Effects of Rainfall on PM2.5 and PM10 in theMiddle Reaches of the Yangtze River

Based on the PM2.5 and PM10 mass concentration data obtained from 51 national air quality monitoring stations and the corresponding rainfall intensity data in automatic meteorological stations in Hubei Province from 1 January 2015 to 31 December 2017, the impact of rainfall intensity on PMmass concentrations under relatively different humidity conditions was analyzed. -e results showed that light rain occurredmost frequently in the pollution process, with Xiangyang being affected for up to 587 h. PM concentration would not change drastically under the effect of precipitation. Mean rainfall intensity responsible for wet growth of PM10 and PM2.5 wasmainly <0.5mm/h, while that responsible for wet removal of PM2.5 was significantly higher (>1.4mm/h) than that of PM10 (>1.0mm/h). Precipitation was more likely to produce a wet removal effect for a greater initial value of PM mass concentration, and on the contrary, a wet growth effect was more likely, with the threshold of PM10 mass concentration being 150 μg/m and that of PM2.5 mass concentration being 95 μg/m.Wet removal played a leading role in lower humidity (∼60%) and greater rainfall intensity, but wet growth played a leading role in higher humidity (∼90%) and lower rainfall intensity. As the precipitation level increased (rainfall ≥1.5mm·h), the wet removal to PM10 mass concentration was enhanced more obviously. -e variations of PM2.5 had similar distributions to those of PM10 under the effect of precipitation, but the wet removal effect of precipitation was weakened and the wet growth effect was enhanced.


Introduction
Atmospheric aerosols, the suspended systems of liquid or solid particles in the air, constitute an important part of the atmosphere [1][2][3] and are a key factor affecting environmental and climate change. Atmospheric pollution is a weather phenomenon caused by accumulation of atmospheric aerosols through local emissions and regional transmission [4][5][6][7]. e emission source intensity of atmospheric pollutants determines the atmospheric pollution degree, while meteorological conditions determine the outbreak, persistence, and elimination of pollution weather.
As a key meteorological element affecting pollutant concentration, precipitation dominates the process of wet removal in the atmosphere as one of the most important mechanisms in self-purification of the atmosphere [8]. Wet removal of aerosol particles is the process in which particles eventually fall to the ground after being removed by atmospheric coagulants, mainly caused by Brownian diffusion and inertial collision [9]. However, the humidification effect of precipitation on the near ground significantly enhances the influence of hygroscopic growth on aerosols and accelerates formation of secondary aerosols, resulting in increased mass concentration [10]. e impact of precipitation on aerosols involves a quite complicated mechanism, which closely concerns not only the characteristics of precipitation intensity, rainfall, raindrop spectrum, and final velocity of raindrops [11,12], but also the aerosol particle size, spectral distribution, and chemical composition [13,14].
Research on wet removal of precipitation has mainly considered efficiency of precipitation in removing aerosol particles of different particle sizes in observational experiments. Related studies have pointed out that precipitation mainly removes aerosol particles of size <1 μm and 2.5-10 μm [15,16], while those of 0.2-2 μm are difficult to remove [17][18][19]. Precipitation intensity directly affects aerosol removal efficiency. Short-term heavy precipitation helps remove particles of diameter <2.2 μm, but long-term weak precipitation facilitates removal of particles >2.2 μm [20]. In general, increased precipitation intensity will enhance its wet removal ability. Yao et al. [21] and Zhao and Zheng [22] further established a model on correlation between precipitation intensity and aerosol removal coefficient. At the same time, wet removal effect of precipitation also concerns initial concentration of aerosol before precipitation. For a higher initial concentration, the wet removal effect is more significant in the same context. Xu et al. [23] pointed out that, under 70 μg/m 3 in winter and 45 μg/ m 3 in other seasons, the PM 2.5 mass concentration decreased in more than 80% of precipitation processes. Related research also shows that PM 2.5 mass concentration does not decrease but increases after precipitation in actual observations [24]. Yue et al. [25] demonstrated that processes with daily precipitation <10 mm produce a humidification effect significantly stronger than its wet removal effect on PM 2.5 .
e wet removal effect of precipitation is closely related to ambient humidity. e scouring coefficient of hygroscopic particles constitutes a function of relative humidity. To achieve the same removal effect, the precipitation intensity required for 50% ambient humidity is twice the intensity required for 95% ambient humidity [26,27]. e rapid formation of secondary aerosols in high humidity and low wind-speed environments is the main reason for the increase of PM 2.5 mass concentration following precipitation [28,29].
e Twain-Hu Basin (THB) with a special geographical location connecting the major haze pollution areas, represented by Hubei Province, is a hub of regional transport of atmospheric pollutants in China. As the development center of the Yangtze River Economic Belt and a city cluster in the middle reaches of the Yangtze River, the THB area has special "subbasin" topography, which promotes accumulation of atmospheric pollutants. In addition, there are quite developed water systems in the THB [30,31]. In recent years, the level of atmospheric aerosols in this area has shown an upward trend [32]. e high humidity weather and high concentration pollution have significantly worsened regional visibility [33,34], leading to gradual formation of a new core region of pollution. e difference in rainfall intensity, variable humidity, and different pollution characteristics result in many uncertainties in the impact of precipitation on PM (PM 2.5 and PM 10 ), which is the main reason for errors in judgment of pollution levels with the occurrence of precipitation. Due to the lack of analysis on the correlation between large-area, long-term sequence precipitation and PM concentration, there is currently no quantitative understanding of the general laws and characteristics of precipitation effects on PM 2.5 and PM 10 mass concentrations. In the middle reaches of the Yangtze River, some weather conditions during heavy pollution, such as low pressure with an inverted trough and a cold front, are often accompanied by precipitation [35].
Precipitation can occur during 70-80% of pollution events under the effects of cold fronts [36]. erefore, it is necessary to analyze the variations of long-sequence precipitation and PM in combination with observation data, explore the distributions of rainfall intensity under different pollution conditions, summarize the variation of PM concentrations at the beginning and ending of precipitation hours under different humidity conditions, and determine the different effects of wet removal and wet growth on PM in the presence of precipitation. is will effectively improve the accuracy of aerosol concentration prediction during precipitation and enhance understanding of the variation mechanisms in pollution processes under high-impact weather. e time resolution is 1 h. e mass concentrations of PM 2.5 and PM 10 were measured using β absorption and microoscillating balance methods. e observation equipment was regularly cleaned and calibrated according to the operational specifications, and observation data were subject to quality control to ensure accuracy and validity. Observational data were taken into account when effective data of PM 2.5 and PM 10 exceeded 20 h in one day. At the same time, according to the standard of the Ministry of Environmental Protection (GB 3095-2012) [37], pollution weather was divided into light (LP), moderate (MP), severe (SP), and terrible pollution (TP), as shown in Table 1.

Data and Methods
e collected data included precipitation, relative humidity, and wind speed. In order to match the precipitation and pollutant data more accurately, 2539 automatic meteorological stations in Hubei Province were used, which were maintained by Hubei Meteorological Information and Technology Support Center.
For the summer half-year, there is more local precipitation; for the winter half-year, regional precipitation is the main phenomenon. In order to ensure that when the meteorological automatic station observed precipitation, there was also occurrence of precipitation in the adjacent national air quality monitoring station, we adopted the following main principles: (1) priority selection of the automatic weather station closest to the national air quality monitoring station; (2) using the observation data of the next closest automatic weather station when the observations of the closest automatic weather station were missing. e timespace consistency could be guaranteed in meteorological and pollutant concentration data.
Because we mainly used hourly precipitation, rainfall intensity was classified into light rain (LR), moderate rain (MR), heavy rain (HR), torrential rain (TR), and downpour (SR) according to the local standard of neighboring province (DB 34/T 1592-2012) [38], as shown in Table 2.
We selected time periods with PM 10 concentration >150 μg/m 3 or PM 2.5 concentration >75 μg/m 3 . Some parameters were determined to explore the impact of precipitation on the change in PM concentration. V PM2.5 , V PM10 , and C PM2.5/PM10 were determined according to (1)-(3), respectively: where M end PM2.5 and M end PM10 are the PM 2.5 and PM 10 mass concentrations at the end moment of the precipitation hour, respectively. M beg PM2. 5 and M beg PM10 are the PM 2.5 and PM 10 mass concentrations at the beginning moment of the precipitation hour, respectively. Some researchers term these the initial concentrations [16,18]. Precipitation data when wind speed exceeded 3 m/s were eliminated to ensure a low wind-speed environment condition [23]. Figure 2 shows the distribution of different-level pollution days in 13 regions of Hubei Province during 2015-2017. Judging from the total days with pollution weather, we found that the pollution weather was most frequent in XY and YC, reaching 335.3 and 304.1 d, respectively. Pollution weather occurs in one-third of days throughout the year in these two cities. ese two cities are also typical representative areas affected by the transmission of an external pollution source and accumulation of a local pollution source. e heavy pollution weather in XY is often accompanied by a strong northerly wind and a weak inversion layer, whereas low wind speed and a multilayer inversion often occur in YC during heavy pollution weather [39,40]. Pollution weather for ES, HG, SY, and XN occurred in less than 160 d, mainly located in the mountainous areas of Hubei Province, where pollution days were significantly lower than other areas.

Distribution of Pollution Weather at Different Levels.
LP days accounted for >60% of total pollution days, especially in areas with less pollution such as XN and HG where LP days represented >80% of the total pollution days, reaching >100 d. MP days mainly accounted for 18-23%, with differences of less than 30 d among regions. SP days were significantly fewer, representing about half of MP days. However, it is worth noting that XY and YC showed no obvious reduction in SP days. In particular, SP days in XY

Characteristics of Precipitation during Pollution Weather
Days. e distributions of precipitation hours during pollution days in Hubei Province from 2015 to 2017 ( Figure 3) revealed that the mean value of LR during pollution weather in each area was 298 h. LR was the most frequent in XY, reaching 587 h, with a mean value of 497 h. Occurrence of MR was lower than LR by an order of magnitude, with a mean value of 21 h. XY, YC, and WH subject to heavy pollution processes were less affected by MR which lasted for only about 10 h. e occurrence of HR was consistent in each region, mainly in the range of 5-12 h. National air quality monitoring stations of WH and EZ located in Wuhan City Circle showed significant differences in occurrence of HR, with values exceeding 10 h. Regions where TR occurred during pollution weather were mainly located in Wuhan City Circle, all within 7 h. e SR occurred rarely, with only 1 h in the four cities. Precipitation hours of the same level differed for the different national air quality monitoring stations in the same city. As heavy precipitation was usually concentrated in local areas, average deviation of occurrence in LR, MR, and HR gradually increased, with 14.75%, 26.42%, and 35.01%, respectively. e greater the rainfall intensity was, the larger the difference in occurrence of precipitation was at national air quality monitoring stations in the same area, showing that the research on the separate occurrences of precipitation in different national air quality monitoring stations was reasonable.
Strong-thick inversion layers and low altitude mixed layers that often occur at night or in the early morning are typical conditions that aggravate pollution weather [41][42][43].
eir strong inhibition of pollutant diffusion means that the peak concentration of pollutants often occurs at night. Bai et al. [44] found that daily variation of PM 2.5 concentrations in Hubei Province showed a bimodal distribution in winter, with peak values during 11:00-14:00 caused by transportation and peak values during 21:00-24:00 caused by accumulation.
e frequency of precipitation during different pollution periods can indirectly reflect the impact degree of precipitation on pollution weather. Figure 4 shows diurnal variation of the occurrence of different levels of precipitation during pollution weather in Hubei Province from 2015 to 2017. e daily variations of precipitation showed a unimodal distribution regardless of precipitation level during pollution weather, with peak values during 20: 00-23:00. Over time, the ratio of LR to MR and HR at each time was reduced from 17 and 110 during 00:00-03:00 to 6 and 30 during 19:00-23:00, respectively; HR was concentrated during 17:00-23:00. ese moments were during the period when PM 2.5 mass concentration reached its peak value. With increased rainfall intensity, precipitation was more concentrated at night, indicating that occurrence of precipitation during the pollution process may have had a great impact on the change of PM mass concentration.

Variation Characteristics of PM Mass Concentration before and after Precipitation during Pollution Process.
e value of PM mass concentration is, on the one hand, affected by wet removal caused by inertia coagulation of the precipitation particles during their falling and, on the other hand, affected by aerosol hygroscopic growth and secondary aerosol formation caused by the humidification effect of precipitation. At the same time, the variation law in background concentration of PM under the joint action of localsource emission, external-source transportation, and air mass diffusion also affects the value of PM concentration. erefore, this section statistically analyzes PM 2.5 , PM 10 , and their ratios before and after precipitation to reveal distribution laws governing PM concentration under the influence of precipitation. Table 3 shows the distributions of V PM2.5 and V PM10 under different C PM2.5/PM10 intervals during rainfall hours. With the ending of precipitation, PM 10 and PM 2.5 concentrations decreased by 61.5% and 52.3% in hours, respectively. Absolute values of the decrease ratios of PM 10 and PM 2.5 were larger than those of the increase ratios of PM 10 and PM 2.5 . e effect of precipitation on wet removal was significant. Moreover, PM 10 was more affected by wet removal of precipitation, which was relatively consistent with the results of Xu et al. on wet removal of precipitation in   Advances in Meteorology Shanghai [23]. ey showed that precipitation in Shanghai mainly removed atmospheric aerosol particles of diameters <1 μm and 2.5-10 μm, while PM 10 contained aerosol particles of both these particle size ranges. At the same time, the formation of secondary aerosols under the influence of precipitation also made PM 2.5 concentration subject to a weaker wet removal effect than PM 10 concentration. e distributions of V PM2.5 and V PM10 were further compared under different intervals of C PM2.5/PM10 . Figure 5 shows the relationships between V PM2.5 and V PM10 for different values of C PM2.5/PM10 . ere was a good positive correlation between V PM2.5 and V PM10 in different intervals, with correlation coefficients >0.7, but there were differences in correlations between them for different C PM2.5/PM10 intervals. e values of C PM2.5/PM10 were mainly concentrated within the range of −0.6 to 0.5. When the values of C PM2.5/ PM10 were negative (−0.6 to −0.1), the fitted line had a slope greater than 1 in each ratio interval, showing that the values of V PM10 exceeded those of V PM2.5 . When C PM2.5/PM10 increased (0.1-0.5), the fitted line had a slope less than 1 in each ratio range, and the values of V PM2.5 exceeded those of V PM10 . When C PM2.5/PM10 approached 0 (−0.1 to 0.1), the fitted line had a slope close to 1, and V PM2.5 and V PM10 demonstrated relatively consistent variability.

Effect of Rainfall Intensity on the PM Mass Concentration.
e analysis in the above two sections provided a preliminary understanding of the distributions of precipitation and PM concentration in the pollution process. erefore, we then combined the two to investigate the law governing the effect of precipitation on variations of PM concentration.

Advances in Meteorology 5
Studies have shown that aerosol concentrations before precipitation are closely related to the wet removal effect of precipitation [16,18], so this section concentrates on analyzing the relationships between V PM2.5 and V PM10 , rainfall intensity, and values of M beg PM2.5 and M beg PM10 . Figure 6 shows the distributions of average M beg PM2.5 and M beg PM10 , average rainfall intensity, and mean values of V PM2.5 and V PM10 with intervals of 0.05. When the average rainfall intensity was <0.5 mm/h, the mean values of V PM10 were in the interval 0.4-0.8, and average M beg PM10 also had a low value (<140 μg/ cm 3 ; Figure 6(a)). is indicated that precipitation resulted in aerosol hygroscopic growth and that formation of secondary aerosol was stronger than its wet removal, which thereby led to the increase in PM 10 mass concentration. When the average rainfall intensity was >1.0 mm/h, the mean values of V PM10 were negative (i.e., −0.66 to −0.20), showing the wet removal effect caused by precipitation. It is noteworthy that when average rainfall intensity was 0.5-1.0 mm/h and average M beg PM10 was 110-180 μg/m 3 , there were both positive and negative values of V PM10 , with absolute values lower than those for rainfall intensities of <0.5 and >1.0 mm/h. e effect of precipitation on PM 2.5 mass concentration (Figure 6(b)) was similar to that of PM 10 . e range of mean rainfall intensity responsible for wet growth was mainly <0.5 mm/h, but the range of mean rainfall intensity responsible for wet removal of PM 2.5 was significantly higher than that of PM 10 , with a value >1.4 mm/h. Particles with smaller size need greater rainfall intensity to be significantly cleared by precipitation. At the same time, the initial value of PM 2.5 or PM 10 mass concentration was the dominant factor under the rainfall intensity range in which wet growth and wet removal had similar effects. Precipitation was more likely to have a wet removal effect for a greater initial value of mass concentration, and, on the contrary, wet growth effect was more likely. e threshold of PM 10 mass concentration was 150 μg/m 3 , and that of PM 2.5 mass concentration was 95 μg/m 3 , relatively consistent with the results of Xu et al. [23] on initial concentration threshold of PM 2.5 corresponding to wet removal in Shanghai.

Effect of Relative Humidity on the Change of PM Mass
Concentration during Precipitation. Aerosol hygroscopic growth and secondary aerosol formation play a key role in formation of heavy pollution. Sun [29] clarified the important role of rapid growth of secondary aerosols in heavy pollution formation in the North China Plain, pointing out that secondary aerosol accounts for 70% of heavy pollution,    [45,46]. Precipitation significantly increases the relative humidity of the environment, and the elimination of data when the wind speed reaches 3 m/s or more will maintain the low wind-speed environmental conditions. erefore, precipitation would also affect the formation of secondary aerosols. Figure 7 shows the distributions of V PM10 in each relative humidity range (0-100% with 1% intervals). Different panels of Figure 7 correspond to different levels of precipitation. e relative humidity was <40% only during LR, with the values of V PM10 ranging from −0.32 to −0.2, showing that precipitation played a significant role in wet removal of PM 10 in a dry environment. As the relative humidity increased, wet growth gradually played a more significant role. e PM 10 mass concentration in the relative humidity range of 40-65% was subject to the combined action of wet removal and wet growth. When relative humidity was 65-90%, PM 10 mass concentration increased under the impact of aerosol hygroscopic growth and secondary aerosol formation, while in the near-saturated relative humidity range (90-100%), PM 10 mass concentration was slightly affected by wet removal, with values of V PM10 concentrated within −0.05 to 0. As the precipitation level increased (MR, HR, and TR), the wet removal to PM 10 mass concentration was enhanced more obviously, especially in the near-saturated relative humidity range (90-100%). For the relative humidity range (40-90%) under the combined action of wet removal and wet growth, wet removal played a leading role in lower humidity (about  Meteorology 60%) and greater rainfall intensity, but wet growth played a leading role in higher humidity (∼90%) and lower rainfall intensity. e variations of V PM2.5 had distributions similar to those of PM 10 (figure omitted), but the wet removal effect of precipitation was weakened in the near-saturated relative humidity range (90-100%), and the wet growth effect was enhanced under the higher humidity (∼90%) and lower rainfall intensity.

Conclusions
In this study, the characteristics of precipitation, distributions of PM concentrations, RH conditions, and evolution of PM under the effect of precipitation during pollution days in Hubei from 2015 to 2017 were comprehensively analyzed. Due to wet removal and wet growth, the PM concentrations varied widely under different rainfall intensities and relative humidity environments.
(1) e LR occurred most frequently in the pollution process, with XY being the most affected, up to 587 h. Occurrence of MR was lower than LR by an order of magnitude, with a mean value of 21 h. e occurrence of HR was consistent in each region, mainly in the range of 5-12 h. SR was rare. e greater the rainfall intensity was, the larger the difference in occurrence of precipitation was at national air quality monitoring stations in the same area. Rainfall was more concentrated during 19: 00-23:00.  8 Advances in Meteorology similar effects, precipitation was more likely to produce a wet removal effect for greater initial values of mass concentration, and, on the contrary, a wet growth effect was more likely. e threshold of PM 10 mass concentration was 150 μg/m 3 and that of PM 2.5 mass concentration was 95 μg/m 3 . (4) Precipitation played a significant role in wet removal of PM 10 in a dry environment (relative humidity <40%), with a weaker wet removal effect in the nearsaturated relative humidity range (90-100%). When relative humidity was 40-90%, the effects of wet removal and wet growth were similar. As the precipitation level increased (MR, HR, and TR), the wet removal of PM 10 mass concentration was enhanced more obviously. Wet removal played a leading role in lower humidity (∼60%) and greater rainfall intensity, but wet growth played a leading role in higher humidity (∼90%) and lower rainfall intensity. e effect of precipitation on PM 2.5 was similar to that on PM 10 , but the wet removal effect of precipitation was weakened in the near-saturated relative humidity range (90-100%), and the wet growth effect was enhanced under higher humidity (∼90%) and lower rainfall intensity.
Data Availability e PM 2.5 and PM 10 mass concentration, precipitation, relative humidity, and wind-speed data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.