Impact of particulate nitrate photolysis on air quality over the Northern Hemisphere

Abstract

We use the Community Multiscale Air Quality (CMAQv5.4) model to examine the potential impact of particulate nitrate (${\text{pN}\mathrm{O}}_3^{-}$) photolysis on air quality over the Northern Hemisphere. We estimate the photolysis frequency of ${\text{pN}\mathrm{O}}_3^{-}$ by scaling the photolysis frequency of nitric acid (${\text{HNO}}_3$) with an enhancement factor that varies between 10 and 100 depending on ${\text{pN}\mathrm{O}}_3^{-}$ and sea-salt aerosol concentrations and then perform CMAQ simulations without and with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis to quantify the range of impacts on tropospheric composition. The photolysis of ${\text{pN}\mathrm{O}}_3^{-}$ produces gaseous nitrous acid (HONO) and nitrogen dioxide (${\text{NO}}_2$) over seawater thereby increasing atmospheric HONO and ${\text{NO}}_2$ mixing ratios. HONO subsequently undergoes photolysis, producing hydroxyl radicals (OH). The increase in ${\text{NO}}_2$ and OH alters atmospheric chemistry and enhances the atmospheric ozone (${\text{O}}_3$) mixing ratio over seawater, which is subsequently transported to downwind continental regions. Seasonal mean model ${\text{O}}_3$ vertical column densities without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are lower than the Ozone Monitoring Instrument (OMI) retrievals, while the column densities with the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis agree better with the OMI retrievals of tropospheric ${\text{O}}_3$ burden. We compare model ${\text{O}}_3$ mixing ratios with available surface observed data from the U.S., Japan, the Tropospheric Ozone Assessment Report – Phase II, and OpenAQ; and find that the model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis underestimates the observed data in winter and spring seasons and the model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis improves the comparison in both seasons, largely rectifying the pronounced underestimation in spring. Compared to measurements from the western U.S., model ${\text{O}}_3$ mixing ratios with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis agree better with observed data in all months due to the persistent underestimation of ${\text{O}}_3$ without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. Compared to the ozonesonde measurements, model ${\text{O}}_3$ mixing ratios with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis also agree better with observed data than the model ${\text{O}}_3$ without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis.

Graphical Abstract

1.0. Introduction

Atmospheric processing of nitrogen oxides (${\text{NO}}_{\text{x}}$) produces nitric acid (${\text{HNO}}_3$), which partitions into particulate nitrate (${\text{pN}\mathrm{O}}_3^{-}$). While the slow photolysis of ${\text{HNO}}_3$ can return some ${\text{NO}}_{\text{x}}$ to the atmosphere, ${\text{pN}\mathrm{O}}_3^{-}$ is thought to be the end reaction product, which can only be removed from the atmosphere via dry and wet deposition. Thus, ${\text{pN}\mathrm{O}}_3^{-}$ has traditionally been considered a permanent sink for ${\text{NO}}_{\text{x}}$. However, recent field and experimental studies challenge this traditional view by suggesting that ${\text{pN}\mathrm{O}}_3^{-}$ can undergo photolysis to produce nitrous acid (HONO) and nitrogen dioxide (${\text{NO}}_2$) with potentially important implications for air quality. Gen et al. (2022) and Cao et al. (2023) completed comprehensive reviews of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis in the atmosphere and reported the details of the mechanism and factors affecting the photolysis. ${\text{pN}\mathrm{O}}_3^{-}$ absorbs UV light under atmospherically relevant condition and transforms into an excited state which can then decompose to nitrite or react to form ${\text{NO}}_2$. Nitrite can then undergo further reaction producing HONO. Ye et al. (2016) measured gaseous and aerosol species during an aircraft campaign over the North Atlantic in July of 2013, and reported elevated HONO and ${\text{pN}\mathrm{O}}_3^{-}$ mixing ratios in the remote marine atmosphere. They found HONO and ${\text{pN}\mathrm{O}}_3^{-}$ were correlated and suggested that ${\text{pN}\mathrm{O}}_3^{-}$ photolysis produces HONO with a photolysis frequency ~300 times faster than that of ${\text{HNO}}_3$. They used measured chemical species and known atmospheric chemistry and predicted atmospheric HONO mixing ratios using a box model. Their predicted HONO mixing ratios were much lower than the observed HONO data over the clean marine environment. When they introduced the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$, the model successfully reproduced the observed HONO mixing ratios. To further support their suggestion, they collected ambient aerosol samples, performed laboratory chamber experiments, and detected HONO and ${\text{NO}}_2$ from the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$.

Reed et al. (2017) reported measurements of ${\text{NO}}_{\text{x}}$ at the Cape Verde Atmospheric Observatory (CVAO) (16° 51’ 49 N, 24° 52’ 02 W) and noted that ${\text{NO}}_{\text{x}}$ mixing ratios peak at around noon. Their box model without photolysis of ${\text{pN}\mathrm{O}}_3^{-}$ did not reproduce the observed diurnal pattern of ${\text{NO}}_{\text{x}}$. However, the addition of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis in the box model (with HONO and ${\text{NO}}_2$ yields as shown in Eq-1) helped reproduce the observed ${\text{NO}}_{\text{x}}$ diurnal cycle and was able to explain the observed HONO data at CVAO.

$${\text{pN}\mathrm{O}}_3^{-}\stackrel{hv}{\to }0.67\times \text{HONO}+0.33\times {\text{NO}}_2$$ (1)

Ye et al. (2017) collected aerosol samples near the ground surface from urban, suburban, and remote areas of New York in 2008–2010 and aloft using an aircraft in the southeast U.S. in the summer of 2013. HONO and ${\text{NO}}_2$ were detected in chamber experiments when the collected aerosol samples were exposed to ultraviolet light, but not when the light was turned off. Ye et al. (2007) concluded that photolysis of ${\text{pN}\mathrm{O}}_3^{-}$ produces HONO and ${\text{NO}}_2$ and reported a mean ${\text{pN}\mathrm{O}}_3^{-}$photolysis frequency greater than ~87 times that of ${\text{HNO}}_3$. The results also suggested that chemical composition, including organic matter and nitrate loading, can affect photolysis frequency. Additional experiments reported by Bao et al. (2018) show that aerosol samples from urban areas of Beijing, China that were exposed to a xenon lamp in a photochemical flow reactor also produced HONO and ${\text{NO}}_2$ only when the light was turned on and showed HONO formation to be correlated with ${\text{pN}\mathrm{O}}_3^{-}$. These authors reported that ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequencies 1–3 orders of magnitude higher than that of ${\text{HNO}}_3$ and suggested that the photochemical aging of ${\text{pN}\mathrm{O}}_3^{-}$ is an important atmospheric HONO production pathway in Beijing, China. Shi et al. (2021) performed laboratory photolysis experiments in a Teflon chamber using particulate sodium and ammonium nitrate. They also suggested that the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$ can release HONO and ${\text{NO}}_2$. They, however, reported a ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency of less than 10 times faster than that of ${\text{HNO}}_3$ and suggested that ${\text{pN}\mathrm{O}}_3^{-}$ photolysis plays a limited role in air quality.

Ye et al. (2018) further reported the measurements of HONO and other relevant chemical species using an aircraft in the southeastern U.S. during the summer of 2013. The daytime mean HONO mixing ratio was 11.2 pptv in the planetary boundary layer and 5.6 pptv in the free troposphere. Ye et al. (2018) calculated a daytime mean HONO source of 53 pptv hr⁻¹, suggesting that nearly 70% of HONO was produced from the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis, and reported a ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency of 2.0×10⁻⁴ s⁻¹ compared to the value of 7.0×10⁻⁷ s⁻¹ for ${\text{HNO}}_3$ photolysis at a solar zenith angle of 0°. Conversely, Romer et al. (2018) analyzed ${\text{NO}}_{\text{x}}$ and ${\text{HNO}}_3$ data over the Yellow Sea measured during the 2016 Korea−United States Air Quality (KORUS-AQ) Study and found that rapid ${\text{pN}\mathrm{O}}_3^{-}$ photolysis was not consistent with their measurements. These authors suggested that ${\text{pN}\mathrm{O}}_3^{-}$ photolysis can only occur at a frequency of 1–30 times faster than that of ${\text{HNO}}_3$.

Zhu et al. (2022) reported HONO and ${\text{NO}}_{\text{x}}$ measurements during the 2019 spring and summer at the Tudor Hill Marine Atmospheric Observatory in Bermuda. They reported higher mixing ratios of HONO and NOx in polluted plumes and suggested that ${\text{NO}}_{\text{x}}$ reactions were the dominant source for daytime formation of HONO. In the clean marine atmosphere with low ${\text{NO}}_{\text{x}}$, they reported a distinct HONO diurnal cycle with a peak occurring around noon and lower values at night. Zhu et al. (2022) suggested that ${\text{NO}}_{\text{x}}$ reactions contributed to ~21% of the daytime HONO source while the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis accounted for the remaining daytime HONO source with an enhancement factor of 29 for ${\text{pN}\mathrm{O}}_3^{-}$ photolysis compared to that of ${\text{HNO}}_3$. Anderson et al. (2023) reported HONO measurements at surface and aloft using an aircraft near CVAO and provided strong evidence that ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances daytime HONO, with a photolysis frequency that increases with relative humidity and decreases with ${\text{pN}\mathrm{O}}_3^{-}$ concentrations. The average enhancement factor for ${\text{pN}\mathrm{O}}_3^{-}$ photolysis compared to that of ${\text{HNO}}_3$ in this study was 70. Their results mostly reconcile the large discrepancies in ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequencies reported in previous field and laboratory studies.

Kasibhatla et al. (2018) implemented photolysis of coarse mode sea-salt ${\text{pN}\mathrm{O}}_3^{-}$ into the GEOS-Chem model with 4° × 5° horizontal grid-resolution and compared model predictions with measurements at the CVAO. They reported that sea-salt ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequencies of 25–50 times greater than that of ${\text{HNO}}_3$ reproduces the observed ${\text{NO}}_{\text{x}}$ data. HONO prediction with the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$ reproduced the diurnal pattern of observed HONO but under-predicted the daytime observed peak data. Their analysis suggests that the upper limit of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency can be ~100 times greater than that of ${\text{HNO}}_3$. They reported that the global tropospheric impacts of the (coarse mode only) sea-salt ${\text{pN}\mathrm{O}}_3^{-}$ photolysis on ${\text{NO}}_{\text{x}}$, ${\text{O}}_3$, and OH are relatively small (1–3%). They performed a sensitivity analysis by also including photolysis of accumulation-mode ${\text{pN}\mathrm{O}}_3^{-}$ and reported 1–2 ppbv (parts per billion by volume) ${\text{O}}_3$ increase over most regions and >6 ppbv increase over northern India and eastern China. Zhang et al. (2022) added six HONO sources in the WRF-Chem model, performed simulations over North China, and compared model predictions with measurements of HONO, ${\text{O}}_3$, and other chemical species. The six HONO sources included traffic emissions, soil emissions, indoor emissions, ${\text{NO}}_2$ heterogeneous reaction on aerosol, ${\text{NO}}_2$ heterogeneous reaction on ground, and ${\text{pN}\mathrm{O}}_3^{-}$ photolysis with a photolysis frequency of 30 times the ${\text{HNO}}_3$ photolysis frequency. The model without the additional HONO sources failed to capture the observed HONO data but the model with the additional sources reproduced observed HONO data and increased ${\text{O}}_3$ and fine particles. They completed additional sensitivity simulations using ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency of 1, 7, 30, and 120 times the ${\text{HNO}}_3$ photolysis frequency and analyzed the resulting model predictions. They reported that selecting a larger value of the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency may cover up other ground based unknown HONO sources but overestimate vertical sources of HONO as well as ${\text{NO}}_{\text{x}}$ and ${\text{O}}_3$, and emphasized the need for additional studies to determine the exact ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency in the atmosphere.

Shah et al. (2023) compared four atmospheric models including GEOS-Chem to observations from three aircraft campaigns including the Atmospheric Tomography Mission (ATom). They used the GEOS-Chem model with 4° × 5° horizontal grid-resolution to perform an annual simulation for 2015 and selected months in 2012, 2013, 2016, 2017, and 2018; and reported that GEOS-Chem underestimates nitric oxide (NO) observations over the Pacific and Atlantic oceans and suggested the existence of a missing source. They were able to account for the missing source by adding the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. The photolysis frequency of ${\text{pN}\mathrm{O}}_3^{-}$ was expressed using an enhancement factor (EF) which indicates how fast it occurs relative to that of ${\text{HNO}}_3$ and proposed the following equation for EF:

$$\mathrm{E}\mathrm{F}=100\times \max \left(\frac{\left[\mathrm{S}\mathrm{S}\mathrm{A}\right]}{\left[\mathrm{S}\mathrm{S}\mathrm{A}\right]+\left[\mathrm{p}\mathrm{N}{\mathrm{O}}_3^{-}\right]},0.1\right)$$ (2)

Where [SSA] is the molar concentration of sea-salt and [${\text{pN}\mathrm{O}}_3^{-}$] is the molar concentration of ${\text{pN}\mathrm{O}}_3^{-}$.

Previous studies (Ye et al., 2017; Ye et al., 2017; Bao et al., 2017; Anderson et al, 2023) suggest that the photolysis frequency of ${\text{pN}\mathrm{O}}_3^{-}$ is higher over the clean marine environment and lower over ${\text{pN}\mathrm{O}}_3^{-}$ rich atmospheres (Romer et al., 2018; Shi et al., 2022). Equation (2) produces a maximum of 100 and a minimum of 10 for EF. The addition of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis in the GEOS-Chem model increased ${\text{O}}_3$ mixing ratios by up to 5 ppbv and improved the low model bias compared to the ozonesonde measurements. It also increased the global tropospheric OH mixing ratios by 19%.

Dang et al. (2023) used the GEOS-Chem model with 4° × 5° horizontal grid-resolution over the globe and 0.5° × 0.625° grid-resolution over North America and performed model simulations for 2009 and 2017 to understand the sources and trends of background ${\text{NO}}_2$ over the contiguous U.S. The model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis underestimated ${\text{NO}}_2$ vertical column densities compared to the retrievals from the Ozone Monitoring Instrument (OMI). Including ${\text{pN}\mathrm{O}}_3^{-}$ photolysis, with the same enhancement factor of Shah et al. (2023), increased ${\text{NO}}_2$ column densities by 13% annually and improved comparisons with OMI data. Here, we examine the potential impact of modeled ${\text{pN}\mathrm{O}}_3^{-}$ photolysis on air quality using the Community Multiscale Air Quality (CMAQv5.4) model (www.epa.gov/cmaq), a widely used air quality model in regulatory analysis and air quality research in many countries (Appel et al., 2021; Kitayama et al., 2019; Mathur et al., 2022) that does not currently include ${\text{pN}\mathrm{O}}_3^{-}$ photolysis.

2.0. Methodology

2.1. Model description

CMAQ contains comprehensive treatments of atmospheric emissions, transport, gas-phase chemistry, aerosols, cloud processes, dry and wet deposition. We use the CMAQ model configured for the Northern Hemisphere with a horizontal resolution of 108 km and 44 vertical layers (Mathur et al., 2017). We use the Weather Research and Forecasting (WRFv4.3.3) (Skamarock et al., 2008) model for generating meteorological fields and the Meteorological-Chemistry Interface Processor (MCIPv5.3.3) for preparing model-ready meteorology needed for the CMAQv5.4 model. We compare WRF predicted precipitation, temperature, wind speed, and water vapor mixing ratio to available observed data in the U.S. and compute Mean Absolute Error (MAE), Root Mean Squat Error (RMSE), Mean Bias (MB) and Correlation Coefficient (r) for each month (Table S.1). Gilliam et al. (2021) previously compared WRF predictions for 12-km Contiguous U.S (CONUS) and 108-km Northern Hemisphere domains for 2002–2019 with observed data over the CONUS. Error, bias and correlation shown in Table S.1 for the 108-km Northern Hemisphere domain are consistent to those shown by Gillam et al. (2021) and represent reasonable model performance considering the coarser grid-resolution. The domain and configuration used in this study was identical to those used in the EPA’s Air QUAlity TimE Series (EQUATES) (https://www.epa.gov/cmaq/equates) (Foley et al., 2023) except the use of lightning assimilation used here.

Foley et al. (2023) describe the anthropogenic and fire emissions over the U.S., Canada, and Mexico which we use in this study. We use anthropogenic emissions over China from the Tsinghua University and anthropogenic emissions for all other regions from the Hemispheric Transport of Air Pollution (HTAPv2) (Janssens-Maenhout et al., 2015) with scaling factors derived from the Community Emissions Data System (Hoesly et al., 2018). Fire emissions are from the Fire Inventory from the National Center for Atmospheric Research (FINN) version 1.5 (Wiedinmyer et al., 2011) for outside North America, Biogenic Volatile Organic Carbon (VOC) and soil nitric oxide (NO) emissions are from version 2.1 of global emissions from the Copernicus Atmosphere Monitoring Service (CAMS) (Sindelarova et al., 2014), and lightning NO emissions are from the Global Emissions InitiAtive (GEIA) monthly climatology (Price et al., 1997).

For gas-phase chemistry, we use the Carbon Bond chemical mechanism version 5 (CB6r5) (Yarwood, et al., 2020) augmented with detailed chlorine (Sarwar et al., 2012), bromine and iodine chemistry (CB6r5m) (Sarwar et al., 2019). CMAQv5.4 contains detailed treatments of elemental carbon (EC), organic carbon (OC), non-carbon organic mass (NCOM), sulfate (${\mathrm{p}\mathrm{S}\mathrm{O}}_4^{2-}$), nitrate (${\mathrm{p}\mathrm{N}\mathrm{O}}_3^{-}$), ammonium (${\mathrm{p}\mathrm{N}\mathrm{H}}_4^{+}$), un-speciated particles (OTHER), trace elements, and secondary organic aerosols species (Binkowski and Roselle, 2003; Murphy et al., 2017). CMAQ contains sea-spray emissions from the ocean (Gantt et al., 2015) and wind-blown dust emissions. Sea-spray emissions in CMAQ are speciated into 7 aerosol species (Millero, 2005) while wind-blown dust emissions are speciated into 28 aerosol species using U.S. EPA’s SPECIATE database (Simon et al., 2010) including ${\text{pN}\mathrm{O}}_3^{-}$ with a mass fraction of 0.00002 (gm/gm). Sea-spray emissions are not speciated into ${\text{pN}\mathrm{O}}_3^{-}$. CMAQ uses ISORROPIA II for calculating partitioning of total nitrate between ${\text{HNO}}_3$ and ${\mathrm{p}\mathrm{N}\mathrm{O}}_3^{-}$ (Fountoukis and Nenes, 2007).

CB6r5m contains photolysis of 68 gaseous compounds including ${\text{HNO}}_3$ and uses absorption and quantum yield data from the International Union of Pure and Applied Chemistry (IUPAC) to calculate photolysis frequencies of ${\text{HNO}}_3$ (Atkinson et al., 2004). However, it does not include photolysis of ${\mathrm{p}\mathrm{N}\mathrm{O}}_3^{-}$. We add the fine mode ${\text{pN}\mathrm{O}}_3^{-}$ photolysis reaction (Eq-1) to CMAQv5.4 using the Shah et al. (2023) proposed expression (Eq-2) for calculating EF. We multiply the photolysis frequency of ${\text{HNO}}_3$, calculated at each timestep, by EF to estimate the photolysis frequency of ${\text{pN}\mathrm{O}}_3^{-}$. We do not use coarse mode ${\text{pN}\mathrm{O}}_3^{-}$ photolysis in this study. Unlike GEOS-Chem, CMAQ does not separately track sea-salt concentrations as a unique species. Instead, the sea-spray emissions are speciated into other aerosol species. For this study, we tracked sea-salt concentration by adding an additional tracer species for sea-salt in CMAQ.

We performed two annual simulations for 2018. Each simulation was initiated on October 1, 2017, and continued through December 31, 2018. The first 3 months were used as spin-up and the results covering the entirety of 2018 were analyzed and presented subsequently. Initial conditions for both simulations were obtained from a previous simulation (the EPA’s Air QUAlity TimE Series (EQUATES)) (https://www.epa.gov/cmaq/equates). Boundary conditions for both simulations are identical and discussed in Mathur et al. (2017). One simulation uses the CB6r5m without the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$ while the other simulation uses CB6r5m with the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$. Differences in the results between the two simulations are attributed to the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$. To better understand the importance of the chemistry in different emissions scenarios, we performed several additional model sensitivity simulations for January 2018. Each of these simulations was also conducted with a 3-month spin-up time (October 1-December 31, 2017). Additional details of these simulations are described in Section 4.0.

Mean model calculated photolysis frequencies of ${\text{HNO}}_3$ and ${\text{pN}\mathrm{O}}_3^{-}$ (mean over the entire seawater and land areas) are shown in Figure 1. Photolysis frequencies of ${\text{HNO}}_3$ over seawater near the surface are similar to those over land. However, photolysis frequencies of ${\text{HNO}}_3$ over seawater at higher altitudes are slightly greater than those over land. Photolysis frequencies of ${\text{pN}\mathrm{O}}_3^{-}$ over seawater and land are greater than those of ${\text{HNO}}_3$ and photolysis frequencies of ${\text{pN}\mathrm{O}}_3^{-}$ over seawater are consistently greater than those over land. Our annual mean enhancement factor at surface over land is ~80 and decreases with altitude reaching ~41 at an altitude of ~14 km. In contrast, our annual mean enhancement factor at surface over ocean is ~98 and also decreases with altitude reaching ~63 at the similar altitude. Values over land vary with locations - lower values occur over ${\text{pN}\mathrm{O}}_3^{-}$ rich areas (for example portion of China) but higher values occur over areas adjacent to oceans where sea salt can be transported easily. These enhancement factors, though higher than those reported in Zhang et al. (2021) and Zhang et al. (2022), are consistent with those used in several other recent studies (Shah et al. (2023), Dang et al. (2023), Fu et al. (2020)) and highlight existing uncertainties in ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequencies and the need for better quantification of its spatial and temporal variability in the troposphere.

Figure 1:

A comparison of CMAQ calculated photolysis frequencies of ${\text{HNO}}_3$ and ${\text{pN}\mathrm{O}}_3^{-}$ over seawater and land.

2.2. Description of observed data used in model evaluation

We use ${\text{pN}\mathrm{O}}_3^{-}$ measurements from the Interagency Monitoring of PROtected Visual Environments (IMPROVE) (http://vista.cira.colostate.edu/Improve), the Clean Air Status and Trends Network (CASTNET) (www.epa.gov/castnet/download-data), and the Chemical Speciation Network (CSN) (www.epa.gov/amtic/chemical-speciation-network-csn) for model evaluation. For evaluating model ${\text{NO}}_2$ and ${\text{O}}_3$ vertical column density, we use data from the OMI satellite retrievals (https://aura.gsfc.nasa.gov/omi.html). We use ozonesonde measurements from the World Ozone and Ultraviolet Radiation Data Centre (WOUDC) (https://woudc.org/home.php) and the National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratories (ESRL) (https://gml.noaa.gov/ozwv/ozsondes) for evaluating model predictions aloft.

For evaluating model surface ${\text{O}}_3$ predictions, we use measurements from the Air Quality System (AQS) (www.epa.gov/aqs), the NOAA ESRL (https://gml.noaa.gov/ozwv/surfoz), Acid Deposition Monitoring Network in East Asia (www.eanet.asia), the Tropospheric Ozone Assessment Report – Phase II (TOAR2) (https://igacproject.org/activities/TOAR/TOAR-II), and OpenAQ sites (https://openaq.org). We use HONO measurements at CVAO in 2015 and 2023. Reed et al. (2017) described the CVAO HONO measurements during November 25 - December 3, 2015. The 2023 HONO measurements are described here. HONO was measured at the CVAO during 7^th to 26^th February 2023 using a long path absorption photometer (LOPAP-03 QUMA Elektronik & Analytik GmbH). The LOPAP measurement technique is described in detail by Heland et al. (2001) and Kleffman et al. (2002). To summarize, HONO is collected into the liquid phase from the sampled air flow as a diazonium salt through reaction with reagent 1 (100 g sulfanilamide dissolved in 9 L deionized water and 1 L HCl) and is then mixed with reagent 2 (0.8 g n-(1-naphtyl)-ethylenediamine dihydrochloride dissolved in 8 L deionized water) to make an azo-dye. The absorption of the azo dye at 550 nm is measured with a spectrometer and is used to determine the mixing ratio of HONO. The LOPAP has two channels, the first channel measures HONO and interferences, the second channel measures only interferences, allowing an interference-free HONO mixing ratio to be determined. At the CVAO, the LOPAP inlet was placed on the roof of a container lab, with a sampling height of roughly 3 m above ground. The instrument was zeroed with ${\text{N}}_2$ (N5 purity) every 6 hours for 30 minutes and was calibrated using a Titrisol nitrite standard solution (1000 mg of ${{\text{NO}}_2}^{-}$ diluted to 0.01 mg L⁻¹). Three calibrations were performed during the campaign, on 9^th, 11^th and 21^st February. The detection limit was calculated to be 0.3 ppt (2σ), using the noise during zero measurements. The relative error of the instrument was estimated as 10% of the measured HONO value.

3.0. Results:

3.1. Impact on annual daytime mean ${\text{pN}\mathbf{O}}_3^{-}$, ${\text{NO}}_2$, HONO, and ${\text{O}}_3$ at different altitudes

Predicted concentrations of ${\text{pN}\mathrm{O}}_3^{-}$ and mixing ratios of ${\text{NO}}_2$, HONO, and ${\text{O}}_3$ without and with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are shown in Figure 2. The concentrations and mixing ratios at each model vertical layer are averaged separately over all land and ocean areas and are shown as a function of altitude. ${\text{pN}\mathrm{O}}_3^{-}$ concentrations over land with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are marginally lower than the concentrations without the photolysis at lower altitude (<3 km); however, the concentrations with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis decrease aloft compared to those without photolysis [Figure 2(a)]. In contrast, ${\text{pN}\mathrm{O}}_3^{-}$ concentrations over the ocean with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are consistently lower than the concentrations without photolysis due to greater ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequencies over ocean than land. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis produces HONO and ${\text{NO}}_2$ and increases their mixing ratios. HONO undergoes photolysis during the day producing OH. The additional OH reacts with ${\text{NO}}_2$ and produces ${\text{HNO}}_3$, which partitions to ${\text{pN}\mathrm{O}}_3^{-}$, compensating for ${\text{pN}\mathrm{O}}_3^{-}$ lost via photolysis. Predicted HONO mixing ratios over land without and with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are similar near the surface (<1 km), suggesting that the chemistry has only a minor impact near the surface over land [Figure 2(b)], where alternative sources of HONO such as ${\text{NO}}_2$ hydrolysis dominate (Finlayson-Pitts and Pitts, 2000). However, mixing ratios aloft with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are higher than without photolysis. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis consistently increases HONO over the ocean at all altitudes. Impacts on ${\text{NO}}_2$ mixing ratios over land and ocean are similar to those of HONO [Figure 2(c)], which is expected as ${\text{pN}\mathrm{O}}_3^{-}$ photolysis directly produces both HONO and ${\text{NO}}_2$ (R1). Enhancements over seawater are greater than those over land due to higher ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequencies over seawater. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis has a substantial impact on ${\text{O}}_3$ and increases mixing ratios over land and ocean at all altitudes [Figure 2(d)]. Enhanced ${\text{NO}}_2$ and OH from the photolysis participate in atmospheric chemistry, producing more ${\text{O}}_3$ over seawater, which is then transported to downwind continental regions resulting in the increased mixing ratios over land from the surface to ~14 km. Percent increases of ${\text{O}}_3$ due to this chemistry are also shown in the figure. Percent increases in ${\text{O}}_3$ over the ocean are greater than those over land near the surface (<~4 km) [Figure 2(d)], however the relative impacts aloft are similar.

Figure 2:

Impact of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis on (a) ${\text{pN}\mathrm{O}}_3^{-}$ concentrations, (b) HONO, (c) ${\text{NO}}_{\text{x}}$, and (d) ${\text{O}}_3$ mixing ratios at different altitudes.

Numerous studies have documented the reduction of ${\text{O}}_3$ by bromine and iodine chemistry over seawater (Saiz-Lopez et al., 2014; Fernandez et al., 2014; Sarwar et al., 2015; Sherwen et al., 2016; Sarwar et al., 2019; Kang et al., 2021). These studies suggest that bromine and iodine emitted from seawater destroy ${\text{O}}_3$ near the surface and aloft; however, the reduction is largest near the surface and the extent of the reduction decreases with altitude. Kang et al. (2021) reported that the bromine and iodine chemistry in CMAQ reduces ${\text{O}}_3$ by ~16% near the surface. In contrast, the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis increases ${\text{O}}_3$ by more than 16% not only over seawater but also over land near the surface as well as aloft. Thus, ${\text{pN}\mathrm{O}}_3^{-}$ photolysis more than compensates for the ${\text{O}}_3$ reduction due to the bromine and iodine chemistry.

3.2. Impact on annual daytime mean surface ${\text{pN}\mathbf{O}}_3^{-}$

Predicted annual mean daytime surface ${\text{pN}\mathrm{O}}_3^{-}$ without the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and the changes in ${\text{pN}\mathrm{O}}_3^{-}$ due to the photolysis are shown in Figures 3(a) and 3(b), respectively. Predicted surface ${\text{pN}\mathrm{O}}_3^{-}$ concentrations without the photolysis are higher over land and lower (generally <0.6 μg m⁻³) over the ocean. Higher concentrations over China, India, western Europe, and eastern U.S. are noticeable. The chemistry reduces ${\text{pN}\mathrm{O}}_3^{-}$ over many ocean areas by 0.1–0.4 μg m⁻³. In contrast, the impacts over land are smaller. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis produces ${\text{NO}}_2$ and HONO which in turn undergoes photolysis producing OH. Enhanced OH then reacts with ${\text{NO}}_2$ to produce ${\text{HNO}}_3$ which partitions to ${\text{pN}\mathrm{O}}_3^{-}$. Thus, the loss of ${\text{pNO}}_3$ due to photolysis is compensated for by the production of ${\text{pN}\mathrm{O}}_3^{-}$ from the ${\text{NO}}_2$ + OH reaction. We compare model predictions with ${\text{pN}\mathrm{O}}_3^{-}$ measurements from the IMPROVE, the CASTNET, and the CSN. We calculate monthly Mean Bias (MB) using data over the entire U.S. and also only over the western U.S. (sites in California, Oregon, Washington, Idaho, Nevada, New Mexico, Wyoming, Colorado, Utah, Arizona, and Montana). As the impacts of the chemistry on ${\text{pN}\mathrm{O}}_3^{-}$ are small over the U.S. (Figure 3(b)), the monthly MB without and with the chemistry do not change appreciably [Figures 3(c) and 3(d)]. We also calculate monthly MB of ${\text{HNO}}_3$ using model predictions and measurements from the CASTNET sites. The ${\text{pN}\mathrm{O}}_3^{-}$ photolysis does not affect the MB for the entire U.S. (Figure 3(e)) or for predictions at the sites in the western U.S. (Figure 3(f)).

Figure 3:

(a) Annual mean daytime surface ${\text{pN}\mathrm{O}}_3^{-}$ concentrations without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; (b) changes in ${\text{pN}\mathrm{O}}_3^{-}$ concentrations with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis compared to those without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; (c) monthly MB of ${\text{pN}\mathrm{O}}_3^{-}$ concentrations at the IMPROVE, CASTNET and CSN sites over the entire U.S.; and (d) monthly MB of ${\text{pN}\mathrm{O}}_3^{-}$ concentrations at the IMPROVE, CASTNET, and CSN sites over the western U.S. (e) monthly MB of ${\text{HNO}}_3$ mixing ratios at the CASTNET sites over the entire U.S.; and (f) monthly MB of ${\text{HNO}}_3$ mixing ratios at the CASTNET over the western U.S. For western U.S., all sites in California, Oregon, Washington, Idaho, Nevada, New Mexico, Wyoming, Colorado, Utah, Arizona, and Montana are used. Red color represents model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and blue color represents model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis.

Kelly et al. (2010) described sea-salt emissions in CMAQ and evaluated the model by comparing the model predictions to measurements from the Bay Regional Atmospheric Chemistry Experiment in Florida. Here, we show a comparison of model fine-mode sodium chloride (NaCl) concentrations to the measurements from the 13 coastal IMPROVE sites (Figure S.1). Model tends to underestimate the observed data in most months in part due to coarse model grid resolution that may not adequately capture the magnitude of surf-zone sea-salt emissions. The underestimation of sea-salt concentrations also implies that model does not overestimate the enhancement factor (equation 2).

3.3. Impact on annual daytime mean surface ${\text{NO}}_2$

Predicted annual daytime mean surface ${\text{NO}}_2$ without the photolysis and the changes in ${\text{NO}}_2$ due to the photolysis are shown in Figures 4(a) and 4(b), respectively. Predicted surface ${\text{NO}}_2$ mixing ratios without the photolysis are higher over land and lower (generally <0.2 ppbV) over the ocean. Higher mixing ratios over China, India, western Europe, and eastern U.S. reflect higher ${\text{NO}}_{\text{x}}$ (oxides of nitrogen) emissions over these areas. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis directly produces ${\text{NO}}_2$ and enhances the levels over many ocean and land areas. It also decreases ${\text{NO}}_2$ over some areas due to enhanced reaction with OH.

Figure 4:

(a) Annual mean surface ${\text{NO}}_2$ without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis, and (b) changes in ${\text{NO}}_2$ with ${\text{pNO}}_3$ photolysis compared to those without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis.

We compare model predicted ${\text{NO}}_2$ vertical column density with the OMI satellite retrievals (Figure 5). Seasonal MBs without the photolysis are generally negative over seawater as well as most continental land areas suggesting the model underpredicts ${\text{NO}}_2$ vertical column density compared to the satellite data. However, MBs with the photolysis are generally positive over much of the seawater and some land areas. The inclusion of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis reduces the magnitude of the negative MBs over many land areas suggesting improvement of model predictions compared to satellite data. Greater improvements are seen in spring, summer, and fall, when photolysis frequencies are higher and lower improvements are seen in winter when photolysis frequencies are lower. CMAQ with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances annual ${\text{NO}}_2$ vertical column density by 28% over the U.S. compared to the 13% enhancement reported by Dang et al. (2023) using the GEOS-Chem model. It is difficult to exactly isolate the factors for such differences. Possible reasons include the differences in the treatment of emissions, vertical layer structures, and other items between the two models. We also compared model predictions with surface ${\text{NO}}_2$ measurements from the AQS sites across the U.S. AQS sites are generally located near urban areas where higher mixing ratios are present. Model changes near the surface are small; consequently, there are no changes in MBs (not shown) since the impacts near the surface are small.

Figure 5:

Top panel: Seasonal MB of model ${\text{NO}}_2$ vertical column densities without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis (calculated using satellite retrievals from the OMI): (a) winter (DJF) (b) spring (MAM) (c) summer (JJA) (d) fall (SON). Bottom panel: Seasonal MB of model ${\text{NO}}_2$ vertical column densities with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis (calculated using satellite retrievals from the OMI): (e) winter (DJF) (f) spring (MAM) (g) summer (JJA) (h) fall (SON).

3.4. Impact on annual daytime mean surface HONO

Predicted annual daytime mean surface HONO without the photolysis and the changes in HONO mixing ratios due to the photolysis are shown in Figure 6(a) and 6(b), respectively. Predicted surface HONO mixing ratios without the photolysis are higher over land and lower (generally <1.0 pptV) over the ocean. CMAQ without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis includes HONO emissions, gas-phase reactions, and heterogeneous reactions (Sarwar et al., 2008). Heterogeneous reactions of ${\text{NO}}_2$ on ground and aerosol surfaces are important for HONO production (Finlayson-Pitts and Pitts, 2000; Aumont et al., 2003). Since ${\text{NO}}_2$ mixing ratios are higher over land, HONO mixing ratios are also generally higher over land. However, ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances HONO over large areas of ocean and also over some areas of land, where ${\text{NO}}_2$ hydrolysis is a negligible source of HONO due to low ${\text{NO}}_2$ levels. We compare CMAQ predicted monthly mean diurnal HONO mixing ratios with the 2015 and 2023 measurements at CVAO (Figure 6c). Both sets of measurements show higher daytime values than nighttime values. However, the 2023 daytime peak value is ~4 times greater than that in 2015. Andersen et al. (2023) reported measurements of HONO at CVAO in August 2019; their daytime peak is also larger than the 2015 data. Reasons for higher observed values in 2023 are not entirely clear. However, the measurements were completed in different years and months. It is plausible that HONO levels have increased during the measurement time period. We consider these two sets of data as the lower and upper bounds of measurements. We calculate monthly mean diurnal values using results from both model simulations and then also calculate annually aggregated diurnal mixing ratios using the monthly mean diurnal values (Figure 6c). Predicted monthly mean diurnal HONO mixing ratios without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are substantially lower than the observed data and do not follow the variation in observed data. In contrast, monthly mean and annually aggregated diurnal HONO mixing ratios with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis follow the variation in the observed data and most of the monthly mean and the annually aggregated diurnal mixing ratios fall within the upper and lower limits of the observed data. Zhang et al. (2022) recently reported HONO measurements at Beijing University, China in October 2018. Using the reported measured data, we calculate an average daytime value of 0.76 ppb (11–16 hours) and nighttime value of 1.95 ppbv (remaining hours). In urban areas, HONO mixing ratios tend to be much higher than those in remote areas as shown in Figure 6 and daytime observed HONO values are lower than corresponding nighttime values. In polluted urban atmosphere, ${\text{NO}}_{\text{x}}$ reactions tend to be the main source for HONO formation. Model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis produces an average nighttime value of 0.91 ppbv and a daytime value of only 0.03 ppbv. Model predictions are lower than observed data due to two reasons: (1) this study uses relatively a coarse horizontal grid which artificially dilutes emitted ${\text{NO}}_{\text{x}}$ over large grid volumes and subsequently produces lower HONO values (2) we previously implemented additional HONO reactions in CMAQ (Zhang et al., 2021) which improves the model comparison with observed HONO data in China. However, this study did not consider those additional reactions. As expected, the model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis has small impact on HONO production in urban areas and produces similar HONO values (nighttime value = 0.92 ppbv and daytime value = 0.04 ppbv). Thus, the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis does not appreciably alter the predicted HONO predictions in urban areas. Model predicts high levels of ${\text{pN}\mathrm{O}}_3^{-}$ over China, yields lower values of EF, and subsequently has low impacts on HONO. This reiterates that the chemistry is more important over remote areas where ${\text{pN}\mathrm{O}}_3^{-}$ can undergo photolysis at a higher frequency.

Figure 6:

(a) Annual daytime mean surface HONO without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; (b) changes in HONO mixing ratio with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis compared to those without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; and (c) a comparison of model predicted diurnal HONO mixing ratios with measurements at the Cape Verde Atmospheric Observatory (CVAO): solid black colors represent measurements – upper black curve represents measurements in 2023 and lower black curve represents measurements in 2015; red solid curve represents the annual aggregate of monthly mean diurnal values from model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; blue solid curve represents the annual aggregate of monthly mean diurnal values from model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; blue dashed curves represent monthly mean diurnal values from model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. Monthly mean diurnal mixing ratios from model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are consistently lower than those with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and are not shown for clarity of the plot.

3.5. Impact on annual daytime mean surface OH

Predicted annual daytime mean surface OH mixing ratios without the photolysis and the absolute and relative changes due to the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are shown in Figures 7(a)-7(c), respectively. OH mixing ratios of <0.15 pptV are predicted over the remote marine atmosphere and higher mixing ratios (>~0.2 pptV) are predicted over some land areas and along the shipping tracks where higher ${\text{NO}}_{\text{x}}$ mixing ratios are predicted. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances OH substantially (0.04–0.08 pptV and >50%) over large areas of ocean and some areas of land. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis directly enhances HONO which undergoes photolysis generating OH. Larger enhancements occur over marine environments than over land areas since more HONO and subsequently OH is produced over marine environments.

Figure 7:

(a) Annual mean daytime surface OH mixing ratio without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; (b) absolute changes in OH mixing ratio with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis compared to those without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; and (c) relative changes in OH mixing ratios with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis compared to those without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis.

${\text{pN}\mathrm{O}}_3^{-}$ photolysis in CMAQ enhances the maximum annual mean OH mixing ratio by a factor of ~2.0 compared to the Kasibhatla et al. (2018) reported largest impact of 1.6 over the marine boundary layer. CMAQ predicted impacts on OH are greater than the estimate of Kasibhatla et al. (2018) likely due to the fact that they used only coarse mode sea-salt ${\text{pN}\mathrm{O}}_3^{-}$ while we include all fine mode ${\text{pN}\mathrm{O}}_3^{-}$. Kasibhatla et al. (2018) performed an additional simulation by adding the photolysis of accumulation-mode ${\text{pN}\mathrm{O}}_3^{-}$ and reported that it further enhanced OH by 4.6% compared to that with the photolysis of coarse mode sea-salt ${\text{pN}\mathrm{O}}_3^{-}$. Shah et al. (2023) reported a global OH mixing ratio increase of 19% which agrees well with our estimated increase of 21% over the Northern Hemisphere. Observed OH data are not publicly available for the simulation period. Whalley et al. (2010) reported the measurements of diurnal OH levels at the CVAO during May 29 - June 2, 2007 using the Fluorescence Assay by Gas Expansion method. We qualitatively compare our model predictions with the reported measurements since the simulation and measurements are from different years. Measured mean peak occurred around noon and reached ~0.24 pptv. Model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis predicts a peak value of 0.34 pptv (average of May and June). Thus, model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis predicts 40% higher value than the measurement. Measurements were completed in 2007 while the simulation was conducted for 2018. Similar to the HONO measurements, it is likely that OH levels at this location have increased in recent years; consequently, model over-estimates the 2007 observed data. Model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances peak OH level to 0.44 pptv; thus, the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances OH by 30%.

3.6. Impact on annual and monthly daytime mean surface ${\text{O}}_3$

Predicted annual daytime mean surface ${\text{O}}_3$ without the photolysis and the changes due to the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are shown in Figures 8(a) and 8(b), respectively. Predicted surface ${\text{O}}_3$ mixing ratios without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis over land are greater than those over seawater. ${\text{O}}_3$ mixing ratios over seawater tend to be lower due to limited precursors and destruction by bromine (Fernandez et al., 2014) and iodine chemistry (Saiz-Lopez et al., 2014). Continental outflow regions have higher ${\text{O}}_3$ mixing ratios than over remote seawater. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances surface-level ${\text{O}}_3$ by 2–10 ppbv over large areas of the ocean as well as over land. The chemistry proceeds faster over seawater than over land, producing higher ${\text{O}}_3$ over seawater, which is subsequently transported over land. Areas of the Himalayas have larger impacts due to their higher elevation while areas of Africa have larger impacts due to wind-blown dust from the Sahara desert (see more in Section 4.0). Time series of monthly daytime mean surface ${\text{O}}_3$ mixing ratios over seawater and land without and with the chemistry are shown in Figure 8(c). Predicted ${\text{O}}_3$ mixing ratios with the photolysis are consistently greater than those without the photolysis over seawater as well as land. Percent ${\text{O}}_3$ enhancements due to ${\text{pN}\mathrm{O}}_3^{-}$ photolysis over seawater and land are also shown in Figure 8(c). Enhancements over seawater are consistently greater than those over land and the highest enhancement occurs in spring over both seawater and land.

Figure 8:

(a) Annual daytime mean surface ${\text{O}}_3$ mixing ratios without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; (b) changes in ${\text{O}}_3$ mixing ratios with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis compared to those without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; and (c) monthly mean surface ${\text{O}}_3$ mixing ratios without and with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis over land and seawater and ${\text{O}}_3$ enhancements by the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis over land and seawater.

Kasibhatla et al. (2018) reported ${\text{O}}_3$ enhancements of 1–5 ppbv over the marine boundary layer while Shah et al. (2023) reported a mean increase of 3.6 ppbv at the surface and a maximum increase of 8 ppbv. CMAQ predicts an increase in annual mean ${\text{O}}_3$ by 2–8 ppbv over the marine boundary layer, with a mean ${\text{O}}_3$ enhancement of 4.2 ppbv at the surface and a maximum increase of 14 ppbv. Thus, CMAQ predicted ${\text{O}}_3$ increases are larger than previous estimates reported by Kasibhatla et al. (2018) and Shah et al. (2023). The reasons for greater impacts in CMAQ are not clear. However, inherent model-to-model differences, and differences in horizontal grids and the representation and speciation of wind-blown dust likely contributed to larger enhancements.

We compare model predicted ${\text{O}}_3$ vertical column density with OMI satellite retrievals (Figures 9a and 9b). MBs without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are generally negative over most of the seawater as well as continental land areas suggesting the model underpredicts ${\text{O}}_3$ vertical column density compared to the satellite data. However, MBs with the photolysis are positive over large areas suggesting the improvement of model predictions. Larger enhancements occur in the spring, summer, and fall season when photolysis is more active.

Figure 9:

Top panel: Seasonal MB of model ${\text{O}}_3$ vertical column densities without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis (calculated using satellite retrievals from the OMI): (a) spring, (b) summer, (c) fall, (d) winter. Bottom panel: Seasonal MB of model ${\text{O}}_3$ vertical column densities with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis (calculated using satellite retrievals from the OMI): (a) spring, (b) summer, (c) fall, (d) winter.

We compare model predicted ${\text{O}}_3$ mixing ratios with observed data from AQS and CASTNET sites in the U.S. and calculate monthly MBs without (red) and with (blue) the photolysis by using data over the entire U.S. as well as the western U.S. (Figure 10a and 10b). For calculating MBs in the western U.S., we use data from all sites in California, Oregon, Washington, Idaho, Nevada, New Mexico, Wyoming, Colorado, Utah, Arizona, and Montana. Model performance without the photolysis is mixed – generally overpredicting observed ${\text{O}}_3$ in summer while underpredicting it in other seasons. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis eliminates the negative bias for January-June both at CASTNET and AQS sites, improving the model performance. However, it increases MBs for the remaining months at both sites, deteriorating the model performance. When CASTNET and AQS sites only over the western U.S. are considered, MBs without the photolysis are negative for all months (except December at AQS sites) indicating that the model predictions are lower than the observed data. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis eliminates the negative bias for all months (except January-March) at CASTNET sites and all months (except December) at AQS sites, leading to a substantial improvement of model performance. The existence of large springtime negative bias in CMAQ has been previously reported (Appel et al., 2021); ${\text{pN}\mathrm{O}}_3^{-}$ photolysis drastically rectifies the persistent negative bias in spring. Thus, the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis is particularly suitable for improving model ${\text{O}}_3$ concentrations during winter and spring, periods in which the model without the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis exhibits systematic underestimation. However, during summer and fall the additional ${\text{O}}_3$ produced from this pathway degrades model performance at sites where the model without the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis already overestimated the observed ${\text{O}}_3$.

Figure 10:

(a) monthly MB of daily 8-hr maximum ${\text{O}}_3$ mixing ratios at the CASTNET sites calculated using data over the entire U.S.; (b) monthly MB of daily 8-hr maximum ${\text{O}}_3$ mixing ratios at AQS sites calculated using data over the entire U.S.; (c) monthly MB of daily 8-hr maximum ${\text{O}}_3$ mixing ratios at CASTNET sites calculated using data over the western U.S.; and (d) monthly MB of daily 8-hr maximum ${\text{O}}_3$ mixing ratios at AQS sites calculated using data over the western U.S. For western U.S., all sites in California, Oregon, Washington, Idaho, Nevada, New Mexico, Wyoming, Colorado, Utah, Arizona, and Montana are used. Red color indicates Bias without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and blue color indicates Bias with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. Red color represents model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and blue color represents model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis.

MBs without and with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are also calculated using surface ${\text{O}}_3$ measurements from the NOAA ESRL sites (Figure S.2(a)). The model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis consistently produces negative bias while the model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis improves the MBs for all months. We also calculate monthly MBs using data from the TOAR2 (Figure S.2(b)) and over Japan (Figure S.2(c)). Consistent with findings over the U.S., the model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis improves the comparison in cooler months and deteriorates the comparison in warmer months.

We show seasonal mean observed data from the OpenAQ sites over the Northern Hemisphere (Figure S.3(a-d)). MBs without the photolysis are generally negative in spring (Figure S.3(e)). In contrast, MBs without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are mixed in summer and fall - positive over some areas and negative over other areas (Figure S.3(f-g)). The model performance is mixed in winter, over-predicting in some areas while under-predicting in other areas. The model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis consistently improves model performance in spring (Figure S.3(i)), and has mixed impacts in other seasons (Figure S.3(j-l)).

Comparison with ozonesonde measurements from the WOUDC and the NOAA ESRL are shown in Figure 11. Data are separated by latitude and altitude. Model predictions without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are consistently lower than observed data at all altitudes and latitudes, revealing a large under-prediction of ${\text{O}}_3$ throughout the troposphere. In contrast, the model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis drastically improves the comparison with observed data, not only near the surface layers but also aloft, indicating a large improvement in model performance. At latitudes of <45° N, the model without the additional chemistry underestimates ${\text{O}}_3$ in several months during June and September (Figure 11-b, d, f). However, model with the ${\text{pN}\mathrm{O}}_3^{-}$ photlysis tends to overestimate ${\text{O}}_3$ during this period likely due to the presence of high solar radiation which increases the photolysis frequency causing an overestimation of model ${\text{O}}_3$, suggesting possible seasonal variations in both the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency and the enhancement factor used in our parameterization. Nevertheless, the comparisons with ozonesonde measurements show that the inclusion of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis helps rectify the systematic underestimation in springtime ${\text{O}}_3$ throughout the troposphere.

Figure 11:

A comparison of model ${\text{O}}_3$ mixing ratios with ozonesonde measurements from the WOUDC and the NOAA ESRL: (a) plot using data from 5.7–9.5 km at latitude >45° N; (b) plot using data from 5.7–9.5 km at latitude <45° N; (c ) plot using data from 1.5–5.7 km at latitude >45° N; (d) plot using data from 1.5–5.7 km at latitude <45° N; (e) plot using data from 0–1.5 km at latitude >45° N; and (f) plot using data from 0–1.5 km at latitude <45° N. Black color represents observed data, red color represents model without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis, and blue color represents model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis.

4.0. Sensitivity study

4.1. Impact of wind-blown dust on ${\text{O}}_3$ without and with ${\text{pN}\mathbf{O}}_3^{-}$ photolysis

Wind-blown dust in CMAQ produces ${\text{pN}\mathrm{O}}_3^{-}$ (section 2.0) which can affect the impact of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis on ${\text{O}}_3$. To isolate the impact associated with wind-blown dust on ${\text{O}}_3$, we performed two additional sensitivity simulations for January 2018. One simulation was conducted without wind-blown dust and using ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. The other simulation was conducted without both wind-blown dust and ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. Impacts of wind-blown dust on ${\text{O}}_3$ were calculated as the differences of mean ${\text{O}}_3$ with and without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis (Figure 12(a)-(c)). Wind-blown dust without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis reduces ${\text{O}}_3$ over some areas of Asia by 1–2 ppbv since wind-blown dust attenuates sunlight, reducing photolysis frequencies and subsequently the ${\text{O}}_3$ formation rate. In contrast, wind-blown dust with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis has a mixed impact on ${\text{O}}_3$ [Figure 12(b)]. It reduces ${\text{O}}_3$ over parts of Asia and Africa but enhances ${\text{O}}_3$ over portions of Africa, South America, Asia, and the Atlantic Ocean. The inclusion of wind-blown dust with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis can affect the chemistry in several ways. In CMAQ, wind-blown dust is speciated into several aerosol components, including ${\text{pN}\mathrm{O}}_3^{-}$. The addition of wind-blown dust affects sunlight and reduces photolysis frequencies and ${\text{O}}_3$ formation rate. The additional ${\text{pN}\mathrm{O}}_3^{-}$ from wind-blown dust reduces EF (Eq. 2), and also reduces ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency and ${\text{O}}_3$ formation rate. However, the presence of additional ${\text{pN}\mathrm{O}}_3^{-}$ from wind-blown dust directly increases the availability of ${\text{pN}\mathrm{O}}_3^{-}$ and enhances the photolysis of ${\text{pN}\mathrm{O}}_3^{-}$ and ${\text{O}}_3$ formation rates. ${\text{O}}_3$ enhancement rates outweigh the ${\text{O}}_3$ reduction rates over some areas, causing an increase in ${\text{O}}_3$ mixing ratios. The changes in wind-blown dust initiated ${\text{O}}_3$ [difference of Figure 12(b) and 12(a)] are shown in Figure 12(c). The enhancements of ${\text{O}}_3$ over the Sahara desert and surrounding areas are evident. Thus, wind-blown dust can have appreciable impact on ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and affect ${\text{O}}_3$ mixing ratios in the vicinity of dust source areas. Using the GEOS-Chem model, Fairlie et al. (2010) reported that wind-blown dust can uptake ${\text{HNO}}_3$ and reduce surface ${\text{O}}_3$ mixing ratios over the North America by up to 1.0 ppbv. Koenig et al. (2021) recently performed aircraft measurements of iodine monoxide and other chemical species over the Atacama and Sechura Deserts and postulated that iodine released from dust can reduce ${\text{O}}_3$ over desert areas. However, CMAQ does not include ${\text{HNO}}_3$ uptake or release of iodine from dust; hence, the impact of such chemistry on ${\text{O}}_3$ is not accounted for in this study.

Figure 12:

(a) Impact of wind-blown dust on ${\text{O}}_3$ without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; (b) impact of wind-blown dust on ${\text{O}}_3$ with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; and (c) difference of the impact of wind-blown dust on ${\text{O}}_3$ with and without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis

4.2. Impact of ${\text{NO}}_{\text{x}}$ reduction on ${\text{O}}_3$ without and with ${\text{pN}\mathbf{O}}_3^{-}$ photolysis

To examine how ${\text{pN}\mathrm{O}}_3^{-}$ photolysis affects the relationship between ${\text{O}}_3$ abundance and ${\text{NO}}_{\text{x}}$ emissions, we performed two additional sensitivity simulations for January 2018. One simulation was conducted without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and using 75% of total ${\text{NO}}_{\text{x}}$ emissions (25% ${\text{NO}}_{\text{x}}$ reduction). The other simulation was conducted with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and using 75% of total ${\text{NO}}_{\text{x}}$ emissions. The impact of lower ${\text{NO}}_{\text{x}}$ emissions on ${\text{O}}_3$ was calculated as the difference of monthly mean ${\text{O}}_3$ with 75% and 100% ${\text{NO}}_{\text{x}}$ emissions. The impacts of lower ${\text{NO}}_{\text{x}}$ emissions on ${\text{O}}_3$ without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are shown in Figure S.4(a). Lower ${\text{NO}}_{\text{x}}$ emissions without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis reduces ${\text{O}}_3$ over most areas of the modeling domain except over a portion of China, suggesting the area is a ${\text{NO}}_{\text{x}}$ saturated region. The spatial impacts of lower ${\text{NO}}_{\text{x}}$ emissions with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are also similar [Figure S.4(b)]. The changes in ${\text{O}}_3$ due to lower ${\text{NO}}_{\text{x}}$ emissions [difference of Figures S.4(b) and S.4(a)] are shown in Figure S.4(c). The inclusion of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis enhances the impact of NOx emission reductions on lowering ${\text{O}}_3$ concentrations over portions of Africa, South America, India, and some oceanic areas by 0.3–3.0 ppbV ${\text{O}}_3$ for 25% ${\text{NO}}_{\text{x}}$ emission reduction. Thus, ${\text{NO}}_{\text{x}}$ plays an important role in affecting ${\text{O}}_3$ through ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. ${\text{NO}}_{\text{x}}$ control is commonly employed for mitigating ${\text{O}}_3$ pollution. This result suggests that neglecting nitrate photolysis can lead to an underestimation in representing the potential benefit of ${\text{NO}}_{\text{x}}$ reduction in lowering ${\text{O}}_3$. As seen in Figure S4(c), greater reductions in ${\text{O}}_3$ are realized over marine areas influenced by continental transport of ${\text{NO}}_{\text{x}}$ reservoir species highlighting the significance of this pathway in modulating large scale tropospheric ${\text{O}}_3$ distributions, especially as emissions across the globe change. The effects of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis on modulating ${\text{NO}}_{\text{x}}$ controls in coastal urban regions, however, cannot be robustly discerned using the coarse grid resolution employed by our current calculations and is an aspect that needs to be explored further with finer spatial resolution model calculations which can better resolve the interactions of this additional chemistry with that in urban plumes.

4.3. Impact of halogen chemistry on ${\text{O}}_3$ without and with ${\text{pN}\mathbf{O}}_3^{-}$ photolysis

To examine the impact of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis on ${\text{O}}_3$ due to halogen (bromine and iodine) chemistry, we performed two additional sensitivity simulations for January 2018. One simulation was conducted without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and using no halogen chemistry and the other simulation was conducted with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and using no halogen chemistry. The impact of halogen chemistry on ${\text{O}}_3$ was calculated as the difference of monthly mean ${\text{O}}_3$ with and without halogen chemistry, without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis. Consistent with results of Sarwar et al. (2015) and Sarwar et al. (2019), halogen chemistry without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis reduces ${\text{O}}_3$ not only over oceanic areas but also over land areas [Figure S.5(a)]. However, the reductions over ocean areas are greater than those over land areas. The halogen chemistry with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis also reduces ${\text{O}}_3$ over oceanic and land areas [Figure S.5(b)]. However, the spatial impacts over the oceans are greater than those obtained without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis; inclusion of photolysis further reduces ${\text{O}}_3$ over large oceanic areas by 0.5–3.0 ppbV [Figure S.5(c)] due to two reasons. ${\text{pN}\mathrm{O}}_3^{-}$ photolysis increases ${\text{O}}_3$ which in turn enhances the iodine atom + ${\text{O}}_3$ and the bromine atom + ${\text{O}}_3$ reaction rates causing an additional ${\text{O}}_3$ loss. CMAQ uses the MacDonald et al. (2014) parameterization for estimating inorganic iodine emissions due to the interaction of iodide in seawater and ${\text{O}}_3$. Enhanced ${\text{O}}_3$ due to ${\text{pN}\mathrm{O}}_3^{-}$ photolysis increases inorganic iodine emissions which enhance the iodine atom + ${\text{O}}_3$ reaction rates causing a further ${\text{O}}_3$ loss.

4.4. Impact of DMS chemistry modulated ${\text{O}}_3$ without and with ${\text{pN}\mathbf{O}}_3^{-}$ photolysis

To examine the impact of ${\text{pN}\mathrm{O}}_3^{-}$ photolysis on ${\text{O}}_3$ due to DMS (dimethyl sulfide) chemistry, we performed two additional sensitivity simulations for January 2018. One simulation was conducted without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and without any DMS chemistry (note that inclusion of DMS chemistry is default in the cb6r5m chemical mechanism). The other simulation was conducted with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and without any DMS chemistry. We calculated the DMS chemistry modulated impact on ${\text{O}}_3$ as the difference of monthly mean ${\text{O}}_3$ with and without DMS chemistry. As expected, the DMS chemistry modulated impact on ${\text{O}}_3$ without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis is small (< ±1.0 ppbV) [Figure S.6(a)]. However, the DMS chemistry modulated impacts on ${\text{O}}_3$ with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis are greater than those without ${\text{pN}\mathrm{O}}_3^{-}$ photolysis as the process reduces ${\text{O}}_3$ over large oceanic and land areas by 1.0–6.0 ppbV [Figure S.6(b)]. DMS chemistry not only enhances sulfate but also reduces ${\text{pN}\mathrm{O}}_3^{-}$ over oceanic areas (Zhao et al., 2021; Sarwar et al., 2023). The model with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and DMS chemistry reduces more ${\text{O}}_3$ since ${\text{pN}\mathrm{O}}_3^{-}$ concentrations are lower compared to the levels without DMS chemistry. The changes in DMS modulated ${\text{O}}_3$ [difference of Figures S.6(b) and S.6(a)] are shown in Figure S.6(c).

5.0. Summary

We implemented ${\text{pN}\mathrm{O}}_3^{-}$ photolysis into CMAQv5.4 by adopting the parametrization of Shah et al. (2023) for calculating the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequency of fine mode ${\text{pN}\mathrm{O}}_3^{-}$ and perform annual simulations without and with ${\text{pN}\mathrm{O}}_3^{-}$ photolysis and examine its impacts on air quality over the Northern Hemisphere. Model results suggest that the chemistry reduces ${\text{pN}\mathrm{O}}_3^{-}$ concentrations and enhances ${\text{NO}}_2$, HONO, and OH mixing ratios, mostly over seawater. Additionally, the incorporation of this pathway increases ${\text{O}}_3$ over seawater as well as land, with peak enhancements occurring in spring. Enhanced ${\text{O}}_3$ mixing ratios compare better with available surface measurements over the U.S., Japan, and TOAR2, and improve the model performance, in particular substantially improving the springtime low bias in CMAQ. The enhancement in modeled ${\text{O}}_3$ occurs not only at the surface but also aloft and results in improved comparisons with three-dimensional ${\text{O}}_3$ distributions represented with available ozonesonde observations. The improvements in aloft ${\text{O}}_3$ distributions result in better characterization of long-range transported ${\text{O}}_3$ in the troposphere (as evident by comparisons with ozonesonde data) and its subsequent modulation of surface ${\text{O}}_3$ in downwind continental regions as also demonstrated by improved comparison of spring-time surface ${\text{O}}_3$ predictions with ground-based measurements. We find that the ${\text{pN}\mathrm{O}}_3^{-}$ photolysis pathway has large impacts on model results, highlighting the need for additional field and experimental studies to better constrain ${\text{pN}\mathrm{O}}_3^{-}$ photolysis frequencies and reduce the uncertainty in its estimated impacts on modulating tropospheric composition and air quality across the globe.

Highlights.

• Particulate nitrate photolysis enhances nitrogen dioxide, nitrous acid, and ozone

• Particulate nitrate photolysis initiates ozone peaks in spring

• Particulate nitrate photolysis produces more ozone than lost due to halogen chemistry

• Dimethyl sulfide chemistry reduces the impact of particulate nitrate photolysis on ozone

Data availability:

CMAQ source code is publicly available from the following website: https://github.com/USEPA/CMAQ.