMS, located 600 m above sea level, the AQM concentrations at 640 m were used. Figure 4 illustrates the spatial distribution of the simulated PM2.5 concentrations with weather charts from KMA for Case 1 and Case 2. The simulated wind fields are consistent with the atmospheric pressure patterns shown in weather chart. As detailed in Section 3.1, Case 1 was simulated such that the formation of the west-to-northwesterly airflow contributed to the high PM2.5 concentrations observed. The particulate matter started to get transported to South Korea on 2 November at 15:00 KST ( Figure 4a) and lasted for approximately three days (until 5 November) due to foreign transport and stagnant air ( Figure  4b,c). The simulated high PM2.5 concentrations began to dissipate on 6 November under the influence of a northerly airflow. For Case 2, a clear north airflow was formed, preventing the high concentrations from being adequately simulated (Figure 4d-f). During the initial stages of high PM concentrations, the pressure placement and air pattern analyses suggested that South Korea was subjected to a (north) westerly airflow in Case 1 and a northerly airflow in Case 2. Their specific roles in creating the high PM concentrations observed in South Korea are described in Sections 3.2-3.4.

AQM
The PM 2.5 concentrations were simulated via AQM over the research period. For the SMA-IAMS and JB-IAMS sites, the simulated concentrations of the lowest layer (z = 32 m) were used; however, for JJ-IAMS, located 600 m above sea level, the AQM concentrations at 640 m were used. Figure 4 illustrates the spatial distribution of the simulated PM 2.5 concentrations with weather charts from KMA for Case 1 and Case 2. The simulated wind fields are consistent with the atmospheric pressure patterns shown in weather chart. As detailed in Section 3.1, Case 1 was simulated such that the formation of the west-to-northwesterly airflow contributed to the high PM 2.5 concentrations observed. The particulate matter started to get transported to South Korea on 2 November at 15:00 KST (Figure 4a) and lasted for approximately three days (until 5 November) due to foreign transport and stagnant air (Figure 4b,c). The simulated high PM 2.5 concentrations began to dissipate on 6 November under the influence of a northerly airflow. For Case 2, a clear north airflow was formed, preventing the high concentrations from being adequately simulated (Figure 4d-f).
The overestimation of Case 1 primarily occurred during the early stages of foreign transport; thus, it can be verified that the foreign influx of PM was higher than that observed, likely as a result of the emission list being based on 2010 data and failing to reflect the true situation of 2015.    The overestimation of Case 1 primarily occurred during the early stages of foreign transport; thus, it can be verified that the foreign influx of PM was higher than that observed, likely as a result of the emission list being based on 2010 data and failing to reflect the true situation of 2015.
The underestimation of high concentrations in Case 2 was likely based on nearby crop residue burning (see Sections 3.3 and 3.4). The foreign emission data, MICS-Asia 2010, used for the air quality simulation in this study classified emission sources as either development, engineering, traffic, residence, or agriculture. For agriculture, only NH 3 is included, and crop residue burning and dust scattering are excluded from the emission calculations [34]; therefore, it was found that the high PM 2.5 concentrations could not be adequately modeled as foreign emissions from crop burning.

Crop Residue Burning in Northeastern China
To determine the driving factors behind the failure to model the high-pressure system observed in Case 2, the atmospheric situation of northeastern China, windward of South Korea, was further examined using MODIS fire and thermal anomaly data. The MODIS products from 4 November to 8 November indicated that relatively extensive heat was produced in northeastern China prior to Case 2 (although cloud cover prevented further detection on 7-8 November; Figure 6). According to the PM 2.5 network (http://www. pm2.5.in, accessed on 8 November 2015) information on China's air quality monitoring network, Changchun City in northeastern China, located between Shenyang and Harbin, exhibited a daily PM 2.5 average concentration of 536 µg·m −3 on 8 November, a 21-fold increase over the recommended WHO standards of 25 µg·m −3 [46]. On 9 November 00:00, the recorded PM 2.5 concentration was 991 µg·m −3 , signaling a severe air pollution event; therefore, it was concluded that the heat production and resulting pollution continued to 7-8 November.
Atmosphere 2021, 12, x FOR PEER REVIEW 8 of 13 Previous research has found that the heat production during Case 2 was due to crop burning. Yin et al. used MODIS data to identify crop residue burning sites in northeastern China [47], concluding that from 4 November to 6 November, daily crop residue burning occurred at 2291, 1050, and 1109 individual sites, respectively. In particular, during the agricultural off-season of northeastern China in November 2015, the highest hourly PM2.5 concentrations reported surpassed 1000 µ g·m −3 . Furthermore, Zhang et al. reported serious air pollution in Shenyang in 8 November 2015, with PM concentrations of 1326 µ g·m −3 [48]. Previous research has found that the heat production during Case 2 was due to crop burning. Yin et al. used MODIS data to identify crop residue burning sites in northeastern China [47], concluding that from 4 November to 6 November, daily crop residue burning occurred at 2291, 1050, and 1109 individual sites, respectively. In particular, during the agricultural off-season of northeastern China in November 2015, the highest

PM 2.5 Composition
The main components of PM 2.5 in Cases 1 and 2 are shown in Table 3 and Figure 7. The majority of the composition data are missing from SMA-IAMS owing to the maintenance schedule, and this site was thus removed from this analysis. The observed PM 2.5 concentrations at JB-IAMS in Case 1 ranged from 36-104 µg·m −3 , with an average of 61.3 µg·m −3 ( Table 3). The relative average contribution was highest for NO 3 − (38.  (Figure 7b). Notably, the relative contribution of NO 3 − diminished in Case 2, whereas that of OC increased.   Table 3). The average component contribution of PM 2.5 was highest for SO 4 2− (41.4%), followed by OC (23.7%) and NH 4 + (18.7%; Figure 7c). In Case 2, the PM 2.5 concentrations ranged from 41-116 µg·m −3 (average 63.4 µg·m −3 ; Table 3). The relative average contribution was highest for OC (63.6%), followed by SO 4 2− (16.7%) and NH 4 + (8.2%; Figure 7d). Accordingly, the relative contribution of SO 4 2− decreased, whereas that of OC significantly increased in Case 2.
For JB-IAMS, NO 3 − maintained the largest contribution in both cases. To analyze the cause, CAPSS, the emission list data provided by the Ministry of Environment of South Korea, was used to investigate the major domestic emission sources of Junggu, Daejeon, where the site is located [49]. In 2015, the largest PM 2.5 emission source was scattered dust and automobile pollution, accounting for 32.8% and 30.9%, respectively (~64% of total emissions). Among automobile pollution sources, NOx accounted for 40% of the pollutants; therefore, it can be inferred that NO 3 − would have maintained the highest contribution compared to the JJ-IAMS area due to its local characteristics.

Conclusions
In this study, two cases (Cases 1 and 2) where the PM 2.5 concentrations exceeded 35 µg·m −3 during 2-12 November 2015 were analyzed. For Case 2, meteorological conditions, air quality simulations, PM 2.5 concentrations, and corresponding composition data were used to study the effects of crop residue burning in northeastern China on the PM 2.5 concentrations of South Korea.
As a result of the AQM, the simulated PM 2.5 concentrations in Case 1 overestimated the observation values by 3.7-17.6 µg·m −3 (6-36%); however, in Case 2, the simulated values were underestimated by 20.0-59.8 µg·m −3 (53-91%). Despite this overestimation, Case 1 properly simulated the high-concentration phenomena that occurred in South Korea over this period; however, in Case 2, the high concentrations could not be simulated due to a spike in crop residue burning that occurred over this time. Based on the satellite data and preceding research, it was confirmed that there were crop residue burning events in the northeastern regions of China throughout this period, and when compared with Case 1, Case 2 had significantly higher contributions of OC, with an average OC/EC ratio of 8.8 (a 1.6-2.3-fold increase; extremely similar to the expected crop residue burning yields of 9.0). It was confirmed that crop residue burning from the northeastern region of China during the non-growing agricultural off-season significantly affected the air quality of South Korea. However, this study has a limitation that AQM modeling could not be directly performed using the emissions of crop residue burning in northeast China in Case 2, because it was not possible to obtain the available emission information (location, field size, fuel load, etc.) for estimating them.
These results can be used as scientific evidence to plan the policies that need to be enacted in the future to improve South Korean air quality. Particularly, the policies that are going to be enacted in the future aimed at improving South Korean air quality should consider crop residue burning in northeast China. Furthermore, this research has highlighted that the air quality forecasting system may fail to predict high concentrations due to a lack of input data regarding foreign crop residue burning; however, as it is difficult to anticipate crop residue burning in advance due to its irregular periodicity, it is suggested here that the performance of any AQM must be improved via the inclusion of air quality data assimilation of satellite and ground data