Impact and pathway of halogens on atmospheric oxidants in coastal city clusters in the Yangtze River Delta region in China

Abstract

Halogens (chlorine, bromine, and iodine) are known to profoundly influence atmospheric oxidants (hydroxyl radical (OH), hydroperoxyl radical (HO₂), ozone (O₃), and nitrate radical (NO₃)) in the troposphere and subsequently affecting air quality. However, their impact on atmospheric oxidation and air pollution in coastal areas in China is poorly characterized. In this study, we use the WRF-CMAQ (Weather Research and Forecasting-Community Multiscale Air Quality) model with full halogen chemistry and process analysis to assess the influences and pathways of halogens on atmospheric oxidants in the Yangtze River Delta (YRD) region, a typical coastal city cluster in China. Halogens cause the annual OH radical increase by up to 16.4% and NO₃ decrease by up to 45.3%. O₃ increases by 2.0% in the YRD but decreases by 3.3% in marine environment. Halogen induced changes in atmospheric oxidants lead to a general increase of atmospheric oxidation capacity by 5.1% (maximum 48.4%). The production rate of OH (P_OH) in the YRD is enhanced by anthropogenic chlorine through both increased HO₂ pathway and hypohalous acid photolysis pathway, while P_OH over ocean is enhanced by oceanic halogens through converting HO₂ into hypohalous acid. Anthropogenic chlorine enhances both O₃ and NO₃ production (P_NO3) rates through influencing their precursors while oceanic halogens reduce P_NO3 and directly destroy ozone. Iodine contributed most (on average of 91% in oceanic halogens) in reducing production rates of oxidants. Thus, halogen emissions and potential effects of halogens on air quality need to be considered in air quality policies and regulations in the YRD region.

1. Introduction

Recently, China has achieved a significant decline in primary pollutants (Zheng et al., 2018). However, the level of surface ozone (O₃), a typical secondary pollutant, has gradually increased and become one of major contributors to air pollution in China, and the Yangtze River Delta (YRD) region suffered a combined pollution of O₃ and fine particulate matter (PM_2.5). Secondary air pollutants are formed through photochemical or oxidation processes (i.e., volatile organic compounds (VOCs) oxidized by radicals to form O₃) in the atmosphere, presenting a nonlinear correlation with their precursors. Therefore, knowledge in mechanisms between atmospheric oxidants and pollutants is required (Lu et al., 2019; Tan et al., 2019).

Halogens are known as a major contributor to stratospheric O₃ depletion events (Lary, 1997). Reactive halogen species (RHS) may also play an important role in the chemistry in the troposphere. Recent studies have given a comprehensive overview on halogen chemistry in the troposphere (Saiz-Lopez and von Glasow, 2012; Simpson et al., 2015). RHS can destroy O₃ through catalytic cycle, change the partitioning of hydrogen oxides (HO_X) and nitrogen oxides (NO_X), and oxidize VOCs to produce peroxy radicals leading to O₃ formation in polluted area (Atkinson et al., 2006). In coastal areas where sea spray aerosols (SSA) are emitted to the atmosphere, heterogeneous reactions between halides and gas-phase compounds occur, activating RHS into the atmosphere (Osthoff et al., 2008).

Several studies using large-scale models have been conducted to examine the effect of halogen chemistry in the troposphere. Sarwar et al. (2012), Parrella et al. (2012) and Saiz-Lopez et al. (2014) implemented initial chlorine, bromine and iodine chemistry in hemispheric CMAQ (Community Multiscale Air Quality), GEOS-Chem and CAM-Chem (Community Atmospheric Model with Chemistry) model. For chlorine, the chemistry reduced the Normalized Mean Bias (NMB) of predicted ozone for 1.4% and predicted levels of nitryl chloride (ClNO₂) are similar to the observed values reported across the United States. For bromine, the modeled BrO reproduced major features of the satellite observations including the increase with latitude and the seasonal maxima in different latitudes. For iodine, the simulated vertical profile of IO was in agreement with field measurements over the tropical oceans. This result proposed the existence of a “tropical ring of atomic iodine” and subsequent ozone loss ranged from 11% to 27%. Sarwar et al. (2019) implemented bromine and iodine chemistry and oceanic emissions in CMAQ to study their role in background O₃ in the northern hemisphere. The bromine/iodine chemistry enhances BrO and IO levels by a factor of ~2.0 and ~1.5 on average, respectively. Predicted IO levels were similar to these reported observed values ranging from 0.4 to 1.0 pptV while BrO levels were underpredicted. Halogen chemistry could reduce ozone biases by 2~6 ppbv at most coastal and oceanic sites with less deterioration in other regions. More regional studies have been conducted with updated halogen chemistry in testbeds among the U. S., tropical east Pacific, Europe and northern Indian Ocean (Badia et al., 2019; Li et al., 2019; Mahajan et al., 2021; Womack et al., 2023). These modeling studies exhibited well in capturing either vertical or horizontal profile of IO and BrO levels against the observations, and suggested a non-negligible influence on ozone as well as other related pollutants.

The modeling studies in China mainly focused on the effect of chlorine chemistry on ClNO₂ uptake from aerosols and surface O₃ in WRF-Chem (Weather Research and Forecasting model coupled with Chemistry) and CMAQ (Li et al., 2016; Qiu et al., 2019; Zhang et al., 2017). Simulations indicated an enhancement on O₃ by 5~20% while ClNO₂ predictions reproduced the same order of magnitude as observed values with marginal differences due to uncertainties in meteorological or chemical schemes. Huang et al. (2021) implemented CMAQ with embedded bromine and iodine chemistry to examine the ocean halogen-mediated O₃ loss and vertical profile over Asia-Pacific. Results showed that NMB for O₃ in January was reduced by 6.3% and a higher index of agreement in O₃ with 0.61 for ocean halogen chemistry. O₃ loss exhibited a strong seasonal cycle and could reach −41~−44% in maximum over the ocean. Li et al. (2020) and Li et al. (2021c) adopted WRF-Chem with state-of-the-art halogen chemistry and sources to study the influence in atmospheric oxidants and haze pollution by halogens in northern China. Simulations with halogen chemistry reproduces the routine gas-phase pollutants successfully although ClNO₂ was underestimated with a NMB of - 25.3% and bromine species were simulated with an average of 71 pptV compared to the observation of 105 pptV. The inclusion of halogen chemistry and sources resulted from the enhancements of atmospheric oxidation and loadings of fine aerosols in northern China by 10~20% and 21% on average, respectively. A few more studies have focused on chemical reactions between oceanic halogens and ship-originated air pollutants and hydroxyl radicals (OH) over the East Asia seas (Fan and Li, 2022; Li et al., 2021b).

Modeling resolutions in previous studies were majorly 27km or 12km (Fan and Li, 2022; Huang et al., 2021; Li et al., 2020; Mahajan et al., 2021; Womack et al., 2023). The coarse resolutions may limit further detailed analysis in a specific region. On the other hand, human activities could also affect the emission of halogen precursors, which were scarcely considered, putting uncertainties in the modeling studies. Therefore, we presented a comprehensive study using the WRF-CMAQ model with 3 km resolution focusing on the YRD region, a populated coastal area. Elaborated halogen chemistry, and an up-to-date anthropogenic chlorine emission inventory were adopted in this study. We quantitatively assessed the impact of RHS on air oxidants and atmospheric oxidation capacity from different perspectives in 2018. The process analysis (PA) was employed to perform a detailed breakdown of individual impacts and contributions on air quality within reactions from halogens. The results of this study may serve as an overview with respect to the role of comprehensive halogen chemistry in the aggravation of atmospheric oxidants and oxidation capacity in the YRD city-cluster.

2. Material and method

2.1. Model configuration

In this study, Weather Research and Forecasting-Community Multiscale Air Quality Model system (WRF-CMAQ) was used to assess the impact of halogen chemistry on regional air quality. The WRF model v4.2.1 was used to simulate meteorological fields and the CMAQ model v5.3.3 with process analysis was used to simulate chemical fields. Specific parameterization schemes were included in Table S1. We adopted detailed halogen chemistry and emission parameterizations by Sarwar et al. (2019), including 38 gas-phase and 4 heterogeneous chemical reactions for bromine, and 44 gas-phase and 9 heterogeneous chemical reactions for iodine with carbon bond 6 mechanism (Luecken et al., 2019) in CMAQ to investigate the overall impact of halogens.

2.2. Data

For meteorological field simulated by WRF, 1° × 1° NCEP FNL Operational Model Global Tropospheric Analyses dataset was used (ds038.2, ds351.0 and ds461.0). Observation datasets ds351.0 and ds461.0 were prepared for meteorological data assimilation (Liu et al., 2008). The Multi-resolution Emission Inventory in China (MEIC) (www.meicmodel.org.cn) (Li et al., 2017; Zheng et al., 2018) with 0.25°×0.25° spatial resolution, and 0.1°×0.1° shipping emission inventory model (SEIM) (Liu et al., 2016; Liu et al., 2019) were used for routine anthropogenic emissions. The temporal allocation and the spatial interpolation for MEIC and SEIM have been conducted in accordance to CMAQ modeling domain. Base year of both MEIC and SEIM is 2017 without routine biogenic emissions.

For halogen emissions, sources of marine halogen species were considered (Fig.S1): an oceanic source of halocarbons from biological activity (CHBr₃, CH₂Br₂, CH₂BrCl, CHBrCl₂, CHBr₂Cl, CH₃I, CH₂ICl, CH₂IBr, CH₂I₂), and inorganic iodine from ocean (HOI and I₂). Marine halogen emissions were calculated online following the parameterization method by Sarwar et al. (2019). Halocarbon emissions were estimated using monthly average climatological chl-α concentrations obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS, oceancolor.gsfc.nasa.gov/cgi/l3). Inorganic iodine was estimated using sea surface temperature, wind speed, surface ozone concentration and sea surface iodide ions from the model (MacDonald et al., 2014).

In our study, an up-to-date (2017) and high-resolution (4 km) anthropogenic inventory (Fig.S2) of major chlorine precursors (HCl, fine particulate Cl⁻, Cl₂ and HOCl) for the YRD region was used (Yi et al., 2021). The emission inventory was developed based on a bottom-up method with detailed emission factors (sources of emissions, emission control policies, etc.) and county-level activity data. Although the influence of anthropogenic chlorine emissions other than natural sources remained uncertain and negligible from a global perspective (Wang et al., 2019), numbers of observations in China have confirmed the potential existence of chlorine species derived from anthropogenic sources (Tham et al., 2016; Wang and Ruiz, 2017; Yang et al., 2018). With the increasing industrial emissions and anthropogenic activities, it is necessary to include anthropogenic chlorine emissions in evaluating the overall impact of halogen species on air quality in the YRD region.

2.3. Simulation setup

A three-layer nested grid was used in this study as shown in Fig.S3. Horizontal resolutions for each domain were 27 km, 9 km and 3 km, respectively, covering Shanghai and most of the YRD region with 15 vertical layers of varying thickness from the surface to 50-hPa altitude. Halogen chemistry would make an impact in the coastline of the YRD, causing influences on both regional transport and atmosphere chemistry. Initial and boundary data for the outer domain were simulated under clean air condition. Initial and boundary data for inner domains were extracted from simulated concentrations of the outer domains using MEIC and SEIM without halogen emissions and chemistry. The simulation with PA was conducted for the entire year of 2018, with a spin-up period of 15 days. Besides, we divided 12 months into 4 seasons: March, April, and May for spring; June, July, and August for summer; September, October, and November for autumn; and January, February, and December for winter.

In this study, we conducted six sets of simulations shown in Table 1. Differences between BASE and HAL cases represented the comprehensive impact of halogen chemistry on air quality. Four more sensitive cases were conducted by zeroing out different emission input of halogens. By comparing BASE and EMIS_OCN cases we could investigate impact of oceanic halogen emissions. EMIS_I, EMIS_BR and EMIS_CL were 3 sensitive cases that used halogen emissions respectively to investigate individual impacts by different halogens.

Table 1.

Simulation Cases Setup

Cases	Anthropogenic chlorine emission	Bromine emission	Iodine emission	Halogen chemistry
BASE	×	×	×	×
HAL	√	√	√	√
EMIS_OCN	×	√	√	√
EMIS_I	×	Zeroed out	√	√
EMIS_BR	×	√	Zeroed out	√
EMIS_CL	√	×	×	√

Note: × refers to emissions that were excluded from emission inventories, Zeroed out refers to inline emissions that were scaled to 0 in emission factors using Detailed Emissions Scaling, Isolation and Diagnostics Module (DESID) in CMAQ.

2.4. Atmospheric oxidation capacity calculation

The physical and chemical states of the atmosphere are influenced in the presence of trace gases such as VOCs. The oxidizing capacity of the atmosphere determines the amount of these trace gases and their removal rate (Elshorbany et al., 2009). In this study, we adopted a mathematical method to quantify the effects of halogens on atmospheric oxidation capacity (AOC). AOC is defined as the multiply of major oxidants, their reactants, and corresponding reaction rates (Equation 1) (Li et al., 2020). AOC is a uniformed numerical value to quantify contributions of different oxidants towards atmospheric oxidation capacity (Xue et al., 2015).

$$\text{AOC}=\sum _{i=1}^l([{\text{Ox}}_i]\times \sum _{j=1}^m(n_j\times [C_j]\times k_{i,j}))$$ (1)

In Equation 1, [${\text{Ox}}_{\mathrm{i}}$] is the concentration of oxidant $i$ (molecules cm⁻³), $l$ and $m$ represents the number of oxidants and reactant species, ${\mathrm{n}}_j$, is the number of carbon atoms in reactant $j$ [${\mathrm{C}}_{\mathrm{j}}$] is the concentration of reactant $j$ (molecules cm⁻³), and ${\mathrm{k}}_{i,j}$ is the second-order reaction rate constant of oxidant $i$ and reactant $j$ (cm³ molecules⁻¹ s⁻¹). We included all VOC species defined in CMAQ, carbon monoxide (CO) and methane (CH₄) as reactants to calculate AOC. Oxidants that react with reactants above were OH, O₃, NO₃ (nitrate radical), and Cl (chlorine radical). The reactions between oxidants and reactants are shown in the Table S2.

3. Results and discussion

3.1. Model evaluation

In order to evaluate the model results, both WRF and CMAQ simulations were evaluated using ground-level observation data obtained from national air quality forecast platform of MEE (https://air.cnemc.cn:18007/). The calculation of the indicators was listed in Table S3. For WRF model, 6 representative monitoring stations from different cities in the YRD region were selected. Simulated meteorological parameters in 2018 by WRF presented consistent temporal trends and close value ranges with observations in daily average time series comparison (Fig.S4 and Table S4). WRF model is capable of reproducing surface temperature and humidity (with −4.3% and −12.7% in Normalized Mean Bias, NMB, respectively) although wind speed was slightly underestimated with a NMB of −22.9%. In general, WRF model produced acceptable meteorological results in the research domain and could be used as input for the CMAQ model.

Model performances of CMAQ in BASE and HAL cases were evaluated by comparing simulated pollutant species with observations from 40 national air quality monitoring sites in the YRD region. Time-series comparison was conducted between simulations and observations of pollutants (Fig.S5), and verifications of simulated daily average concentration were listed in Table 2.

Table 2.

Evaluations of simulated pollutants by CMAQ in BASE and HAL cases

Pollutants (μg/m3)	R		NMB (%)		RMSE (μg/m3)
	BASE	HAL	BASE	HAL	BASE	HAL
MDA8-O3	0.7320	0.7227	13.78	14.73	33.29	34.03
NO2	0.7768	0.7801	7.05	6.78	10.72	10.63
PM2.5	0.8629	0.8643	46.04	45.86	27.60	28.10
SO2	0.7405	0.7403	11.43	11.21	2.77	2.78

In BASE case, CMAQ model presented relatively good correlation and less bias on simulating MDA8-O₃, NO₂ (nitrogen dioxide) and SO₂ (sulfur dioxide), with NMB range from 7% to 14% and Root Mean Squared Error (RMSE) between 3-33%. The implementation of halogen chemistry and oceanic emissions in EMIS_OCN case could represent the influence of halogens on model performances (Table S5). The results indicated that halogen chemistry could slightly improve model performance especially for O₃. NMB of O₃ was reduced by 1.44% while NMB of other pollutants were reduced by 0.3~0.6%. The inclusion of anthropogenic chlorine emissions in HAL case indicated the enhancement effect on the burden of ozone pollution by chlorine, which would be further discussed. Still, the NMB for other routine pollutants were lower than NMB in BASE case, representing an acceptable result. Simulated PM_2.5 was overestimated with NMB of 46% but in good correlation with R of 0.86.

The CMAQ simulations were subjected to some uncertainties in emission inventories. For MEIC, calculated uncertainties were the lowest for NO_X (−28%~33%) and highest for NH₃ (−58%~117%). The overall uncertainties were reduced in comparison of previous YRD emission inventories (An et al., 2021). For SEIM, the Monte Carlo method reported a total uncertainty for CO₂ at about 3%, and other pollutants were about 4~6% (Liu et al., 2016). For anthropogenic chlorine emissions, the uncertainties were estimated using Monte Carlo methods with uncertainties range of −26.8~16.8% (HCl), −57.4%~56.8% (Cl⁻), −105.7%~166.5% (Cl₂), and −75.1%~94.5% (HOCl) (Yi et al., 2021). Furthermore, aerosol species mapping in different chemical mechanisms might account for overestimate in PM_2.5 (Chen et al., 2022) and biases from meteorological fields (i.e. wind speed underestimation) might affect the diffusion and advection of pollutants. Overall bias and R of all simulated pollutants between observations were within the acceptable range, suggesting that CMAQ model results could be used to investigate the effect of halogen chemistry on air quality over the YRD region.

3.2. Simulated reactive halogen species

Spatial distributions of simulated RHS were shown in Fig.1. In HAL case, CMAQ model simulated a prominent loading (over 1000 pptV) of inorganic chlorine species (Cl_y = Cl + Cl₂ + ClO + HCl + HOCl + ClNO₃ + ClNO₂ + BrCl + ICl) in the YRD region, especially over the region with higher anthropogenic chloride emissions. Cl_y loading peaked in winter and was lowest in summer on lands, while the loading of Cl_y exhibited an opposite trend along the coastal area. The difference between temporal variation of Cl_y on the land and the ocean might be due to the low temperature induced accumulations of N₂O₅ in winter and higher biogenic activities and activations of chloride from SSA in summer (Fig.S6).

Fig.1.

Simulated seasonal average of daily-maximum of Cl_y (left column), Br_y (middle), and I_y (right) (pptV) in HAL case

In our study, inorganic bromine (Br_y = Br + Br₂ + BrO + HBr + HOBr + BrNO₃ + BrNO₂ + BrCl + IBr) and iodine (I_y = I + I₂ + IO + I₂O₂ + I₂O₃ + I₂O₄ + HI + HOI + INO₃ + INO₂ + INO + ICl + IBr) were also estimated with a non-negligible amount (over 0.4 pptV) on lands and a higher level along the coast. Br_y and I_y exhibited a consistent seasonal variation which were generally influenced by the solar radiation and sea surface temperature. Connections between emissions and simulations of RHS were examined by evaluating IPR (integrated process rate) results. The spatial distribution and seasonal variation of Br_y and I_y were roughly in line with the emissions of oceanic bromine and iodine species (Fig.S8) which was higher in summer (over 5 pptV and 15 pptV in coastal areas) and lower in winter. Meanwhile, wind direction in China is governed by the Asian monsoon (Li et al., 2020). In summer, the southeaster monsoon blows oceanic emissions of RHS into the land, enhancing RHS level on lands and along the coast. In contrast, the norther monsoon in winter blows the air from the land to the ocean and thus prevents the ocean emitted RHS from entering Mainland China (Fig.S9).

Simulated halogen species were evaluated using available observations and measurements till now. For ClNO₂, a reservoir of NO_x (oxides of nitrogen) and reactive chlorine radicals, simulations in the current study showed that it peaked in winter and reached the lowest in summer (Fig.S6). For the higher ClNO₂ in coastal area than the inlands in Fig.S6a, the result is probably due to the chlorine in SSA in the HAL case. Differences of simulated N₂O₅ concentration between HAL and BASE cases and particulate chlorine (PCl) were shown in Fig.3-1. The increase of ClNO₂ exhibits a similar pattern to the decrease of N₂O₅ in the HAL case compared to BASE case (Fig.S7a, b), indicating the enhancement effect of chlorine in SSA on ClNO₂. The ambient atmosphere is blown from urban YRD to coastal area due to west-northern monsoon in cooler seasons and creating a favorable condition for N₂O₅ to accumulate over the coastline because of shorter sunlight duration and atmosphere circulation. Besides, simulated PCl concentration in spring is the second highest among 2018 and in winter it is the lowest. Although NO_x emissions from vessels may compensate for the loss of N₂O₅ along ship tracks, these conditions may lead to a higher ClNO₂ in coastal area in spring. Several observations of reactive chlorine have been made in different locations of China. Despite no available public datasets, the simulated results of ClNO₂ and other reactive chlorine species were compared with observations throughout YRD within the same time periods, as listed in Table 3. For observations outside the YRD, we intended to demonstrate the magnitude of the results. Generally, CMAQ simulated reasonable results of reactive chlorine species that were the same order of magnitude and within the range of observations.

Fig.3.

Annual average of relative changes in major oxidants (%) due to anthropogenic chlorine emissions (EMIS_CL-BASE) and oceanic halogen emissions (EMIS_OCN-BASE)

Table 3.

The comparisons of simulated reactive chlorine species with observed concentrations in China

Time period	Species	Location	Observations	Simulations**	References
Jun 20~Jul 9, 2014	ClNO2	Northern China Plain*	>350 at most of the nights and peaked at 2070 (pptV)	72.9±90.3 in daily-average, peaked at 1155	Tham et al. (2016)
Jul 24~Aug 27, 2014	ClNO2	Mt. Tai, Shandong*	54±106 on average and 2065 on maximum (pptV)	82.2±39.6 in daily-average, peaked at 1834	Wang et al. (2017b)
Jun 11~18, 2017	ClNO2	Beijing*	174.3±262.0 on average and reached 1440 on maximum (pptV)	104.4±139.9 in daily-average, peaked at 924	Zhou et al. (2018)
Sep 8~14, 2018	ClNO2	Hok Tsui, Hong Kong*	Peaking at 840 and 940 (pptV)	164.0±173.0 in daily-average, peaked at 990	Dai et al. (2020)
May 17~Jun 17, 2018	HCl	Taizhou	0.6 in average and peaked more than 1 ppbV	0.25±0.17 in daily-average, peaked at 0.8	Li et al. (2021a)
May 17~Jun 17, 2018	Particulate Cl−	Taizhou	2.63 in average and peaked at 9.0 μg m−3	1.72±1.0 in daily-average, peaked at 13.8	Li et al. (2021a)
May 26~Jun 14, 2019	ClNO2	Changzhou	0.15±0.20 averaged with 1-min maximum concentration of 1.3 (ppbV)	0.15±0.18 in daily-average, peaked at 0.8	Li et al. (2023)
May 26~Jun 14, 2019	Cl2	Changzhou	0.026±0.036 in average with maximum of 0.52 (ppbV)	0.018±0.012 in daily-average, peaked at 0.66	Li et al. (2023)
Oct 28~Dec 4, 2019	ClNO2	Shanghai	0.094±0.11 averaged with 1-min maximum of 5.7 (ppbV)	0.25±0.11 in daily-average, peaked at 1.5	Li et al. (2023)
Oct 28~Dec 4, 2019	Cl2	Shanghai	0.024±0.039 in average with a maximum of 1.1 (ppbV)	0.031±0.041 in daily-average, peaked at 0.4	Li et al. (2023)
Oct 30~Nov 18, 2020	ClNO2	Shanghai	1h diurnal averaged maximum of 0.16 and peaked at 0.4 (5min averaged) (ppbv)	0.19±0.06 in daily-average, peaked at 0.34	Lou et al. (2022)

Note: * For observations conducted outside the YRD, results were compared with simulation results at 40 monitoring sites in the YRD.

^**For simulation results, values were presented in the same unit as “daily average” ± “standard deviations” and peak values during the same time slices.

BrO (bromine oxide) and IO (iodine oxide) are two critical RHS. The seasonal variation of simulated BrO and IO in HAL case was generally consistent with the results of Br_y and I_y, which peaked at 0.9 and 7.1 pptV in summer, respectively. The spatial distribution of BrO and IO in different seasons were shown in Fig.S6. BrO and IO were lower in inland regions and higher at coastal areas, but they were significantly influenced by the interactions between ship-originated plumes and oceanic halogens, presenting a coincident pattern with ship tracks in SEIM. NO_X emitted from shipping activities could transform BrO and IO into halogen nitrates (XNO₃) which would eventually deposit onto the sea surface, serving as a sink for halogen oxide (XO) to decrease their levels. As BrO and IO both rapidly transform into XNO₃ or other species, the decreasing effect of ship emissions becomes obvious (Li et al., 2021b), thus exhibiting a trajectory in line with ship tracks.

For BrO and IO, no measurements have been made in the YRD hitherto, thus we could only compare the simulation results with observations outside the domain of this study. On the other hand, previous modeling studies should also reproduce the magnitude of simulated BrO and IO and might support the results from this study. The comparisons between CMAQ simulations and available observations and simulations of BrO and IO over the ocean were listed in Table 4. Saiz-Lopez and von Glasow (2012) compiled earlier observations of BrO and reported maximum range of 0.5 to 2.0 pptV for ground-based and 3.0 to 3.6 pptV for open sea measurements. For IO, the levels observed were between 0.2 to 2.4 pptV and 3.5 pptV for land-based and cruise measurements. Other studies have reported observed BrO and IO concentrations in MBL with an order of 1 pptV (Allan et al., 2000; Mahajan et al., 2009; Peters et al., 2005; Read et al., 2008), which were generally comparable to our simulation results.

Table 4.

The comparisons of simulated BrO and IO over the ocean with observations and simulations reported from previous studies

Species	Time period	Sources	Observation or simulation results	CMAQ simulation results*
BrO (pptV)	Jan-Feb, 2014	Flight campaign in west Pacific1	<0.5 to 1.7 during daytime below 500 m	0.11 in daily-average, peaked at 0.35
	4 months of 2016	WRF-CMAQ simulations in west Pacific2	1.0 to 2.04 in maximum from different seasons	0.1 to 1.0 in daily-maximum
	Aug, 2018	WRF-Chem simulations in west Pacific3	0.45 in daytime average and 1.4 in maximum	0.26 in daytime average, peaked at 0.97
	Jul, 2019	WRF-CMAQ simulations in the Philippine Sea4	0.25 in average and 1.2 in maximum	0.14 in daily-average, peaked at 0.93
IO (pptV)	Dec, 2010-Jul, 2011	Malaspina 2010 global circumnavigation5	0.4 to 1.4 in the west Pacific marine boundary layer	0.36 to 2.07 in daily-average
	Jan-Feb, 2015	Cruise campaign in the Indian Southern Ocean6	Peaked at 0.57±0.27 on 6.2°N	0.61 in daily-maximum
	Aug, 2018	WRF-Chem simulations in west Pacific3	1.2 in average and 2.6 in maximum at the surface	1.9 in daily-average, peaked at 6.9
	Jul, 2019	WRF-CMAQ simulations in the Philippine Sea4	1.4 in average and 2.5 in maximum	2.0 in daily-average, peaked at 7.0

Note: *CMAQ simulation results were averaged over the ocean from the same time slices according to observations.

¹ Koenig et al. (2017)

² Huang et al. (2021)

³ Li et al. (2021b)

⁴ Fan and Li (2022)

⁵ Prados-Roman et al. (2015)

⁶Mahajan et al. (2019).

CMAQ produced lower BrO levels probably due to the uncertainties in the speciation of halogen contents in sea spray aerosols and the parametrization of bromocarbon emissions. The slight underestimation of BrO from simulations was also reported in previous CMAQ halogen chemistry studies (Li et al., 2019; Sarwar et al., 2019). For IO simulations, IO levels were expected to be higher in nearshore areas due to higher bioactivity and higher outflow of O₃ pollution from the coast. It should be emphasized that the comparisons were indirect due to limited observations and we urged for more relevant RHS measurements to improve model validations. Nonetheless, the level and order of magnitude of RHS simulated in this study were comparable to previous modelling studies and observations, indicating that the CMAQ model was suitable to investigate the effect of halogens on air quality in the YRD region.

3.3. Impact of halogens on atmospheric oxidation

3.3.1. Impact of halogens on major atmospheric oxidants

The annual average impact of halogens on individual oxidants in the YRD region were shown in Fig.2, and relative changes of oxidants (OH, HO₂, O₃ and NO₃) derived from different emission sources of halogens (EMIS_CL and EMIS_OCN) were shown in Fig.3. Seasonal variations of changes in these oxidants were consecutively shown in Fig.S10~S13. Halogen emissions and chemistry induced a uniformly positive impact on OH radical. The highest increase of OH occurred in locations with larger anthropogenic chlorine emissions (Fig.S2) in the YRD region, which was more than 4% in annual average. Chlorine would influence OH production mainly through the reactions with VOCs to produce HO_X and changes in the partitioning of HO₂ and OH. Halogen-mediated changes in HO₂ were consistent with OH on the land with a higher increase up to 8% in provincial capitals in the YRD region. The seasonal variation of OH and HO₂ had similar patterns which peaked in summer in BASE run due to stronger solar radiation and photochemical reactions (Fig.S10 and S11). The influence on HO_X was the strongest in winter, since the mixing ratio of OH and HO₂ was the lowest in winter.

Fig.2.

Annual average of relative difference of simulated oxidants (%) between HAL and BASE cases

The increase of HO_X due to halogens on land was primarily influenced by anthropogenic chlorine (Fig.3a, b). Although simulated bromine and iodine species on the land were non-negligible, their impact on HO_X mixing ratio was overall neutral (less than ±1%), whereas they would exert a larger influence in the marine environment (Fig.3e, f). The impact of oceanic halogens (mainly bromine and iodine) on OH and HO₂ presented an opposite result with increased OH up to 12% and decreased HO₂ for 8%, indicating the role of halogens in changing HO_X partitioning. In the marine environment, the transformation of HO₂ to OH and decrease in the ratio of HO₂/OH were mainly driven by iodine and bromine chemistry (Saiz-Lopez and von Glasow, 2012) and the relative changes in our study were in accordance to previous modelling studies (Mahajan et al., 2021; Sarwar et al., 2015; Stone et al., 2018). The influence of oceanic halogens on HO_X along ship tracks was distinct. Relative changes in OH induced by oceanic halogens along ship tracks were weaker than that outside shipping trajectories (Fig.3e). The mixing ratio of HO₂ was lower along the ship tracks in BASE case and was significantly increased due to oceanic halogens in all seasons along ship plumes. Changes of HO_X along the ship plumes were complex in different pathways: (1) O₃ reduction due to halogens in marine environment (Fig.2c) and hence reduced production of OH by ozone photolysis; (2) VOC oxidation by halogens along ship tracks and increased production of HO₂; (3) competitions between ship-emitted NO_X and atmospheric HO_X.

Surface O₃ was moderately increased around 1% annually in the YRD region while in marine environment O₃ was decreased by 3% (Fig.2c). In coastal areas where the atmosphere was both influenced by anthropogenic chlorine and oceanic halogens, the changes of O₃ became neutral, indicating that the decreasing effect of bromine and iodine on O₃ is compensated for by the enhancement effect of chlorine. For seasonal variations (Fig.S12), O₃ decreased by 1% along the coast in summer, covering the entire of Shanghai due to a stronger horizontal transport of emitted bromine and iodine species (Fig.S9), and decreased more than 2% in the marine environment. On the contrary, O₃ was the lowest in winter and was mostly enhanced by more than 2% in the YRD region in HAL run. Chlorine emissions and chemistry were identified to aggravate O₃ pollution in areas where anthropogenic activities and pollutant emissions were heavy through oxidizing VOCs to form peroxy radicals and reactions of Cl radicals and NO₂ to form radical precursors ClNO₂, which would both lead to O₃ formation. Li et al. (2019) found a maximum O₃ increment of 0.22 ppbV due to chlorine chemistry over the land of Europe in summer. Wang et al. (2019) reported a neutral to negative effect of chlorine chemistry on O₃ in China, but they did not include anthropogenic chlorine emissions. For bromine and iodine impact, halogens would directly destroy O₃ and reduce the level of NO₂ that subsequently reduce the production of O₃ (Fig.3g). Huang et al. (2021) reported a depletion in O₃ by 6% over the coast line of China due to oceanic halogen emissions (bromine and iodine) and chemistry and the sharpest gradient of O₃ loss occurred near the coastline offshore.

NO₃ was generally reduced over the YRD region for up to 10% in annual average, and was further reduced in the marine environment (Fig.2d). The decrease in NO₃ was majorly resulted from oceanic halogen emissions, while along road networks in the YRD region, NO₃ mixing ratio would increase due to chlorine (Fig.3d, h). As for the seasonal variation (Fig.S13), NO₃ was significantly reduced in summer even for the inland YRD region due to strong photolysis and oceanic halogen emissions. In winter, NO₃ was generally increased owing to the combination effect of increasing O₃ and NO₂ (Fig.S12h and S14h), although its mixing ratio was relatively low. In marine environment, less decreased NO₃ was marked along the ship tracks in all seasons. NO_X emitted from ship plumes might compensate for the loss of NO₃ but still was overwhelmed by the decreasing effect of halogens. NO₃ was a nighttime radical primarily formed by the reaction of O₃ and NO₂, and was removed by rapid photolysis during daytime. NO₃ precursors would react with HO_X and RHS (OH+NO₂ and XO+NO₂ to form nitric acid and subsequently lost to aqueous phase) and form nitrate aerosol through heterogeneous reactions, therefore the increase of HO_X by halogens could result in a general decrease in NO₃. Similarly, Li et al. (2019) found a monthly average decrease of NO₃ over continental Europe by 1~20 pptV in summer and NO₃ was less reduced along ship trajectories due to overall halogen chemistry. Muñiz-Unamunzaga et al. (2018) reported a reduction of both NO₃ and HNO₃ by 20~50% and 6~10%, respectively, from marine halogens in Los Angeles during August to September of 2006.

3.3.2. Impact of halogens on atmospheric oxidation capacity

Atmospheric oxidation capacity was estimated using Equation 1 and shown in Fig.4. In the BASE case, AOC influenced by major oxidants (OH, O₃, and NO₃) presented a clear seasonal variation that peaked in summer with high levels of OH and O₃ while rapidly declined in winter, which was in line with the seasonal variations of oxidants. For halogen-influenced AOC which included Cl radicals, denoted as HAL-AOC, was consistently enhanced by halogens between 5~20% over the YRD region and also exhibited a similar seasonal pattern in accordance to the relative changes of atmospheric oxidants in the marine environment. The strongest enhancement of HAL-AOC occurred during winter by up to 48.4%.

Fig.4.

Seasonal average of atmospheric oxidation capacity (10⁶ molecules cm⁻¹ s⁻¹) in BASE case and relative changes (%) induced by halogens (HAL-AOC) between HAL and BASE cases

OH and Cl radicals are daytime oxidants while NO₃ is a nighttime oxidant. O₃ can exist both at night and during the day. AOC was controlled by different oxidants during a day and would exhibit a diurnal pattern. In this study, Shanghai was selected to examine the impact of overall halogen chemistry on AOC. Diurnal profile of annual average AOC was shown in Fig.S15. Daytime AOC was approximately 2.5 to 5 times higher than nighttime AOC and the dominating radical was OH in both BASE and HAL cases. AOC peaked rapidly before noon, indicating the influence of heavy traffic emissions during rush hours. Nighttime AOC was controlled by O₃ and NO₃ accumulated after sunset, while the former was photolyzed before noon and the latter was gradually extinguished before dawn. The increase of HAL-AOC was achieved via increasing OH and adding Cl radicals, while changes in O₃ and NO₃ induced by halogens were less significant (Fig.S15 and S16). Similarly, Xue et al. (2015) reported a rapidly increase of AOC in the early morning due to chlorine chemistry at a coastal site of Hong Kong.

Anthropogenic chlorine emissions from industrial processes and residential activities led to high loadings of reactive chlorine species in the YRD region. The increase of HAL-AOC (up to 46.2%) induced by chlorine emissions presented several noticeable hot spots over the YRD region that were consistent with the spatial distributions of heavy chlorine emissions (Fig.S17). The results of chlorine’s impact on AOC suggested a further investigation on anthropogenic halogen emissions as they would exert influences on VOCs and NO_X. In polluted areas, O₃ formation was controlled by its precursors, VOC or NO_X, and the impact of HAL-AOC would be closely related to O₃ formation regime. Li et al. (2020) found that AOC increased in VOC and mixed-limited regions while AOC declined in NO_X-limited regions over China. Chen et al. (2024) reported a higher AOC in summer and O₃ formation was controlled by both VOCs and NO_X at Ningde in Fujian Province in China.

Oceanic emissions of halogens would produce nonnegligible loadings of halogens (Br_y and I_y) in marine environment and along the coast of the YRD region. Oceanic halogens resulted in the increase of HAL-AOC through enhancing OH radical levels which peaked in spring by up to 23% over the ocean and more than 5% along the coast. Therefore, influence of ocean-emitted halogens cannot be ignored and their emissions should be included in air quality modelling studies in coastal areas of China.

3.4. Contributions of halogens influencing individual oxidants through process analysis

3.4.1. OH

In this study, the integrated reaction rate (IRR) provided by process analysis module within CMAQ was used to investigate the strengths of different pathways influencing individual oxidants by different halogen species. IRR can calculate production rates of chemical species from different reactions (i.e., production rate of O₃ from halogens) (Jeffries and Tonnesen, 1994). Chlorine, bromine and iodine were isolated by conducting sensitivity simulations. The annual average of absolute changes of production rate of OH (ΔP_OH) were shown in Fig.5a. In our study, pathways of OH production were distributed into 3 major sources: (1) from O₃ photolysis (reactions of O(¹D) + H₂O to produce OH, here denoted as ΔP_OH_O3); (2) from HO₂ conversion (HO₂ + NO₃, O₃, etc., ΔP_{OH_HO2}); (3) from hypohalous acid photolysis (for example HOI photolysis to produce I and OH radicals, ΔP_{OH_HAL}). In accordance to changes in the OH radical mixing ratio, halogen impacts on the production of OH throughout the domain remained positive. The increase of P_OH was achieved mainly through both HO₂ and hypohalous acid enhancement by chlorine on lands while hypohalous acid photolysis enhanced OH radical production over the ocean that compensated for the loss of HO₂ by marine halogens (Fig.5b~d). Changes of ΔP_OH from O₃ and other sources were less significant compared to the other pathways (with an annual average of −0.71 and 0.83 pptV, respectively), hence they were not included in the further discussion (Fig.S18a). The annual P_{OH_O3} was reduced by iodine up to 28 pptV (1.5 pptV in annual average and peaked in summer for 4.2 pptV), indicating a combination effect of enhanced O₃ deposition (Chang et al., 2004; Pound et al., 2020) and destruction by iodine chemistry, especially in the marine environment (Fig.S18b). The decrease of P_OH in O₃ photolysis pathway by iodine would result in a total reduction of ΔP_{OH_O3} from 2~8 pptV in marine environment in comparison of HAL and BASE cases.

Fig.5.

Annual average of absolute changes of (a) total Poh and (b-d) decomposed pathways of P_OH (pptV) between HAL and BASE cases

We further probed into the impact exerted on P_OH through pathways of HO₂ conversions and hypohalous acid photolysis by individual halogens. Since the most distinct season was summer with the highest oxidative radical mixing ratio and oceanic halogen emissions, we focused on the analysis of changes in OH production in summertime. Seasonal variations of monthly average of ΔP_{OH_HO2} and ΔP_{OH_HAL} by separate halogen species were shown in Fig.6 and 7. For chlorine chemistry, P_OH from HO₂ pathway was enhanced by up to 479 pptV in summer and 213 pptV in winter. The largest increase in P_{OH_HO2} by Cl chemistry occurred where anthropogenic emissions were higher, thus illustrating several hot spots in the city clusters such as Shanghai, Nanjing and Hefei, over the YRD region. Chlorine could oxidize VOCs emitted from anthropogenic sources including ship plumes and produce HO₂ and RO₂ radicals, therefore exerting a positive impact on P_OH from HO₂ conversion pathway. The production of OH from hypohalous acid photolysis was also intensified since the chlorine emission included HOCl in this study, but with a smaller magnitude. ΔP_{OH_HAL} by chlorine peaked in summer at 86 pptV and 17 pptV in winter, and the spatial distribution of ΔP_{OH_HAL} was also in line with the intensity of chlorine emission inventory. HOCl was produced from both primary emission and the reaction of HO₂ with ClO in the area and would ultimately photolyzed to produce OH and Cl radicals.

Fig.6.

Monthly average changes in P_OH from HO₂ conversion pathway (ΔP_{OH_HO2}) (pptV) induced by different halogens in summer and winter

Fig.7.

Monthly average changes in P_OH from hypohalous acid photolysis pathway (ΔP_{OH_HAL}) (pptV) induced by different halogens in summer and winter

The changes in P_OH by bromine and iodine from two pathways separately were consistent, but their combination effect on total ΔP_OH was the opposite especially in summertime marine environment (Fig.S19). Iodine would aggravate the production of OH more than 60 pptV over the ocean in summer while bromine exhibited a relatively weaker impact on ΔP_OH with a negative effect outside ship tracks. For HO₂ conversion pathway, changes in P_{OH_HO2} represented the indirect impact by halogens as we focused on the reactions of HO₂ with NO, O₃, and etc. Bromine and iodine could both inhibit direct HO₂ conversion to OH in marine environment by competition with NO and O₃ to consume HO₂ as illustrated in Fig.S11. The decrease in P_{OH_HO2} by iodine was higher than bromine (up to 134 and 10 pptV, respectively) and remained negative along the coast. ΔP_{OH_HO2} was strengthened by 10~50 pptV along the ship tracks, which was probably due to the increased ship-emitted NO_X and the formation of HO₂ through reactions of VOCs with oceanic halogens (Li et al., 2021b). The photolysis of hypohalous acid was another contributor to the increase of P_OH which was mainly resulted from iodine chemistry because of the emissions of oceanic HOI and I₂. The increase of ΔP_{OH_HAL} by iodine could peak at 277 pptV in the coastal area in summer and 3.8 pptV for bromine due to higher biogenic activities. Along ship tracks where NO_X emissions were elevated, the formation of hypohalous acid was suppressed by NO radicals through lowering HO₂ concentrations or serving as a sink for XO and subsequently would reduce P_OH in hypohalous acid photolysis pathway (Simpson et al., 2015).

In regard to the pathways of ΔP_OH influenced by individual halogens, the increase of P_OH was achieved through both strengthened HO₂ conversion and HOCl photolysis in the YRD region by chlorine chemistry. For bromine and iodine chemistry, XO would shift the HO_X partitioning to OH by consuming HO₂ to produce hypohalous acid and further photolyzed to produce OH radicals. The influence of bromine on P_OH was generally weaker than iodine probably because of the lower reactivity of BrO with HO₂ (Whalley et al., 2010). Therefore, iodine chemistry would uniformly increase P_OH in most of the areas where bromine could not. The decrease of P_OH in Shanghai indicated further investigations on the impact of atmospheric chemistry by iodine were needed especially in coastal areas (Fig.S19c). Inorganic iodine species (mostly HOI and I₂) were released to the atmosphere that were correlated with the amount of O₃ deposition to the ocean (Cuevas et al., 2018; Legrand et al., 2018). Oceanic iodine emissions were subjected to increase in the future as feedbacks of increasing anthropogenic O₃ pollutions and subsequently would exert further influences on O₃ and other pollutants in coastal areas.

3.4.2. O₃ and NO₃

Halogens could cause direct O₃ depletion through catalytic cycling or indirect effect on O₃ precursors through complex reactions which might cause opposite influences on the O₃ concentration, resulting in a rather neutral impact on O₃ by comprehensive halogen chemistry. IRR was used to examine the pathways of O₃ loss (L_O3) induced by direct halogen destruction and production (P_O3) influenced by halogens on its precursors. The pathways of O₃ concentration changes were shown in Fig.8. Halogen induced annual average direct O₃ destruction reached more than 600 pptV in ocean areas and more than 50 pptV along the coast, while halogen could have an obvious direct influence on O₃ over Shanghai. The enhanced O₃ loss rate was substantially contributed from marine iodine emissions for up to 90% of the O₃ destruction (Fig.S20c). Reactive iodine species were activated through decomposition of natural iodocarbons and ocean surface reaction of iodide ions and O₃ to produce inorganic precursors, HOI and I₂ (MacDonald et al., 2014). The rate coefficient of IO self-reaction was so high that it could produce OIO and I₂O₂ through the reaction of IO + IO, and OIO could be further photolyzed to produce I and O₂, increasing the efficiency of iodine cycling and subsequent O₃ depletion (Saiz-Lopez and von Glasow, 2012). Therefore, iodine species dominated the depletion rate of O₃. Although the accordance with anthropogenic chlorine emission hot spots indicated the role of chlorine to destroy O₃, reactions with hydrocarbons could predominate other than O₃ depleting pathway and the magnitude of O₃ loss rate by chlorine was incomparable to that of iodine (Simpson et al., 2015). The indirect influence on O₃ production rate by halogens was generally positive over the land and negative over the ocean, which could be attributed to anthropogenic chlorine and oceanic halogen sources, respectively (Fig.S21). The complex impact on O₃ precursors by halogens as well as cross halogen reactions required a further elaboration in the modelling study which could be improved in the future. For the unusual decrease of O₃ production rate in Shanghai (Fig.S21a), one possible reason was that Shanghai was a VOC-limited region where the consumption of VOC by chlorine could lead to a decrease in O₃, presenting a similar hot spot with the RO₂ production by chlorine (Fig.S22). Several observation-based modelling studies reported that Shanghai was usually a VOC-limited region during different time periods and was gradually changing to a transition regime in 2020 (Wang et al., 2017a; Wang et al., 2022; Zhang et al., 2020).

Fig.8.

Annual average changes in (a) O₃ destructions (pptV) induced by halogens and (b) indirect O₃ production rates induced by halogens between HAL and BASE cases

NO₃ is a nocturnal radical formed from the reaction of NO₂ and O₃. Annual average changes in NO₃ production rate (P_NO3) induced by halogens were shown in Fig.9. The increase of NO₃ production rate over the YRD region by halogens was somewhat similar to that of O₃ since the production of NO₃ was related to its precursors NO₂ and O₃. Chlorine would increase the production of NO₃ up to 19.3 pptV (more than 4 pptV in urban areas) by enhancing O₃ production and complex interactions with NO_X. On the contrary, the decrease in P_NO3 (2.3 pptV in average and −20 pptV in maximum) was achieved by both NO₂ and O₃ reduction due to iodine chemistry (Fig.9d and Fig.S23), and bromine exert a rather neutral effect on P_NO3 (less than 2 pptV in general) compared to that of chlorine and iodine. The effect of comprehensive halogen chemistry on P_NO3 was weaker than the sum of individual halogens, probably due to the cross halogen reactions since the reaction rate of these cross reactions were an order of magnitude faster than that of self-reactions (Simpson et al., 2015). Notably, the significant decrease of NO₃ production in Shanghai implied a further study on the chemical reactions between halogens and NO_X in similar coastal areas with air pollutions. Total loss rates of NO₃ were 4~7 times higher than the production rate. The changes of NO₃ in this study might be resulted from complex combination of different pathways: (1) O₃ and NO₂ were both influenced by halogens over the YRD region (Fig.S23); (2) halogens would react directly with NO₃ and reduce its level; (3) complex changes of NO_X induced by halogens, including NO/NO₂ partitioning, removing NO₂ into halogen nitrates, and forming radical precursors such as ClNO₂ (Simpson et al., 2015). Note that we included in-line DMS (dimethyl sulfide) emission parameterization for marine halogen emissions revised by Zhao et al. (2021) that might also serve as a sink for marine and coastal NO₃ (Andreae, 1985).

Fig.9.

Annual average changes in NO₃ production (pptV) induced by different halogen species

4. Conclusions

In this study, the WRF-CMAQ model with high-resolution and comprehensive halogen chemistry were conducted to simulate and investigate the impact of halogen emissions on atmospheric oxidants over the YRD region in China. Changes in major atmospheric oxidants (OH, HO₂, O₃ and NO₃) affected by halogens resulted in the increase of atmospheric oxidation capacity by 5.1% annually. AOC could reach more than 20% in several cities over the YRD region. Daytime OH radical was enhanced by 7.5% on diurnal average while nighttime NO₃ radical was reduced by 4.4% and changes in diurnal O₃ remained neutral. Contributions of OH toward AOC was approximately 2 times of the sum of O₃ and NO₃, and chlorine radical also exerted non-negligible influence (−5.7%) on aggravating daily AOC over the YRD region. Specifically, the contributions of anthropogenic chlorine emissions on the atmospheric oxidants were up to 16.4% for OH, 19.4% for HO₂, 2.2% for O₃ and 6.5% for NO₃ in populated cities over the YRD. Despite the general decrease of NO₃ radical due to halogen chemistry, anthropogenic chlorine emissions caused an annual increase of AOC by 3.0% and reached 46.2% in the YRD. Chlorine also contributed 58.8% to the increase of AOC in the research area.

IRR was applied to investigate pathways of oxidative radicals controlled by separate halogens. For OH radical, P_OH was majorly enhanced through the conversion of HO₂ and hypohalous acid photolysis pathways induced by halogens. Specifically, chlorine accounted for both pathways while bromine and iodine contributed to P_OH mainly through converting HO₂ into hypohalous acid. In regard to O₃, chlorine contributed most of the increase in O₃ production rate by influencing O₃ precursors. The pathway of direct O₃ destruction by halogens was also evaluated and iodine accounted for the most while chlorine and bromine presented a mild increase in destroying O₃. Changes in the production rate of NO₃ had a similar pattern to that of O₃ as chlorine increased its precursors while iodine chemistry influenced the opposite. The sharp decrease of NO₃ in Shanghai suggested further investigations into the impact of the reaction between DMS and NO₃.

We acknowledge the possible uncertainties lay in this modelling study. For the emission inventories, the uncertainties in anthropogenic chlorine emissions adopted in this study come from industrial processes and activity data (Yi et al., 2021), and more local emission measurements and observations for chlorine precursors are required. The parameterization of halocarbon emission rates using chl-α data from MODIS observations may not reproduce the variations in a specific region since the interpolation from coarse (9km) to fine (3km) could cause biases. Moreover, recently developed halogen emission inventories besides chlorine species, such as iodide inventories and anthropogenic bromine inventories based on emissions factors and machine-learning (Sherwen et al. (2019); Li et al. (2021c)) could be adopted in future studies. The results in this study indicates a nonnegligible influence on the atmosphere by iodine and bromine, though their emissions are less comparable to that of chlorine and may be underrated. For halogen chemistry scheme, incomplete reactions due to calculation efficiency may exert limitations on our simulations. For example, several heterogenous reactions of bromine species from Fernandez et al. (2014) are excluded due to modeling uncertainties. The chemical reactions of iodine oxide photolysis and heterogeneous uptake rates of halogen species on aerosols remain uncertain in different simulation and observations studies (Badia et al., 2019; Saiz-Lopez et al., 2014; Sarwar et al., 2019). Though simple parameterizations with constant uptake coefficients of halogens in this model report acceptable simulation results, the update of halogen chemistry is still needed to improve model performances and close the gap between simulations and observations, especially for iodine species. A previous observation study suggests a higher uptake coefficient of HOI and recycling rate of interhalogen product species, which results in higher ozone loss (Tham et al., 2021). Despite the noted uncertainties, this comprehensive process analysis with sensitivity simulations revealed the different roles of individual halogens in the increase of atmospheric oxidation capacity in the YRD region.

Oceanic halogen emissions in coastal areas would be affected in response to anthropogenic activities since the activation of RHS was closely related to NO_X emissions and O₃ levels in the area. Meanwhile, industrial processes and residential activities (i.e., crop straw burning) could also release halogens. As a mega-city and one of the largest ports in the world, air quality in Shanghai and its surrounding city-clusters in the YRD would be affected by halogens since elevated AOC inevitably influences the production of secondary air pollutants. We identified the potential effect of halogens on air quality and recommend considering of chemical interactions between anthropogenic emissions and RHS on formulating air quality policies and regulations. A recent modeling study revealed that shipping emissions might enhance the burden of atmospheric RHS and in turn affect air pollutants (Li et al., 2021b). We propose regulations on low-chlorine fuels and disinfectants used by vessels besides existing lower-sulfur fuel policies (Zhu and Wang, 2021). Shore power facilities during ship berthing incorporating clean energy are also recommended for reducing air pollution in ports to decrease the feedbacks on RHS activation by NO_X and O₃. Price subsidies by the government may improve the utilization rate of shore power, and the cultivations on environmental awareness for both consumers and producers are suggested. Halocarbons emitted from seawater are influenced by anthropogenic inputs (Qi et al., 2023), therefore potential regulations on colored or fluorescent dissolved organic matter in coastal water are needed. Finally, further field, laboratory, and theoretical studies are needed for future studies evaluating the impacts of halogen chemistry on air quality and for assessing air quality policy implications in the YRD region.