Impact of Wheat Harvest on Levels and Sources of PM2.5-associated PAHs in an Urban Area Located at the Center of Beijing-Tianjin-Hebei Region

Wheat harvest and subsequent straw burning for maize planting can cause severe PM2.5 and PAH pollutions and deteriorate the air quality of nearby cities in consequence. PM2.5 samples were collected in Baoding urban area (BUA) from June 18 to July 7 of 2019, during and after wheat harvest (DWH and AWH, respectively). The “Migration Effect” (i.e., PM2.5 and PAHs transferred from rural to urban during DWH and AWH, respectively) was proved by both the later time for appearance of peak values of PM2.5 and PAHs and the air mass origins in BUA. The daily average PM2.5 (reported in μg m) 137 of DWH was 2.58 times 53.1 of AWH for BUA, regardless of its lower levels than the corresponding 156 and 75.6 for an adjacent rural site (ARS). The reverse trend was found for PAH mass contents (in μg g), AWH possessed much higher value of 139 than 27.8 of DWH, while no significant fluctuations occurred for PAH concentrations due to the varied PM2.5 levels. Four PAH sources including biomass burning (BB), coal combustion (CC), vehicle exhaust (VE) and industrial emissions (IN) were identified using positive matrix factorization (PMF). BB was the biggest contributor during whole sampling period (WSP), followed by CC, VE, and IN. BB increased from 17.3% to 37.1% along with the transition from DWH to AWH, indicated the impact of straw burning for maize planting. High share of CC suggested that coal was still an important civil fuel. The strict emission reduction measures made the industry smallest contributor in BUA. Lower VE share in BUA than ARS demonstrated that influence of high-intensity operation of wheat harvesters and rotary cultivators.


INTRODUCTION
Environmental contamination has also occurred along with the process of industrialization and urbanization in the past few decades in China. In recent years, more attention has been paid to the issues with frequent occurrence of haze and photochemical smog characterized by high loadings of particulate matter and ozone (Elorduy et al., 2016;Cao et al., 2018;Li et al., 2020a, b). Fine particulate matter (especially PM2.5) has received widespread attention due to its negative impact on air quality and human health (Li et al., 2019a. As a carrier for various toxic impact of wheat harvest activities on urban PAHs. The main purposes of this study were to: 1) analyze the changes of PM2.5 and PAHs in BUA associated with wheat harvest and subsequent straw burning activities; 2) to analyze the origins of air masses using backward trajectory clustering; and 3) identify the sources of PAHs using Positive Matrix Factor (PMF) model.

Sampling Area Description
As shown in Fig. 1, BUA locates in the northern part of the North China Plain (NCP) and central area of BTH region, which is 140 kilometers north of Beijing, 145 kilometers east of Tianjin and 125 kilometers southwest of Shijiazhuang. The sampling site is on the rooftop of a six-story Teaching Building (approximately 30 m above ground) in North China Electric Power University without tall buildings and industrial enterprises around. The other sampling site in our previous work, ARS was also shown in Fig. 1 2.2 PM2.5 Sampling PM2.5 samples were consecutively collected by the medium-volume air samplers (TH-150C III, Wuhan Tianhong Ltd., China) at a flow of 100 L min -1 and a total of 40 samples were obtained. The 23 h integrated samples were gathered during June 18 to July 7, 2019 and each was conducted from 9:00 a.m. to 8:00 a.m. of the next day. Two air samplers were used to collect daily samples. One sampler was equipped with a quartz filter (QF, 1 µm pore size and 90 mm diameter, Pall USA Inc.) was used to analyze PAHs, and the another sampler installed with a Teflon filter (TEF, 1 µm pore size and 90 mm diameter, Whatman Co. UK) was used to obtain the PM2.5 mass by subtracting pre-weight from post-weight. A set of field blank samples were simultaneously gathered by exposing filters in the sampler without airflow in drawing, to evaluate the subsequent analysis bias and precision. Prior to sampling, QFs were heated at 450°C for 4 h and TEFs were also heated at 60°C. TEFs were stored in a room with a constant 20°C and 50% of relative humidity before and after sampling for weighting (SartoriusME-5F, readability: 1 µg).
Meteorological parameters including the temperature, wind speed and direction were simultaneously recorded. The sensitivity of the weighing scale was set to ± 0.010 mg. The PM2.5 concentrations were finally calculated as the quotient between PM2.5 mass and corrected sampling air volume.
The entire pre-treatment and analysis procedures comply with the quality control/quality assurance (QC/QA) programs were strictly done. A series of experiments including sample blank, sample duplication, matrix spiked sample, and procedural blank were conducted on schedule every 5 samples to evaluate the accuracy of the analysis. No target chemicals were found in the solvent or procedural blank experiments. The method detection limits (MDLs) (reported in ng g -1 ) of the 18 PAHs were in the 0.12-1, with a mean value of 0.230 ± 0.202. The recovery efficiency for the 18 PAHs in matrix added samples ranged from 80% to 121%. The surrogate standards including 14-deuterium substituted terphenyl and 4-bromo-2-fluorobiphenyl in 20 samples, possessed the recoveries of 90 ± 15% and 95 ± 20%, respectively. The variations of PAH detected concentrations in 4 duplicated samples were all less than 10%.
A total of 11 PAHs including BgP, IP, BbP, PHE, FA, BaA, CHR, PY, BaP, BeP and BkF were detectable in this study. Therefore, the sum of 11 PAHs (Σ11PAHs) was used to represent the total PAHs.

Source Identification by PMF Model
The positive matrix factorization (PMF), a multivariate factor analysis tool, has advantages over PCA and CMB lies in elimination of negative factor loadings and no need of source profiles, is widely used for PM2.5 source appointment (Bari and Kindzierski, 2016;Chen et al., 2016;Yao et al., 2016;Tan et al., 2017). It can acquire more realistic source information based on the nonnegative constraints. The EPA PMF (version5.0) was employed to identify and apportion the potential source contributions of PM2.5-bound PAHs in this study. The model input datasets were consisted of PAH concentrations and corresponding uncertainties. The abnormal values were excluded to prevent error of PMF results.
The half of MDL was used to represent the PAH concentration below than MDL. The geometric mean of the observed values was used as substitute for missing values and associated uncertainties were set as four times the geometric mean (Yao et al., 2016). Uncertainty, a key input, included detection limits and error fraction. Uncertainty data could be obtained by following equations: For ci ≤ MDL, unc = 5/6 × MDL (1) A total of 20 runs were used for each chemical component. The lowest Qrobust value was 3069.30, and the ratio of Qrobust/Qtrue was 0.90.

Backward Trajectory Clustering Analysis
Backward trajectory clustering analysis can identify the potential geographic origins of PAHs and is a key tool to simulate the trajectories of air pollutants. Three dimensional 72 h backward trajectories of air masses arriving at Baoding were calculated using the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT-4) model. According to the properties and methods of the clustering trajectory map, the trajectories were divided into four paths and the uniformity of the trajectory was reflected at the minimum angle.
The PM2.5 concentrations showed a generally downward trend from June 18 to June 20 reflected the wheat harvest activities for surround rural areas of BUA entered the final stage. BUA showed the similar PM2.5 fluctuation trends with ARS in Wangdu County during DWH in spite of the its relative low PM2.5 levels, which implied the migration effects of regional particles (Li et al., 2020b).
Triggered by the crisis of high PM2.5, China launched the "Air Pollution Prevention and Control Action Plan" (APPCAP) during 2013-2017, the national emission of primary PM2.5 reduced by 33% in consequence and PM2.5 levels in Beijing-Tianjin-Hebei (BTH) region basically reached the National Air Quality Standard Grade II (Cheng et al., 2019;Zhai et al., 2019). However, PM2.5 associated with wheat harvest in a central rural area of BTH and BUA incompatibly reached up to 156 µg m -3 and 137 µg m -3 , indicating it was necessary to develop effective measures to reduce PM2.5 emissions from wheat harvest and subsequent open straw burning (Li et al., 2020b). It should be noted that the elevated PM2.5 levels during DWH were still lower than those before the implementation of APPCAP confirmed its effectiveness for PM2.5 decline (Li et al., 2020b). However, DWH possessed substantially higher PM2.5 than those after APPCAP.
Much higher PAH mass contents were found during AWH compared with those during DWH implying the strong impact of PAH emissions from open wheat straw burning (Fig. 2). The PAH mass contents during DWH varied from 18.0 to 43.0 µg g -1 and averaged as 27.8 µg g -1 , while they fluctuated from 34.9 to 825 µg g -1 with a mean up to 179 µg g -1 during AWH. It should be noted that total PAH contents peaked on June 22 and July 2 in ARS and the highest PAHs of BUA accordingly appeared on subsequent June 23 and July 3 in consequence, which suggested the wheat straw burning associated PAHs migrated from ARS to BUA (Li et al., 2020b). More than 40% of air masses originated from southern Baoding and trajectories arrived at BUA on June 23 and July 3 even reached up to 81.2%, which further verified the occurrence of the "Migration Effect" for PAHs. It should be concluded that straw burning after harvest characterized as low cost and easily operate still remained as an important practice in some rural areas regardless of the open straw burning prohibition in China. The mass volume concentrations of PAHs exhibited the different trends with PAH contents ascribed to the varied PM2.5 concentrations. The PAH volume concentrations (reported in ng m -3 ) for DWH were in the range of 2.80-4.64 and averaged as 3.57, and they varied from 1.11 to 8.55 with a mean value as 4.51 for AWH. The lower levels of PM2.5 of AWH compared with DWH weaken the difference of PAH volume concentrations between them.
The 5 top PAHs in ARS of Wangdu County were BeP, BbF, IP, BgP, and COR based on both mass contents and concentrations. The dominant PAH congeners between two places were roughly the same, indicating that the PAH pollution sources in two regions were roughly similar (Li et al., 2020b). Certain concentration of BaP was found for both DWH and AWH, which should be paid more attention due to its severe toxicity . Fig. 4 showed the ring size distribution patterns of PAHs during DWH and AWH for each sampling day. The 5-ring PAHs dominated in all the sampling days, followed by the 6-and 4-ring PAHs, while the 3-ring PAHs possessed the lower contributions. During whole sampling period, HMW-PAHs were predominant congeners. Along with the transition from DWH to AWH, the contribution ratios of 5-and 6-ring PAHs decreased from 49.3% to 48.9% and from 25.4% to 24.1%, while 4-and 3-ring PAHs increased from 23.1% to 24.6% and from 2.21% to 2.34%, respectively, which reflected the differences of source contributions between DWH and AWH. For ARS in Wangdu County, 5-ring PAHs increased from 42.3% to 46.7%, while the 6-ring PAHs decreased from 30.9% to 26.6%. The similar contribution ratios between ARS and BUA indicated that the surrounding straw burning phenomenon had a certain impact on the urban area. The 2-, 3-, and 4-ring PAHs dominated in particles emitted from the straw burning would be the explanation of enhanced levels of 4-ring PAHs .

Source Apportionment of PAHs by PMF Model
PMF model was employed to apportion the potential pollution sources and four source profiles were obtained (Fig. S1). Factor 1 was characterized by high loadings of BbF, IP, FA, BeP and CHR, and medium to high loadings of PY and BaA. Also, high proportions of 4-5 ring PAHs were found in Factor 1 and it was assigned to represent vehicle exhaust (VE) (Zhou et al., 2005). The high number of private cars and the peak summer travel during summer vacation verified the VE exist. Factor 2 possessed high loadings of HMW-PAHs like IP and BgP and low levels were found for LMW-and MMW-PAHs including PHE, PY, FA and CHR, which was identified as coal combustion (CC) (Zhou et al., 2005;Li et al., 2020b). In consideration of Factor 3, the highest loading of BeP and BaP, and the low contents of LMW-PAHs such as FA, CHR and PY were found. Also the highest levels of 5-ring PAHs and certain amounts of PAHs with other ring numbers were found and Factor 3  6 -1 8 6 -1 9 6 -2 0 6 -2 1 6 -2 2 6 -2 3 6 -2 5 6 -2 6 6 -2 7 6 -2 8 6 -2 9 6 -3 0 7 -1 7 -2 7 -3 7 -4 7 -5 7 -6 7 -7 AWH DWH was consequently recognized as biomass burning (BB) (Li et al., 2020b). In regard to Factor 4, there was a high loading of 3-to 5-ring PAHs such as FA, CHR, and BbF, while low contents occurred for PAHs with higher ring number than 5 including IP and BgP. It was designed as industrial emissions (IN) accordingly There are a lot of factories in the east of BUA and adjacent rural areas, which may be the PAH emission source. Fig. 5 showed the contributions of four sources including CC, VE, BB, and IN for DWH, AWH and WSP. For USA during WSP, BB possessed the highest contribution ratio of 34.4%, followed by 25.3% of CC, 22.5% of VE, and 17.8% of IN. The substantially different source contribution pattern was found in ARS, which was VE (30.8%) > BB (27.6%) > CC (27.5%) > IN (14.1%) (Li et al., 2020).
The lowest IN contribution ratio was found for both BUA and ARS, which implied the effectiveness of recent emission control measures. The higher VE proportion surprisingly occurred in ARS instead of BUA suggested the strong impact of running of harvesters and rotary tillers with high intensity in ARS during WSP. The implausible appearance of highest BB contribution ratio in BUA was mainly ascribed to the decreasing shares of VE and CC. CC possessed slightly lower share in BUA than that of ARS, which indicated coal was still the important fuel regardless of coal prohibition in ARS and had a certain impact on air quality of BUA. In consequence, coal combustion was still a main contributor of atmospheric PAHs although the civil coal combustion was completely forbidden in BUA.
The shares of four types of emission sources fluctuated significantly during WSP due to the varied emission intensities and weather conditions. The VE and IN peaked on July 4 and June 29, which were 73.4% and 56.7%, respectively. BB possessed the largest number of peaks and they were 52.5% of June 21, 65.8% of June 23, 61.4% of June 27, and 80.7% of July 6. The time with occurrence of the BB peaks was almost same between ARS and BUA reflected the straw burning related PAHs had a certain impact on BUA.

Backward Trajectory Clustering
Fig . S2 showed the backward trajectory curves during different periods including DWH, AWH, and WSP. There were four wind pathways for the three periods. In regard to DWH, 11.1% of air trajectories came from northern Mongolia, 30.0% from Cangzhou City and Bohai Bay, 33.3% from Fuping, Laiyuan County and east Shanxi province, 25.6% from southwest Shandong province, northern Henan Province, Handan, Xingtai and Shijiazhuang. It should be noted that the 33.3% of air masses originated from mountainous areas within Fuping County, Laiyuan County, Quyang County and Shanxi Province could bring large amounts of PAHs and PM2.5 associated with indoor cooking fueled with biomass into BUA during DWH (Li et al., 2020). The inputs of fugitive dust derived from the wheat straw stirring by harvesters in adjacent rural areas and desert of Mongolia resulted in high PM2.5 levels of BUA during DWH.
The significant difference existed between BUA and ARS based on their air mass trajectories during AWH. 7.53% of the tracks came from northern Baoding, Beijing, Zhangjiakou and southern Mongolia, 43.55% from Bohai Bay, 40.86% from Shijiazhuang and Xingtai, and southern Baoding,

CONCLUSIONS
Baoding, locates in the central part of BTH, is frequently nagged by the severe air pollution events in recent years. BTH is an important wheat producing base in China and its air quality can be deteriorated by PM2.5 and PAHs associating with the activities of wheat harvest and subsequent straw burning for maize planting. Our previous study investigated the impact of wheat harvest associated behaviors on PM2.5 and PAHs for ARS within Baoding city, which indicated the wheat harvest associated fugitive dust dramatically elevated the PM2.5 levels in DWH and PAHs levels were also markedly promoted by straw burning in AWH.
In this study, PM2.5 samples were systematically collected in BUA during DWH and AWH to investigate the levels of PM2.5 and 18 PAHs, and PAH sources. 1) "Migration Effect" occurred during WSP, which caused the similar PM2.5 fluctuation trends between ARS of Wangdu County and BUA. Compared with the ARS, the emergence of the peaks of PM2.5 concentrations and PAH contents in BUA has a certain lag, and also the corresponding values in BUA were lower than those of ARS. Combination of the wind speed and direction during WSP with occurrence of pollutant peaks in BUA suggested that input effects of PM2.5 in DWH and PAHs in AWH from rural areas. 2) Higher PM2.5 of 137 µg m -3 was found in DWH, 2.58 times 53.1 µg m -3 of AWH, which was similar to ARS mainly due to the large amounts of fugitive dust during DWH. Unlike PM2.5, the reverse fluctuation trend was found for PAHs and higher PAHs appeared during AWH due to the straw burning for maize planting. AWH possessed substantially higher PAH content (reported in µg g -1 ) as 179 than 27.8 of DWH. 3) BB was the biggest contributor, followed by CC, VE and IN. BB share was significantly elevated from 17.3% of DWH to 37.1% of AWH, indicating the influence of straw burning regardless of the prohibition of straw burning issued by government. It was worth mentioning that high VE portion occurred in ARS instead of BUA due to the intensive running of wheat harvesters and Rotary cultivators. The contributions of IN and CC in BUA were slightly higher and lower than those of rural site. In a word, the wheat harvest associating activities promoted the assimilation of air quality in urban area into that of the ARS. 4) Backward trajectories clustering analysis showed that more than 40% air masses arrived at BUA passed through ARS of Wangdu County. The enterprises in surrounding rural areas and Shijiazhuang City would be the industrial source of PAHs in BUA.