Previous studies have emphasized that the decrease in photolysis rate at the surface induced by the light extinction of aerosols could weaken ozone photochemistry and then reduce surface ozone. However, quantitative studies have shown that weakened photochemistry leads to a much greater reduction in the net chemical production of ozone, which does not match the reduction in surface ozone. This suggests that in addition to photochemistry, some other physical processes related to the variation of ozone should also be considered. In this study, the Weather Research and Forecasting with Chemistry (WRF-Chem) model coupled with the ozone source apportionment method was applied to determine the mechanism of ozone reduction induced by aerosols over central East China (CEC). Our results showed that weakened ozone photochemistry led to a significant reduction in ozone net chemical production, which occurred not only at the surface but also within the lowest several hundred meters in the planetary boundary layer (PBL). Meanwhile, a larger ozone gradient was formed in the vertical direction, which led to the high concentrations of ozone aloft being entrained by turbulence from the top of the PBL to the surface and partly counteracting the reduction in surface ozone. In addition, contribution from dry deposition was weakened due to the decrease in surface ozone concentration. The reduction in the ozone's sink also slowed down the rate of the decrease in surface ozone. Ozone in the upper layer of the PBL was also reduced, which was induced by much ozone aloft being entrained downward. Therefore, by affecting the photolysis rate, the impact of aerosols was a reduction in ozone not only at the surface but also throughout the entire PBL during the daytime over CEC in this study. The ozone source apportionment results showed that 41.4 %–66.3 % of the reduction in surface ozone came from local and adjacent source regions, which suggested that the impact of aerosols on ozone from local and adjacent regions was more significant than that from long-distance regions. The results also suggested that while controlling the concentration of aerosols, simultaneously controlling ozone precursors from local and adjacent source regions is an effective way to suppress the increase in surface ozone over CEC at present.
Ozone in the troposphere, especially in the planetary boundary layer (PBL),
is a well-known secondary air pollutant that is seriously harmful to human
health and vegetation (Haagen-Smit and Fox, 1954). As an important source of
tropospheric ozone, the photochemical production of ozone is significantly
affected by ozone precursors (i.e.,
Quantitative studies have suggested that, because of the impact of aerosols
via their affecting photolysis rates, 2 %–17 % of surface ozone has decreased
(Jacobson, 1998; J. Li et al., 2011; Wang et al., 2016). However, these
studies also showed that ozone net production decreased much more (Cai et
al., 2013; Wang et al., 2019), which did not match the magnitude of the
reduction in surface ozone. For example, a modeling study conducted by J. Li et
al. (2011) showed that the average reduction in surface ozone over central
East China (CEC) was
At present, air pollution in China is characterized by the “air pollution
complex”, which shows both aerosols (especially fine particulate matter
PM
In this study, the fully coupled “online” model system, the Weather Research and Forecasting with Chemistry (WRF-Chem) model, was applied to simulate air pollutants over CEC in October 2018. The impact of aerosols on ozone via influencing the photolysis rate was quantitatively studied by using process analysis, through a comparison between control and sensitivity simulations. In addition, with the application of the ozone source apportionment method (Gao et al., 2016, 2017) that we developed and coupled with the WRF-Chem model system, the ozone contributions and their changes induced by aerosols over typical cities in CEC were discussed quantitatively in this study. This paper is organized as follows. A description of the model setting, used data, and scenario design is presented in Sect. 2. The results and discussion of the subject are presented in Sect. 3. And finally, we end with the conclusions in Sect. 4.
The model system used in this study, the WRF-Chem model, is a fully coupled online 3-D Eulerian meteorological and chemical transport model that has been globally applied in air quality research (Tie et al., 2013; Zhang et al., 2014; M. Gao et al., 2018; Hu et al., 2019). The version of the WRF-Chem model we used in this study is 3.9.1.1, and detailed introductions of the meteorological parts and chemical parts can be found in Skamarock et al. (2008) and Grell et al. (2005), respectively.
Regarding the simulation settings, two nested domains (Fig. 1) were set up
with grid sizes of
Model domain. Hundreds of observations are used for model
validation; locations and types of observation stations are shown in
Major configuration options of WRF-Chem used for this study.
Since the light extinction of aerosols can impact ozone in two ways, it is necessary to distinguish the direct impact on ozone in this study. Thus, two parallel experiments were designed in this study: (1) photolysis rate calculation without the presence of aerosol optical properties (Exp1) and (2) photolysis rate calculation including considering the optical properties of all kinds of aerosols (Exp2). By comparing the results between Exp1 and Exp2, the impact of aerosols on ozone via influencing the photolysis rate can be determined. Both experiments started at 00:00 UTC on 29 September 2018 and ended at 00:00 UTC on 31 October 2018. The first 2 d were designated the spin-up period.
Many kinds of data were used in this study. The initial and boundary
meteorological and chemical conditions were provided by the National Centers
for Environmental Prediction (NECP) final (FNL) operational global analysis
data and outputs of the Community Atmosphere Model with Chemistry (CAM-chem;
Lamarque et al., 2012). Regarding the emissions used in this study,
anthropogenic emissions were provided by the Multi-resolution Emission
Inventory for China (MEIC;
Meteorological observations (temperature, wind direction and wind speed)
from 110 stations and air pollutants (ozone,
Due to secondary pollutant properties, tropospheric ozone is highly
dependent on the photochemical reactions of its precursors (
In this study, 20 geographic source regions were set up in the model domain.
The North China Plain and eastern China are two economic hubs in China and have
suffered serious air pollution in recent years (L. T. Wang et al., 2014; Z. F. Wang, 2014;
Ding et al., 2016; Kang et al., 2019). As shown in Fig. 1, the two areas are
separated into 10 source regions based on administrative divisions. Other
provinces belonging to China and areas outside of China in the model domain
are far from CEC but may also influence the air quality of CEC under
favorable synoptic conditions. Thus, these regions were combined and
defined as several source regions. Other details of the source regions are
listed in Table S1, which can be seen in the Supplement. In
addition to the geographic source regions, chemical boundary conditions
provided by MOZART-4 outputs, named O
Although the WRF-Chem model has been widely used in air quality research,
the performance varies dramatically when dealing with different domains,
episodes and parameterization settings. In this study, common model
performance metrics (IOA, index of agreement; MB, mean bias; RMSE, root mean
square error; MNB, mean normalized bias; MFB, mean fractional bias) were
used to validate meteorological factors (
For meteorological factors and air pollutants, observation data from more than 100 stations distributed in D2 (Fig. 1b) were collected. Considering the large dataset size, averaged model performance metrics are listed in Table 2. The benchmarks shown in brackets follow the recommended values suggested by Emery et al. (2001) and EPA (2005, 2007). In addition, the model performance of meteorological factors and air pollutants at each station is displayed by the Taylor diagram (Taylor, 2001; Gleckler et al., 2008) as shown in Figs. S1 and S2, which are available in the Supplement.
Mean model performance metrics for meteorological factors and air pollutants. The values that do not meet the benchmarks are denoted in bold.
Regarding meteorological factors,
For air pollutants, good agreement was found between the simulations and
observations since the IOAs of ozone,
Figure 2a and b show the comparison of observed (dark gray dots) and
predicted (red line, denotes results in Exp2)
Time series of simulated
As shown in Fig. 2, when the concentrations of PM
The impact of aerosols on the photolysis rate occurs not only at the surface
but also along the vertical direction. To investigate the aerosols'
impact on the photolysis rate, the
Mean profiles of
Under clean conditions (Fig. 3a), PM
At the surface, the mean distributions of daytime PM
Mean distributions of PM
High concentrations of PM
Chemical and physical process analysis (Zhu et al., 2015; Gao et al., 2016) was implemented to discuss the mechanism of the surface ozone reduction induced by aerosols via influencing the photolysis rate in the four representative cities. The following processes were considered: chemistry (CHEM, which is the sum of ozone chemical production and loss of ozone in atmosphere; this contribution is the same as the “ozone net production” which has been mentioned in other studies), advection (ADV, which is caused by the transport effects of wind fields) and vertical mixing (VMIX, which is caused by turbulence in the PBL and is closely dependent on turbulence intensity and the vertical gradients of ozone). In addition, for surface ozone, the contribution of dry deposition (DRY, which is an important sink of ozone and is highly related to concentration of surface ozone and dry-deposition velocity) should also be considered. More information on process analysis of the WRF-Chem system is available in Zhang et al. (2014), Gao et al. (2016) and the Supplement.
Figure 5 illustrates the mean surface ozone concentrations and process analysis results of the four cities during 07:00–18:00 (the results of each city are presented in Fig. S5 in the supplementary material). As shown in Fig. 5a, surface ozone began to be reduced by the impact of aerosols starting at 08:00. From then, ozone reduction accumulated until the afternoon, with a maximum value of 11.7 ppb at 14:00. Similar to the process analysis results of other studies (Kaser et al., 2017; Tang et al., 2017; Xing et al., 2017; Xu et al., 2018), the variation in surface ozone was mainly controlled by VMIX, DRY and CHEM during the daytime (Fig. 5b and c). The contribution of CHEM at the surface was generally below zero, which showed that the chemical consumption of ozone was equal to or stronger than the chemical production of ozone at the surface level. As an important removal of surface ozone, the contribution of DRY was always negative during the daytime. On the contrary, the contribution of VMIX was positive which was the key factor leading to the increase in surface ozone during the daytime.
Averaged surface ozone concentrations and process analysis
results of the four cities in the daytime. Mean ozone concentrations from Exp1
and Exp2 are presented in
The reduction in surface ozone induced by aerosols can be decomposed into
changes in process contributions (Exp2–Exp1), which are shown in Fig. 5d.
The contributions of CHEM decreased significantly during the daytime, which
was mainly due to the reduction in ozone chemical production caused by
weakened ozone photochemistry. Distinct from the change in CHEM, the changes
in DRY and VMIX were increased during daytime. From 08:00 to 14:00, the
reduction in CHEM was more significant than the increases in VMIX and DRY,
which made surface ozone continue to decrease during this period. After
14:00, the increases in VMIX and DRY almost counteracted the reduction in
CHEM. Quantitative results (Table 3) showed the ozone reduction and the
accumulated changes in each process contribution at 14:00. The reduction in
CHEM (
The reduction in surface ozone at 14:00 LT and the corresponding accumulated changes in process contributions.
Because the contribution of DRY is usually negative to surface ozone, the increase in the change in DRY suggested that the strength of dry deposition was weakened during the daytime. Contribution from dry deposition is highly related to surface ozone concentration and dry-deposition velocity. In Exp1 and Exp2, factors impacting on dry-deposition velocity such as land use and vegetation were not changed which indicated that dry-deposition velocity did not change (Wesely, 1989). However, the concentration of surface ozone decreased due to the impact of aerosols which finally led to the weakening of dry deposition of ozone. By contrast, the increase in change in VMIX suggested the enhancement of the vertical mixing process. Since vertical mixing occurs in the entire PBL, the change in VMIX can impact not only surface ozone but also the ozone aloft, which suggests that the change in ozone may also occur in the entire PBL.
Averaged vertical distributions of process contributions as a
function of time from 06:00 to 18:00 LT. Data are spatially sampled. All
the grids within the administrative regions of the four cities are collected
and averaged which can represent the situation of the four cities.
The averaged vertical changes in process contributions of the four representative cities are presented in Fig. 6 (the results of each individual city are quite similar and are presented in Figs. S6–S9 in the Supplement). CHEM showed positive contributions aloft in both Exp1 and Exp2 (Fig. 6a and e, respectively), which resulted from strong ozone photochemical production. At the surface, it showed negative or weak positive contributions which was attributed to the much stronger chemical loss at the surface caused by ozone-consuming species (i.e., NO). Figure 6i shows that the reduction in CHEM induced by aerosols occurred not only at the surface but also within the lowest 500 m during the daytime. VMIX (Fig. 6b and f) showed a negative contribution in the upper layer and a positive contribution in the lower layer, which indicated a high concentration of ozone aloft being entrained downward to the surface by turbulence during the daytime (Zhang and Rao, 1999; J. Gao et al., 2018). The impact of aerosols enhanced the contributions of VMIX; thus, the change in VMIX showed a positive value within the lowest 300 m and negative values in the upper layer in the PBL. ADV (Fig. 6c and g) showed small contributions, and there was no significant change in ADV caused by the impact of aerosols. NET_DIF reflects the sum of the changes in all of the process contributions, and its distributions showed that, by affecting photolysis rate, the impact of aerosols led to the reduction in ozone occurring not only at the surface but also in the whole PBL (Fig. 6l). In the lower layer of the PBL, the reduction in CHEM was primarily responsible for the reduction in ozone, while the increase in VMIX partly counteracted the reduction in ozone. In the upper layer of the PBL, the decrease in VMIX played an important role in decreasing ozone aloft.
The contribution of VMIX is closely related to ozone vertical gradients and turbulence exchange coefficients. Studying the changes in the two factors is helpful to investigate the enhancement of VMIX induced by aerosols. As shown in Fig. 7a and b, via influencing the photolysis rate, the impact of aerosols did not cause obvious changes in the exchange coefficients since the exchange coefficient profiles were almost the same as those from Exp1 and Exp2. However, the ozone gradient from Exp2 was larger than that from Exp1, which suggested that the enhancement of VMIX induced by aerosols was mainly associated with the increase in the ozone gradient. Because of the impact of aerosols, the chemical reduction in ozone was more significant in the lower layer than in upper layer in the PBL (Fig. 6i), which led to smaller concentrations of ozone in the lower layer and consequently formed a larger vertical gradient (Fig. 7c). Therefore, high concentrations of ozone aloft would be entrained from the top of the PBL to the surface, which led to the enhancement in VMIX. In addition, similar features also occurred in each representative city which can be seen in Fig. S10 in the Supplement.
Averaged vertical profiles of turbulence exchange coefficients
Figure 8 illustrates the average ozone contributions from geographic source
regions to surface ozone in the four cities from Exp1 and Exp2, and the
changes in each ozone contribution induced by aerosols are also presented.
For the representative cities, surface ozone was mainly contributed by local
contributions and the contributions from adjacent source regions (Fig. 8a, b, d, e, g, h, i and k). For example, surface ozone over BJ and TJ was
mainly contributed by ozone from these cities themselves and Hebei province. For SJZ and
ZZ, ozone from their respective provinces (HB and HN) contributed significantly more
than ozone from other regions did. In addition,
O
Averaged ozone contributions and changes induced by aerosols from
geographical source regions to BJ
With the impacts of aerosols, ozone from local and adjacent source regions
decreased more significantly than ozone from further source regions did
(right column in Fig. 8). For each city, the first four source regions whose
ozone contribution changed the mean ozone concentration the most from
13:00 to 16:00 are listed in Table 4. For BJ and TJ, which are defined as
independent source regions, ozone from local regions to BJ and TJ decreased by
The first four source regions from which ozone contribution changes the mean ozone concentration the most from 13:00 to 16:00 LT in each city. Local region and source region where the city is located are denoted as bold.
Currently, in China, the concentrations of surface ozone increase annually,
which is considered closely related to the decrease in PM
To more clearly understand the impact of aerosols on ozone via affecting the photolysis rate, the WRF-Chem model was applied to simulate air pollutants over CEC in October 2018. Comprehensive model validations demonstrated the model performance in simulating air quality over CEC during this period. By comparing the results between the control and sensitive simulation, the mechanism of the impacts of aerosols on ozone was quantitatively studied. With the application of the ozone source apportionment method that we coupled into the WRF-Chem model, the impact of aerosol on the source–receptor relationship of ozone was also discussed.
Our results showed that, because of the light extinction of aerosols, the attenuation of incident solar irradiance caused the decrease in the photolysis rate below the PBL and then weakened ozone photochemistry. In this case, the net chemical production of ozone was significantly decreased within the lowest several hundred meters in the PBL. The decrease in surface ozone led to the weakening of dry deposition of ozone which slowed down the decrease in surface ozone to a certain extent. More importantly, the significant reduction in the net chemical production formed a larger ozone vertical gradient. And more air mass aloft with a high concentration of ozone was entrained downward from the top of the PBL to the surface, which also partly counteracted the reduction in ozone net chemical production. Changes in the three processes together led to the reduction in surface ozone. In addition, ozone in the upper layer of the PBL was also reduced, which was also induced by much ozone aloft being entrained downward. Therefore, by affecting the photolysis rate, the impact of aerosols reduced ozone not only at the surface but also in the entire PBL during the daytime over CEC in this study.
The ozone source apportionment results showed that, for the four representative cities in CEC (BJ, TJ, SJZ and ZZ), ozone from local and adjacent regions decreased by 6.9, 6.8, 4.6 and 5.8 ppb, respectively, which accounted for 41.4 %–66.3 % of the reduction in surface ozone in these cities. This suggested that the impact of aerosols on ozone from local and adjacent regions is more significant than that from more distant regions. In recent years, with the implementation of the toughest-ever clean air policy in China, aerosols have decreased, whereas ozone increases year by year. Our results suggest that while controlling the concentrations of aerosols, controlling ozone precursors from local and adjacent regions is an effective way to suppress the increase in surface ozone.
All the observations and model outputs mentioned in this study are publicly
available. Observations on Ozone,
The supplement related to this article is available online at:
JG, YL and BZ designed the research. BH and LW provided
the observed data of
The authors declare that they have no conflict of interest.
We want to acknowledge the support from the SUSTC Presidential Postdoctoral Fellowship. The simulated results in this study were calculated using computational resources provided by the Southern University of Science and Technology. All the observations and model outputs mentioned in this study are available by contacting Ying Li via liy66@sustech.edu.cn. Finally, we hope people all over the world can be healthy and defeat COVID-19. Jinhui Gao hopes Hua Wang can overcome the difficulties that have occurred during this tough period and find her inner peace.
This research has been supported by the National Key Research and Development Program of China (grant no. 2016YFA0602003), Shenzhen Peacock Teams Plan (grant no. KQTD20180411143441009), National Natural Science Foundation of China (grant nos. 41905114, 41961160728 and 41575106), Science and Technology Planning Project of Guangdong Province of China (grant no. 2017A050506003), and China Postdoctoral Science Foundation (grant nos. 2019M662169 and 2019M662199).
This paper was edited by Kostas Tsigaridis and reviewed by two anonymous referees.