Role of sea fog over the Yellow Sea on air quality with the direct effect of aerosols

Abstract

In this study, we investigate the impact of sea fog over the Yellow Sea on air quality with the direct effect of aerosols for the entire year of 2016. Using the WRF-CMAQ two-way coupled model, we perform four model simulations with the up-to-date emission inventory over East Asia and dynamic chemical boundary conditions provided by hemispheric model simulations. During the spring of 2016, prevailing westerly winds and anticyclones caused the formation of a temperature inversion over the Yellow Sea, providing favorable conditions for the formation of fog. The inclusion of the direct effect of aerosols enhanced its strength. On foggy days, we find dominant changes of aerosols at an altitude of 150–200 m over the Yellow Sea resulted by the production through aqueous chemistry (~12.36% and ~3.08% increases in sulfate and ammonium) and loss via the wet deposition process (~-2.94% decrease in nitrate); we also find stronger wet deposition of all species occurring in PBL. Stagnant conditions associated with reduced air temperature caused by the direct effect of aerosols enhanced aerosol chemistry, especially in coastal regions, and it exceeded the loss of nitrate. The transport of air pollutants affected by sea fog extended to a much broader region. Our findings show that the Yellow Sea acts as not only a path of long-range transport but also as a sink and source of air pollutants. Further study should investigate changes in the impact of sea fog on air quality in conjunction with changes in the concentrations of aerosols and the climate.

1. Introduction

The Yellow Sea, located between East China and the Korean Peninsula, has frequent episodes of sea fog. A number of studies have reported various mechanisms of the formation of sea fog over the Yellow Sea such as longwave radiation cooling at the fog top (Yang et al., 2018; Yang and Gao, 2019) and ocean upwelling caused by strong tidal mixing (Cho et al., 2000). Researchers have typically categorized sea fog in this region as advection fog (Yang et al., 2018; Yang and Gao, 2019; Zhang et al., 2009), which mainly occurs from April to July. Owing to the contrast between the thermal inertia of land and the ocean, quickly warmed continental air flows over the relatively colder sea surface, resulting in the formation of a temperature inversion. Moreover, anticyclone winds transport humid air from the southern region to the Yellow Sea.Zhang et al. (2009) observed a temperature inversion at a 100–350 m height over the Yellow Sea from March to June; however, stable layers at 150–450 m height, also a favorable condition for sea fog formation, remained in July and were accompanied by larger amounts of moisture supplied by the ocean. From the analysis of sounding data, the number of foggy days exceeded 15 days per month, the Chinese coastal region showing the most frequent occurrence of fog with more than 80 days per year.

Sea fog is a hazardous phenomenon that occurs within the marine atmospheric boundary layer. As it lowers visibility, sea fog adversely impacts a number of activities in marine and coastal regions (Yang and Gao, 2019). It also affects trace gases and aerosols by chemical and physical processes such as wet deposition and secondary aerosol formation via aqueous-phase chemistry, which causes more rapid formation of theses pollutants than gas-phase chemistry (Ervens, 2015; Faust et al., 1993; Wang et al., 2016; Yuan et al., 2015). Many studies have identified the aqueous-phase production of sulfate (${\mathrm{S}\mathrm{O}}_4^{-}$) from S(IV) by hydrogen peroxide (${\mathrm{H}}_2{\mathrm{O}}_2$), ozone (${\mathrm{O}}_3$) catalyzed by iron ($\mathrm{F}\mathrm{e}$) and manganese ($\mathrm{M}\mathrm{n}$), methyl hydroperoxide (MHP), peroxyacetic acid (PAA), and nitrogen dioxide (NO2) depending on the acidity of water droplets (Li et al., 2018; Sarwar et al., 2013; Seinfeld & Pandis, 2006; Wang et al., 2016). In the presence of clouds or fog, sulfate production by aqueous-phase oxidation, particularly that caused by ${\mathrm{H}}_2{O}_2$, is dominant, while sulfate production by gas-phase oxidation caused predominantly by OH is relatively negligible (Ervens, 2015; Itahashi et al., 2018; Shen et al., 2012). Many recent studies have focused on the aqueous-phase production of secondary organic aerosols (SOAs) (Carlton et al., 2008; Fahey et al., 2017; Lim et al., 2010; Lu et al., 2019; Mandariya et al., 2019).

A few studies have examined aqueous-phase aerosol production over East Asia. With elevated levels of aerosol concentrations from various biogenic (e.g., biomass burning and dust) and anthropogenic emission sources, several countries in East Asia, especially China, have suffered severe regional-scale haze characterized by large fractions of secondary aerosol species under high humidity and stagnant conditions (Gao et al., 2016; Huang et al., 2019; Li et al., 2017, 2018; Seo et al., 2017; Shao et al., 2019; Wang et al., 2016; Yuan et al., 2015). As a pathway of long-range transport of air pollutants, the Yellow Sea, with sufficient supply of moisture, provides a medium for the various chemical and physical processes of inorganic and organic secondary aerosols (Jeon et al., 2018; H. Lee et al., 2019; S. Lee et al., 2019). Therefore, their adverse effects on not only source and coastal regions but also a much broader array of downwind regions have also been a source of concern.

The addition of aerosols directly affects the solar radiation budget by scattering and absorption. A reduction in solar radiation reaching the ground changes the surface air temperature and planetary boundary layer height (PBLH) (Giorgi, 2002; Matus et al., 2019; Wong et al., 2012). Indirectly, aerosols acting as cloud condensation nuclei (CCN) and ice nuclei (IN) generate more cloud droplets, smaller cloud droplets, and narrowed droplet size distribution which results in altered cloud characteristics such as albedo, lifetime, and subsequent cloud or fog duration and cycles (Ervens, 2015; Guo et al., 2019; Kim et al., 2018; Liu et al., 2019; Souri et al., 2020; Twomey et al., 1984; Yu et al., 2013). In addition, some aerosol species, such as mineral dust, black carbon, and brown carbon, heat the atmosphere by absorbing solar radiation. The resulting warming may locally enhance atmospheric stability, leading to a decrease in cloud cover through the semi-direct effect (Johnson et al., 2019; Peng et al., 2020; Talukdar et al., 2019; Tuccella et al., 2020). Meteorological changes such as solar radiation, air temperature, PBLH, wind speed, precipitation, and cloud coverage may also affect air quality. Previous studies have shown that concentrations of major gaseous and particulate pollutants at the surface increase due to reduced ventilation and vertical transport associated with the direct effect of aerosols (Jung et al., 2019; Sekiguchi et al., 2018; Wang et al., 2014; Xing et al., 2017; Yoo et al., 2019). Over East Asia, which exhibits a high aerosol burden, the direct effects of aerosols generally increase concentrations of surface air pollutants by 7.87% - 34% over land, while negative responses were found over the ocean, the result of a reduction in precursors from continents and slight increases in the PBLH (Jung et al., 2019; Sekiguchi et al., 2018).

The aim of this study is to examine the effect of sea fog in conjunction with the direct effect of aerosols on air quality. Using the WRF-CMAQ two-way coupled model over East Asia for the entire year of 2016, we conducted four model simulations to study the impact of the direct effect of aerosols on the formation of sea fog and its impact on the chemical and physical processes of gaseous and aerosol pollutants. The structure of this paper is as follows: Section 2 explains the WRF-CMAQ two-way modeling system and observational data, Section 3 presents the results of the four model simulations and Section 4 the summary and discussion.

2. Overview of the Modeling System and Observational Data

2.1. WRF-CMAQ two-way coupled model and input data

We used the WRF-CMAQ two-way coupled model, which has two sub-models, Weather Research and Forecasting (WRF) model version 3.8 and CMAQ model version 5.2, developed and released by the U.S Environmental Protection Agency (EPA). Receiving aerosol information from the chemistry model within this coupled model framework enables us to study the aerosol radiative feedback effect. A detailed description of the WRF-CMAQ two-way model can be found in Wong et al. (2012). Unlike the conventional off-line CMAQ model, which is driven by hourly meteorological data, the WRF-CMAQ two-way coupled model furnishes more frequent (e.g., every 360 seconds based on a run-time parameter) meteorological information to the CMAQ sub-model. This is one of the key features of the WRF-CMAQ two-way coupled model we relied on to capture the behavior of sea fog, a frequently diverging phenomenon in this study.

The two sub-models share a single domain with a 27 km horizontal grid spacing over East Asia, covering the eastern part of China, Korea, and Japan (Figure 1). The detailed configuration of the WRF-CMAQ two-way model is listed in Table 1. For this study, we used SAPRC-07 (Carter, 2010) for the gas phase and AERO6 for the aerosol chemical mechanisms in CMAQ. Anthropogenic emissions (KORUS-AQ Emission inventory version 5.0) data, collected by Konkuk University (Woo et al., 2012), consists of multiple emissions inventories such as the modified Clean Air Policy Support System (CAPSS) 2015, provided by the National Institute of Environmental Research and Konkuk University in South Korea, and the scaled Comprehensive Regional Emissions for Atmospheric Transport Experiments (CREATE) version 3.0 as well as Studies of Emissions and Atmospheric Composition, Clouds and Climate Coupling by Regional Surveys (SEAC4RS) for the remaining countries in Asia. The spatial resolution is 3 km in South Korea and 0.1° for the rest of Asia. Additionally, to estimate dynamic biogenic emissions, we used the Model of Emissions of Gases and Aerosols from Nature (MEGAN) version 3.0 (Guenther et al., 2020) with various input data such as the leaf area index of vegetation (LAIv = LAI/vegetation cover fraction (VCF)) and meteorological data. We provided WRF output for the meteorological input data and a modified version of the MODIS LAI product (Yuan et al., 2011) for the LAI, which is generated based on the background values from 10 years of MODIS data. For the VCF, we used the Visible Infrared Imaging Radiometer Suite (VIIRS) global green vegetation fraction from the Joint Polar Satellite System (JPSS).

Figure 1.

Map of the model domain.

Table 1.

Configuration of the WRF-CMAQ two-way model

Version	WRF version 3.8 and CMAQ version 5.2
Microphysics	Morrison double-moment scheme
Longwave and shortwave radiation	RRTMG scheme
Land Surface	The Pleim-Xiu land surface model (Pleim and Xiu, 1995; Xiu and Pleim, 2001)
Surface layer	Pleim-Xiu surface layer (Pleim, 2006)
Planetary boundary layer	The ACM2 planetary boundary layer model (Pleim, 2007a, 2007b)
Cumulus parameterization	Kain-Fritsch (KF2) scheme with sub-grid cloud fraction interaction with radiation (Alapaty et al., 2012; Herwehe et al., 2014)
Four-Dimensional Data Assimilation (FDDA)	∙ Indirect soil moisture and temperature nudging technique (Pleim and Gilliam, 2009; Pleim and Xiu, 2003) ∙ A FDDA option every 6 hours above the PBL for the temperature, the water vapor mixing ratio, and wind components (magnitude of ${10}^{-5}$) (Hogrefe et al., 2015)
Initial and boundary conditions for meteorology	National Centers for Environmental Prediction FNL (final) operational global analysis data
Chemical mechanism	SAPRC-07 and AERO6
Horizontal advection	YAMO
Vertical advection	WRF omega formula
Horizontal diffusion	Multiscale
Vertical diffusion	ACM2
Initial and boundary conditions for chemistry	The CMAQ model version 5.3 with the in-line dust module covering the entire Northern Hemisphere
Call frequency between the WRF and the CMAQ	3:1 (We set the timestep for the WRF at 120 seconds. The WRF and the CMAQ exchange meteorological data and aerosol information every 360 seconds)

To generate chemical initial and boundary conditions for the CMAQ model, we set up the CMAQ model with the in-line dust module covering the entire Northern Hemisphere (HCMAQ) with a 108 km × 108 km spatial resolution grid and 44 vertical layers. We simulated three months for the model spin-up; detailed information about the preparation of the emission inventory over the Northern Hemisphere is available in Vukovich et al. (2018). Unlike the traditional approaches to creating boundary conditions (e.g., CMAQ default profile (static), GEOS-CHEM), HCMAQ can provide dynamic chemical boundary conditions with a consistent physical configuration while accounting for the outflow of dust emissions originating from the Taklamakan and Gobi Deserts and biomass burning primarily from Southeast Asia. We compared the output of HCMAQ to multiple satellite datasets such as the NASA Ozone Monitoring Instrument (OMI)/Aura vertical Ozone (O₃) profile 1-Orbit level 2 data for tropospheric O₃ and formaldehyde (HCHO), NASA OMI tropospheric NO₂ level 2 version 3 data for tropospheric NO₂, the Measurement of Pollution in the Troposphere (MOPITT) level 2 version 8 data for carbon monoxide (CO), and the Moderate Resolution Imaging Spectroradiometer (MODIS) Aerosol Optical Depth (AOD). These spatial comparisons with statistical results appear in Table S1 and Figures S1 – S5. Based on satellite comparisons, we seasonally scaled gaseous species to ensure more reliable boundary conditions.

We ran the simulation for all of 2016 and spent December 2015 on model spin-up prior to the time period of interest. We conducted four experiments: without and with aerosol feedback (NF and YF, respectively), and without aqueous chemistry for stable layers with dT/dz > —4K/km (moist-adiabatic lapse rate) over the ocean which indicates layers with a temperature inversion (NFAQ and YFAQ). Differences between the experiments of NF and NFAQ (YF and YFAQ) are considered as the impact of the aqueous chemistry of sea fog under two different conditions. A brief explanation of the four simulations is shown in Table 2.

Table 2.

Description of the four experiments in this study.

Experiments	Aerosol Feedback	Aqueous chemistry in layers with temperature inversion over the ocean
NF	No	Yes
YF	Yes	Yes
NFAQ	No	No
YFAQ	Yes	No

2.2. Ground measurements

For surface concentrations in South Korea, we used surface observational data from the Air Quality Monitoring Station (AQMS) network operated by the National Institute of Environmental Research. The network measures the concentrations of hourly air pollutants such as CO, NO₂, O₃, PM₁₀, PM_2.5, and SO₂. We obtained the meteorological data, such as air temperature and wind fields, from the Automatic Weather System network by the Korea Meteorological Administration (KMA) and downloaded atmospheric sounding data from the Upper-Air Data Center, which operates in the Department of Atmospheric Science at the University of Wyoming. The data contain the vertical distribution of atmospheric properties such as atmospheric pressure, geopotential height, temperature, dew-point temperature, frost-point temperature, relative humidity, water vapor mixing ratio, wind speed, wind direction, potential temperature, equivalent potential temperature, and virtual potential temperature at certain pressure levels. The measured data are generally available two times per day (00 and 12 UTC) or three times per day (00, 06, 12 UTC). We gathered the data from two stations near the Yellow Sea: Qingdao (#54857, 120.33°E, 36.06°N) and Baengnyeongdo (#47102, 124.63°E, 37.97°N).

3. Results

3.1. The response of fog formation over the Yellow Sea to the direct effect of aerosols

The direct effect of aerosols reduces the amount of solar radiation reaching the ground which moderates meteorological variables such as air temperature, the PBLH, wind fields, and the cloud fraction proportional to aerosol loading (Jung et al., 2019; Wong et al., 2012; Xing et al., 2015). These meteorological changes caused by aerosol feedback can lead to the formation of sea fog over the Yellow Sea.

To validate the meteorological variables of the model, we compared modeled output to the KMA measurements for air temperature and wind fields as well as the sounding observations at the two stations located in Qingdao and Baengyeongdo for water vapor mixing ratios for 2016 (Table S2 and Figure S6). We found that the model performance was relatively comparable to these measurements.

Figure 2 shows the spatial distribution of 10 m and 400 m wind fields, the difference between the temperature at 400 m and 2 m, and the total liquid water content (LWC) in the PBL over the Yellow Sea for April 2016. The model clearly simulated the patterns of anticyclone winds, which transport warm and moist air to the Yellow Sea, at 10 m for both NF (Figure 2(a)) and YF (Figure 2(e)). YF showed slightly lower wind speed (–0.05 ms^–1) because of a reduced difference between the temperature on land and at sea (shown in Figure S7) as the anticyclone forming mechanism (Zhang et al., 2009). A southwesterly wind at 400 m was also slightly weaker for YF ($1.87 m{s}^{-1}$) (Figure 2(f)) than for NF ($1.94 m{s}^{-1}$) (Figure 2(b)) because of the stable conditions over land and reduced thermal contrast resulting from the direct effect of aerosols. Anticyclones at the surface (10 m) and southwesterly winds at 400 m led to occurrences of radiative fog over the Yellow Sea, and the significant temperature difference between 400 m and 2 m appeared, especially along the coasts of China and the Korean Peninsula, as shown in Figures 2(c) and (g). The higher temperature differences indicated higher stability of the atmosphere, which helped trap moisture brought from the southern region by anticyclones. As we can see from Figures 2(d) and (h), the spatial distribution of the LWC in the PBL displayed circular patterns along the anticyclones and showed a high LWC near the Chinese coast because of the presence of a larger temperature difference there than near the coast of the Korean Peninsula. YF showed a slightly larger temperature difference over the Yellow Sea near the Chinese coast (shown in Figure S8) as well as higher LWC ($0.08 g{m}^{-3}$) than NF ($0.05 g{m}^{-3}$).

Figure 2.

Spatial distribution of 10 m wind fields, 400 m wind fields, temperature differences (400 m air temperature –2 m air temperature), and the total LWC in the PBL for April 2016.

Figure 3 displays a vertical cross-section of air temperature and the total LWC in the PBL along with the purple dotted line depicted in Figure 2 (c). Compared to that of NF, the warm air mass originating from the Chinese coast in YF did not spread as far toward the Korean Peninsula (Figure 3(c)). The manifestation of colder air associated with the direct effect of aerosols, however, caused the temperature gradient near the Chinese coast to increase, stabilizing the atmospheric conditions and creating increasingly favorable conditions for the formation of sea fog. As a result, YF showed increased total LWC. Although the atmosphere at altitudes below 400 m contained most of the LWC, the atmosphere at higher altitudes, up to 820m, still contained an abundance of LWC (Figure 3(d)).

Figure 3.

Vertical cross-section of the air temperature and the total LWC along the purple dashed line in Figure 2(c) for April 2016.

We plotted the daily total and cloud LWC over the Yellow Sea from the NF and YF simulations in Figures 4(a) and (b). Both NF and YF showed high peaks of total LWC, mostly from March to July, which is known for prevailing advection sea fog over the Yellow Sea (Zhang et al., 2009). Most of the high peaks of total LWC during the sea fog season were associated with clouds (Figure 4(b)). Relatively small peaks appearing in the winter may have been caused by snow (Jan. 23) or the cooling of air masses moving over cold water (Dec. 20). As we mentioned, YF yielded higher LWC than NF, with annual means of 0.05 gm^–3 and 0.04 gm^–3 respectively. In April, they increased to 0.41 gm^–3 (YF) and 0.37 gm^–3 (NF), respectively.

Figure 4.

Daily variation of the total and cloud LWC and comparisons between the cloud LWC and temperature differences between 400 m and 2 m for 2016.

To identify cases of advection sea fog caused by temperature inversions, we displayed the hourly cloud LWC and temperature differences between 400 m and 2 m over the Yellow Sea from NF and YF in Figure 4 (c) and (d). We defined cases of sea fog as those in which the cloud LWC exceeded 0.15 gm^–3 with a positive temperature difference between 400 m and 2 m and analyzed only the most common cases in both NF and YF (350 hours from February to June).

3.2. The response of air quality to the direct effect of aerosols and sea fog over the Yellow Sea

The direct effect of aerosols on meteorological variables further induces notable changes in air quality. For instance, reduced solar radiation reaching the ground lessens photochemical reactions and ventilation, resulting in reduced ozone formation and radical production, both of which can affect aerosol formation. Conversely, reduced ventilation with a lower PBLH can increase concentrations of chemical species near their source regions (Jung et al., 2019; Xing et al., 2017). To validate model performance regarding air quality, we compared hourly modeled fine particulate matter (PM_2.5) to the measurement data at the AQMS sites. Both the NF and YF showed promising model performance. The statistical results are shown in Table 3.

Table 3.

Statistical results of the comparison between the Korean $P{M}_{2.5}$ ground measurements and the model output for 2016 (RMSE: root-mean-square error, CORR: correlation, MAE: mean-absolute error, IOA: index of agreement, MB: mean bias)

Variables	Experiments	RMSE	CORR	MAE	IOA	MB
PM2.5	NF	13.11	0.71	10.55	0.73	−8.99
	YF	13.00	0.70	10.31	0.74	−8.15

Figure 5 shows the average spatial distribution of concentrations of PM_2.5 and fine $S{O}_4^{2-}$, nitrate ($N{O}_3^{-}$), and ammonium ($N{H}_4^{+}$), and the impact of the direct effect of aerosols (YF-NF) at the surface for the fog cases. Because of the direct effect of aerosols, concentrations of PM_2.5 and fine $S{O}_4^{2-}$, $N{O}_3^{-}$, and $N{H}_4^{+}$ increased over land, especially over eastern China and South Korea. The increases over eastern China were as high as $4.83 \mu g{m}^{-3}$ (11.53%) for PM_2.5, $1.01 \mu g/m^3$ (13.39%) for fine $S{O}_4^{2-}$, $2.21 \mu g{m}^{-3}$ (18.74%) for fine $N{O}_3^{-}$, and $1.00 \mu g{m}^{-3}$ (18.30%) for fine $N{H}_4^{+}$. Compared to these increases, those over South Korea were relatively small ($1.14 \mu g{m}^{-3}$ (5.19%) for PM_2.5, $0.25 \mu g{m}^{-3}$ (5.46%) for fine $S{O}_4^{2-}$, $0.64 \mu g{m}^{-3}$ (8.89%) for fine $N{O}_3^{-}$, and $0.28 \mu g{m}^{-3}$ (7.78%) for fine $N{H}_4^{+}$), owing to the smaller aerosol loadings. Along the coasts of eastern China and the Korean Peninsula, we found slightly negative contributions of the direct effect of aerosols over the ocean (~ -4.93%) because of the reduced outflow of PM_2.5 precursors from the land under stable conditions (Sekiguchi et al., 2018).

Figure 5.

The direct effect of aerosols on concentrations of PM_2.5, fine sulfate, fine nitrate, and fine ammonium for fog cases.

Illustrating the effect of sea fog on air quality, Figure 6 displays the vertical distributions of the LWC and fine $S{O}_4^{2-}$, $N{O}_3^{-}$, and $N{H}_4^{+}$ concentrations at Qingdao (120.33°E, 36.06°N), Incheon (126.71°E, 37.46°N), and the Yellow Sea. The difference between NF and NFAQ (YF and YFAQ) represents the production and loss of aerosol concentrations by chemical and dynamical processes in the presence of fog over the Yellow Sea. The LWC was greatest at an altitude of 100 – 150 m and YF exhibited greater values, as we noted in Section 3.1. In both Qingdao and Incheon, the enhancement of sulfate concentrations was significant near the surface ($0.73-1.07 \mu g{m}^{-3}$), and the increase in the direct effect of aerosols was greater by 6.84 – 16.30% because of the enlarged $S{O}_2$ concentration as a precursor as well as accelerated aqueous-phase chemistry with a higher LWC. Similarly, $N{H}_4^{+}$ concentrations exhibited similar changes ($0.24-0.35 \mu g{m}^{-3}$) since $S{O}_4^{2-}$ and $N{O}_3^{-}$ must be neutralized (Seinfeld and Pandis, 2006). For $N{O}_3^{-}$, the contributions of sea fog with and without the direct effect of aerosols were opposite near the surface. Despite the high solubility of $HN{O}_3^{-}$, acting as a major sink, the production of $N{O}_3^{-}$ was dominant with the direct effect of aerosols, in which the weather conditions of higher relative humidity, lower air temperature, and excess $N{H}_4^{+}$ allowed an enhancement in the production, $HN{O}_3+N{H}_3 \leftrightarrow N{H}_{4}N{O}_3$.

Figure 6.

Vertical distribution of the cloud LWC, changes in aerosol concentrations (fine sulfate, nitrate, and ammonium) at three locations (Qingdao, Incheon, and the Yellow Sea) for fog cases.

Interestingly, maximum changes in aerosols appeared at an altitude of 150 – 200 m over the Yellow Sea. The higher LWC with the direct effect of aerosols induced not only active sulfate oxidation at high altitudes ($0.48 \mu g{m}^{-3}$ for NF-NFAQ, $0.51 \mu g{m}^{-3}$ for YF-YFAQ) but also stronger wet deposition of sulfate in the PBL. According to the results of integrated process rates (IPRs), which helped us identify dominant sinks or sources of pollutants (shown in Figure S9), the cloud processes, including aqueous-phase chemistry, in-cloud scavenging, and wet deposition, acted as a sink for sulfate in the PBL; above the PBL, however, it was a major source. For the same reason, $N{H}_4^{+}$ exhibited similar changes, with high peaks above the PBLH ($0.05 \mu g{m}^{-3}$ for NF-NFAQ, $0.09 \mu g{m}^{-3}$ for YF-YFAQ). For nitrate, we found significant decreases in concentrations at high altitudes because of the high solubility of $HN{O}_3^{-}$ ($-0.27 \mu g{m}^{-3}$ for NF-NFAQ, $-0.23 \mu g{m}^{-3}$ for YF-YFAQ), but we noted the production in conjunction with the direct effect of aerosols, especially near the surface. In addition to the chemical changes, the production and loss of aerosols through aqueous chemistry modulated the incoming solar radiation, showing more or less scattering and absorption (Ervens, 2015) which affects the stability of the atmosphere as well as the contributions of vertical advection and dry deposition processes to aerosol concentrations (Figure S9).

Figures 7 and 8 show the spatial distribution of the concentrations of gaseous and aerosol pollutants at the surface and at an altitude of 150 m. Because of the transport of aerosols formed in a fog over the Yellow Sea and removal of aerosols and trace gases in the fog by wet deposition and in-cloud scavenging, we observed mixed increases and decreases in both aerosol and gaseous concentrations overland (Figure S10). These results indicate that the effects of sea fog on air quality are not limited to coastal regions. In fact, they showed enhanced concentrations of surface air pollutants on the coasts of eastern China and the Korean Peninsula. The direct effect of aerosols, resulting in stronger temperature inversions near the Chinese coast, also enhanced the concentration of air pollutants at the surface as well as at an altitude of 150 m. The increased $S{O}_4^{2-}$ concentration, together with decreased $S{O}_2$ and $H_2{O}_2$ concentrations strongly indicated the aqueous-phase production of $S{O}_4^{2-}$ over the Yellow Sea. Average increases in sulfate on foggy days were $0.48 \mu g{m}^{-3}$ (11.73%) for NF-NFAQ and $0.51 \mu g{m}^{-3}$ (12.36%) for YF-YFAQ. On extremely foggy days, with the highest LWC (April 13) and second highest LWC (April 1) (Figure 4(b)), sulfate concentrations increased by $0.78-1.03 \mu g{m}^{-3}$ (9.73 – 26.10%) (shown in Figures S10 and S11). Fog over the Yellow Sea changed the concentrations of $N{O}_3^{-}$ and $N{H}_4^{+}$ by an average of ~ -3.79 % and ~3.08 %, respectively.

Figure 7.

Spatial distribution of the differences among aerosol species caused by fog over the Yellow Sea at the surface and at an altitude of 150 m for fog cases.

Figure 8.

Spatial distribution of the differences among gaseous species caused by fog over the Yellow Sea at the surface and at an altitude of 150 m for fog cases.

4. Summary and Discussion

This study examined the effect of sea fog over the Yellow Sea with the direct effect of aerosols on air quality over East Asia in 2016. We applied the WRF-CMAQ two-way model, which not only linked air quality and meteorology, including the feedback of aerosols and its subsequent effects on meteorology but also received more frequent meteorological information from the WRF model to generate the effects of sea fog on air quality, which is sensitive to meteorological variability. The Yellow Sea, located between the eastern coast of China and the Korean Peninsula, is a path of the long-range transport of air pollutants. The high anthropogenic emissions over East Asia, in addition to biogenic emissions from various sources, exacerbated the impact of the direct effect of aerosols as a function of aerosol loading.

From April to July 2016, sea fog frequently accumulated over the Yellow Sea due to temperature inversions, which occur when warm continental air flows over the colder ocean. Therefore, the inclusion of the direct effect of aerosols, resulting in reduced solar radiation reaching the ground and lower air temperatures, increased the temperature gradient near the Chinese coast even though it reduced the outflow of continental air masses under stable conditions over land. The increased temperature inversion trapped more moisture brought from southern regions and eventually increased the LWC, accelerating aqueous-phase oxidation and removals of aerosols by wet deposition and in-cloud scavenging.

With regard to the impact of sea fog on air quality, we found that changes in gaseous and aerosol concentrations over land and the ocean varied considerably. Overland, even far from coastal regions, we found mixed effects on both gaseous and aerosol concentrations resulting from the transport of aerosols newly-formed in fog and removed by wet deposition and in-cloud scavenging. In coastal areas, enhancements of $S{O}_4^{2-}$ and $N{H}_4^{+}$ concentrations were dominant, and the magnitude of changes increased with the direct effect of aerosols. Because of stabilized atmospheric conditions, both aerosols and precursors remained near the surface, so their production occurred through aqueous chemistry. The occurrence of fog negatively contributed to $N{O}_3^{-}$ concentration because of the high solubility of $HN{O}_3^{-}$; meteorological changes by the direct effect of aerosols (e.g., lower temperature and higher relative humidity), however, enhanced aerosol formation near the surface. Over the Yellow Sea, we observed similar changes in aerosol concentrations, but only above the PBLH. Higher relative humidity in the PBL, together with the direct effect of aerosols, caused stronger wet deposition that greatly exceeded their production by aqueous-phase chemistry. As a result, maximum changes in aerosol concentrations appeared at an altitude of 150–200 m over the Yellow Sea.

Even though we observed a significant impact of sea fog over the Yellow Sea on air quality, it is worth noting that the slight underestimation of aerosol concentrations (Table 3) could have underestimated the impact of the direct effect of aerosols and changes in sea fog on air quality. Thus, determining the true impact would require additional work such as assimilating surface or satellite observations and updating the bottom-up emission inventory with inverse modeling.

This study found that the direct effect of aerosols could enhance temperature inversions over the Yellow Sea and that it modulates the impact of sea fog on air quality as a source or sink of gaseous and aerosol concentrations. As the frequency and intensity of sea fog over the Yellow Sea can be affected by not only aerosol concentrations related to emission control policies but also climatological changes that enhance or reduce differences between the temperature of the land and the ocean, further studies could examine such variability in the impact of sea fog over the Yellow Sea.