Fogs and Air Quality on the Southern California Coast

Fog acts as a reservoir and transport vector for chemicals in the atmosphere, altering the distribution of species between the gas and particle phases, and allowing deposition of nutrients and pollutants onto ecosystems and crops. Fog water and trace gas samples were collected from Casitas Pass along the Santa Barbara Channel in June 2015 to identify emissions sources and aqueous processes impacting Southern California air. Fog water composition was dominated by NH 4+ (volume weighted mean, VWM = 232 µM, range = 85–640 µM), with lesser contributions from NO 3– (126 µM, 30.4–778 µM) and SO 4 2– (28.3 µM, 12.1–90.0 µM), pushing the VWM pH to 5.92 (5.34–6.67). Organic carbon contributed substantially to fog composition (8.27 mg C L –1 , 4.70–16.8 mg C L –1 ). Carboxylic acids, products of aqueous oxidation, were abundant (20.1% of carbon mass on average), with > 1% contributions by acetate, formate, oxalate, malonate, succinate, and lactate. Sulfur-and nitrogen-containing organic species were detected, often after 3–5 hours of fog, suggesting aqueous formation. Sampled air was advected over the coastline near oil extraction operations, urban and agricultural areas; regional oil and natural gas processing and mobile sources were the most influential organic emissions at Casitas Pass. Fog composition in 2015 was contrasted with that from a study in July-September 1985/6. Concentrations of major fog constituents appear to have decreased in response to successful air quality regulations. While natural species concentrations in fog were similar (e.g., 2015 VWM [Na + ] = 101 µM, range = < 30–320 µM; 1985/6 [Na + ] = 129 µM, 12–1000 µM), anthropogenic species concentrations were lower in 2015 (e.g., 1985/6: [NO 3 – ] = 236 µM, 141–2800 µM vs. 2015: 126 µM, 30.4–778 µM). These results overall highlight changes in Southern California air quality issues, including improvement of some anthropogenic emissions and the current influence of organic emissions from industrial and mobile sources.


INTRODUCTION
Interaction of trace atmospheric compounds with fogs, clouds, and wet aerosols can substantially alter chemical fates and effects.Chemical transformations within the aqueous phase could decrease atmospheric lifetimes of organic species from those in the gas phase (e.g., ~2 times shorter lifetime for phenol; ~3.5 times shorter for acetic acid; † Now at Air Quality Research Center, University of California Davis, Davis, CA, USA ‡ Now at National Exposure Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC, USA Monod et al., 2005).Unfortunately, aqueous atmospheric organic processing is not well understood, despite its potential to strongly impact air quality; for example, oxidation reactions can alter partitioning of organic compounds into fog droplets from theoretical expectations (Herckes et al., 2002a, b;Ehrenhauser et al., 2012).
Fog is a key source of water in many regions of the world, including along parts of the west coast of the United States, where approximately 30% of water collected in redwood forests originates from fog (Dawson, 1998).The chemical composition of fog can therefore be influential in ecosystem, agricultural, and residential water resources: fog deposition can exceed annual dry deposition of abundant chemical species (Collett et al., 1989;Benedict et al., 2013).Fog composition varies with regional emissions sources, temperature, liquid water content, and sunlight (Kaul et al., 2011;van Pinxteren et al., 2016).Even on the well-studied Southern California coastline, where frequent fog events are observed due to lowering/advection of North Pacific High stratus layers (Johnstone and Dawson, 2010), mean seasonal concentrations of carboxylic acids and inorganic ions in fog vary widely as a result of these complex factors (Jacob, 1985;Munger, 1989;Munger et al., 1989aMunger et al., , b, 1990;;Collett et al., 1999;Herckes et al., 2002bHerckes et al., , 2007a)).
Land use and climatological changes can also give rise to spatial and temporal changes in fog water chemistry.Industrial processes such as oil and natural gas (O&NG) extraction and production, large-scale agriculture, and fossil fuel combustion are the main anthropogenic contributors to non-methane VOCs (Middleton et al., 1990), and can also emit sulfur oxides and nitrogen oxides (Finlayson-Pitts and Pitts, 2000).The state of California has increased in population dramatically between the 1980s and 2015: from 26 million people in 1985 to more than 39 million in 2015 (U.S. Census Bureau, 2015).Agricultural land use has also increased: the value of crops in Ventura County was $613 million in 1986, and had increased to $2.09 billion by 2013 (Ventura County Office of the Agricultural Commissioner, 2014).Despite these increases in population and agriculture, mean annual measured [NO 2 ] gas , [O 3 ] gas , and [PM 2.5 ] steadily decreased between 1990 and 2014 in Ventura (along the Santa Barbara Channel), in response to regulations on emissions.Similar trends were observed just northwest in Santa Barbara, although in both locations, [SO 2 ] gas remained fairly constant between 1990 and 2014 (U.S. Environmental Protection Agency, 2015).The Ventura Oil Field in the Ojai River Valley has been a thriving oil extraction location since the mid 1920s (Adamson, 2008), but production has decreased in recent years: in 1986, 7.4 million barrels were extracted; in 2014, 5.0 million barrels of oil were extracted (California Department of Conservation, 2015, 1987).Deposits of O&NG also exist offshore along the Santa Barbara Channel, which seep naturally from the ocean floor (Hornafius et al., 1999) and are extracted from offshore platforms.The emission rates of non-methane VOCs from natural offshore O&NG deposits were estimated to be double that from on-road vehicle traffic affecting nearby Santa Barbara County in 1999 (Hornafius et al., 1999).
Fog water and VOC samples were collected at Casitas Pass (CP), north of Ventura, California and inland of the Santa Barbara Channel in June 2015.The objectives of the field campaign were three-fold: (1) to build upon current understanding of aqueous atmospheric organic processing via fog chemical characterization; (2) to help elucidate sources of volatile/semi-volatile organic emissions on the Southern California coast; and (3) to examine changes in fog chemical composition over time by contrasting results between the June 2015 campaign and a study in 1985/6 (Jacob et al., 1985;Munger et al., 1989b).It was hypothesized that substantial changes between current and previous fog chemistry in the Santa Barbara Channel coastal region would be observed, due to increases in population density, changes in land use, and increased regulation of pollutant emissions.Organic markers of urban activities, O&NG extraction, and agricultural activities were expected to be present in the fog water, as well as possible aged particle-and gas-phase emissions.

MATERIALS AND METHODS
Fog water was collected between 8 and 14 June 2015.Four fog events will be discussed that offered sufficient liquid water for chemical analyses (~50 mL; events ending on the mornings of 11-14 June, and referred to as events one through four, respectively).As shown in Fig. 1, fog events one and three were captured from the summit of CP (34,3863°, -119.3809°, 350 m A.S.L.), event two from a location below the CP summit along Highway 150 near an avocado grove (34.3896°, -119.4158°,290 m A.S.L.), and event four from a third location along Highway 150 near an orange grove (34.3904°, -119.4197°,approx.250 m A.S.L.).All three sampling sites were on gravel pull-offs adjacent to Highway 150, 2-10 m from the roadway.Motor vehicle traffic was infrequent between 10:00 pm and 4:00 am (during most sampling periods), and increased somewhat through the final morning hours of each fog sampling period.
Meteorological and satellite data retrieved for the time periods of the four fog events (GOES 15 satellite and Oxnard Airport weather data; Figs.S1, S2 and S3) confirmed that widespread, patchy fog developed near the California coast at approximately 10 pm on the nights of 11-14 June 2015.Sufficiently dense fog was not observed on 10 June, coincident with a frontal passage as indicated by atmospheric pressure and wind observations.The lowest liquid water contents at CP, measured by collection of foggy air, were observed during event three (into the morning of 13 June) of our study while the greatest were observed during event four (into the morning of 14 June).Temperature inversion layers observed at the nearby Vandenburg Air Force Base (at 4 am; Supplementary Information) extended from the ground to ~1000 m during events two and four, but were elevated during events one and three (observed between ~400 and ~700 m during event one and ~700 and ~1400 m during event three).No apparent change in wind direction before each fog event was observed that could explain differences in fog chemical composition during the four evenings.
A 1985/6 study was carried out at a site along Highway 150, 50-150 m from the roadway, and only ~200 m from the event two sampling site (34.3865°,-119.4171°).Based on the similarities between fog composition at the sites used in the present study, it is unlikely that the micrometeorology or location of the sites used in 1985/6 and 2015 were substantial contributors to differences in fog compositions between the two studies.
Two Caltech Active Strand Cloudwater Collectors (CASCCs; Demoz et al., 1996) were deployed at CP to collect fog water (Fig. S4).Droplets were impacted onto rows of forward-tilted strands and pulled by aerodynamic drag imparted by an electric fan and by gravity into a sampling trough at the bottom of each collector.A stainless steel CASCC (ss-CASCC; Herckes et al., 2002a) with stainless steel strands and sampling trough was used to collect fog droplets for analysis of organic fog constituents.A smaller collector (a CASCC2) with Teflon collection surfaces was used to collect fog droplets for analysis of ionic fog constituents (Demoz et al., 1996).Fog water was collected for durations of one to three hours, depending on the Fig. 1.Maps of (a) fog sampling locations along CP Highway (between the Ocean to the west and Lake Casitas to the east), and (b) all fog water, VOC, and aerosol sampling sites.Red points denote cities/landmarks, dark blue points denote whole air grab sampling sites, and light blue markers represent fog sampling sites.Events one and three were collected at CP summit, event two at the avocado grove, and event four at the orange orchard.Labels are as follows: "Background" = upwind of main VOC sources at Gaviota State Park, "Oil spill" = along I-5 at oil spill, "Downwind" = road above Goleta downwind of oil spill, "Freeway" = site along I-5 near CP Highway, "Proc.Plant" = O&NG processing site, and "Serra Cross" = Serra Cross Park above Ventura.

(b) (a)
time required to collect sufficient volume (~50 mL) for fog chemical analyses.Samples of fog water were immediately massed on-site, and pH was also measured on-site using a Cole-Parmer microelectrode and pH meter calibrated with pH 4 and 7 buffers.
Preservation methods of aliquots for chemical analyses have been described previously (Boris et al., 2016).Samples for inorganic ions and total organic carbon (TOC) were aliquoted without added reagents.Chemical analyses were carried out using procedures applied previously (Collett et al., 1999;Benedict et al., 2012;Boris et al., 2016), but will be described here in brief.Formaldehyde was quantified using chemical derivatization and fluorescence spectrometry (Dong and Dasgupta, 1987), TOC was measured using a Sievers Model 800 Turbo TOC Analyzer in Turbo mode, and major inorganic ions as well as carboxylic acids (see species listed in Fig. 2) were quantified using ion chromatography (IC; Dionex).Additional organic molecules (see examples in Fig. 3) were identified and/or quantified via high performance liquid chromatography followed by negative electrospray ionization high-resolution time-offlight mass spectrometry (HPLC-(-)-ESI-HR-ToF-MS; Agilent 1100 series LC with Agilent LC/MSD-ToF).
Carbohydrates (see species in Table 1) were analyzed using high performance anion exchange chromatography with pulsed amperometric detection (HPAE-PAD) as described previously (Sullivan et al., 2008).Fog samples remaining after analyses were composited (Table S1 in the Supplementary Material) for analysis of volatile trace organic compounds via gas chromatography/mass spectrometry (HP Model 6890 with 5973 mass spectrometer and 30 m × 250 µm × 0.25 µm i.d.HP-5 capillary column).Additional information about fog water analyses can be found in the Supplementary Material.
Discrete whole-air VOC grab samples of ambient air (containing gases and particles/fog, but analyzed for trace gases) were collected using evacuated electropolished stainless steel canisters (2 L, 6 mm inlet diameter) during fog events and near areas of possible hydrocarbon emissions.A background measurement was made upwind of the Refugio State Beach site at Gaviota State Park and canisters were additionally collected near possible VOC sources (see below and Supplementary Material).Two canister batch blanks were generated by transporting cleaned, evacuated canisters to the sampling site without collection.Individual VOCs (48) were quantified using a three gas chromatography Fig. 2. Organic constituent concentrations of 2015 CP fog water.Box plot shows concentrations (µM) with box outer bounds at 25th and 75th percentiles, center line at the median value, and whiskers extending to outliers.Inset pie chart shows percent TOC contributed by specific fog water constituents (colors are the same in the box plot and pie plot).On average, carboxylic acids account for 20.1% of TOC.Percentages were calculated using carbon mass concentrations for all organic fog water constituents (in ppmC = mg C L -1 ).For all species, n = 20.The category "other orgs."includes carbohydrates, formaldeyde, and nitrophenols; "other acids" includes methanesulfonate, pyruvate, propionate, maleate, valerate, azelate, adipate, pinonate, and pinate; "carbohydrates" includes galactosan, glucose, mannose, mannitol, arabinose, xylose, levoglucosan, and mannosan; "nitrophenols" includes 4-nitrophenol, 2-methyl-4-nitrophenol, and 2,4-dinitrophenol.

Percent TOC (as ppmC)
Fig. 3. Polar, organic compounds identified via HPLC-(-)-ESI-HR-ToF-MS and IC in event three (other events were similar and are plotted in Fig. S4).Molecular composition within the fog is illustrated via the oxygen/carbon atom ratio and molecular mass (Da).Small, red points represent the first sample in the fog event while large, purple points represent the final sample collected.Patterns in identified molecular structures are demonstrated by the arrows and inset.Formulae of species identified only in later fog samples are labeled.
analytical system with three flame ionization detectors, one electron capture detector and one mass spectrometer as described previously (Sive, 1998;Russo et al., 2010;Zhou et al., 2010) and in the Supplementary Material.Limits of detection (LODs) were calculated as LOD = ts√((N s + N blk )/(N s × N blk )) where N blk = number field blanks, N s = number samples, t = Student's t-statistic value for N blk at the 95% confidence level (two-tailed), and s = standard deviation of blanks.Uncertainties were calculated as CI (95%) = ts/√N s for replicate analyses of calibration standard solutions; N s = number replicate standards (≥ 3), t = Student's t-statistic value for N s at the 95% confidence level (twotailed), and s = standard deviation of replicate standards.Air equivalent concentrations (also called the cloud water loading, and denoted by [X] air eq. ) were calculated for fog water constituents assuming 85% collection efficiency of fog liquid water (Demoz et al., 1996), using the calculated flow rate through the cloudwater collector, duration of each sample, and mass of each sample collected (see Supporting Information).The [X] air eq.values account for dilution/concentration associated with changes in fog liquid water content.The volume weighted mean (VWM) concentration for each chemical species quantified was calculated by multiplying the mass of each sample by the corresponding chemical concentration, summing all values, and dividing by the total summed mass of the samples.
Sea salt (ss) contributions to fog water constituents were estimated using ratios of ionic concentrations in seawater (Cheng et al., 2000).The measured molar ratio of Cl -/Na + Fog, Kanpur, India Kaul et al., 2011 ranged between 0.65 and 3.0, with a mean value of 1.1, while the expected ratio in seawater is 1.16 mole Cl -/mole Na + (Eriksson, 1960;Mouri and Okada, 1993;Cheng et al., 2000).Molar Cl -/Na + ratios less than 1.16 may be the result of acid displacement to form volatile HCl, such as the reaction of HNO 3(g) or carboxylic acids with NaCl (Ault et al., 2013;Wang et al., 2015).Soil contribution of Na + was minimal, as inferred from trace metals analysis in the fog samples (Napolitano et al., In Preparation).Samples with higher ratios may have been affected by HCl scavenged to the nucleating particles or fog (Zhang and Iwasaka, 2001).Sea salt contributions of other ions, using Na + as a conserved tracer, were as follows: 21% of K + , 58% of Mg 2+ , 11% of Ca 2+ , and 22% of SO 4 2-.
Low molecular mass carboxylic acids are atmospheric oxidation products of organic molecules from a variety of sources, including anthropogenic, biogenic or marine processes (Ervens et al., 2011;Lim et al., 2013).There are substantial primary sources of carboxylic acids as well, including fossil fuel and biomass combustion (Chebbi and Carlier, 1996;Kawamura and Bikkina, 2016).A large number of carboxylic acids contributing to the percOA, as in the CP fog water, suggests the fog samples contain oxidized organic matter (acetate, oxalate, formate, malonate, succinate, lactate, and glutarate each contributed > 1% of TOC).
Organic material from marine sources was a likely precursor to some of the carboxylic acids observed in the fog at CP. Methanesulfonate is a chemical marker of marine emissions, formed in the atmosphere exclusively as an oxidation product of dimethylsulfide via either  OH or  NO 3 reaction (Seinfeld and Pandis, 2006).The air equivalent concentration of methanesulfonate ([methanesulfonate] air eq. ) was correlated (r 2 ≥ 0.87; p < 0.06) with that of sulfate, [SO 4 2-] air eq , during events one, two and four (Fig. S8).
Oxalate has also been suggested as a product of aqueous oxidation from marine carbon emissions (Warneck, 2003;Rinaldi et al., 2011), and [oxalate] air eq. was correlated with [SO 4 2-] air eq. in the same events (r 2 ≥ 0.83; p < 0.08), suggesting in-cloud aqueous oxidation of marine emissions as a possible source of organic material (see Supplementary Material).
Aged biogenic and aged anthropogenic organic species were identified in the 2015 CP fog water, including nitrophenols, long-chain (C 4 -C 10 ) saturated and unsaturated carboxylic acids, and α-pinene and isoprene oxidation products (Table S1; tentative identifications made based on formulae).Patterns in identified molecular formulae are shown for fog water collected during event three, which contained the greatest concentrations of organic species (Fig. 3; all events are plotted in Fig. S9).Families of similar species, especially carboxylic acids, were present in the samples, differing by CH 2 , O, and CH 2 O, indicating that there were multiple contributing oxidation pathways of precursor VOCs leading to the complex mixture of aged organics found in these fog samples.The molecular difference of CH 2 O has been reported previously in products of high humidity oxidation of biogenic secondary organic aerosol (Nguyen et al., 2011), and observed within products of aqueous oxidation (Boris, 2016).It is currently unclear whether this difference results from a loss or a gain of CH 2 O; i.e., whether these products are overall a result of the process of fragmentation, functionalization, or oligomerization.During the later hours of each fog event (after 2:00 am, or after the typical maximum liquid water content), differing high molecular mass, oxygenated species were apparently formed, including organic sulfur (CHOS), organic nitrogen (CHNO), and organic nitrogen and sulfur-containing compounds (CHNOS).The CHO species C 12 H 22 O 6 , C 10 H 20 O 4 , and C 6 H 8 O 5 were also found in only mature fog water.These species may have formed through aqueous oxidation, as suggested in previous studies (e.g., CHOS via radicalcatalyzed esterification of -OH groups, Surratt et al., 2008, and CHNO via dark reaction with HONO and H 2 O 2 or via  NO 2 reaction, Vione et al., 2005).The formation of these species could alternatively reflect the lower (and varying) vapor pressures of organic compounds at the lower temperatures observed during late hours of the night, or evaporation of water to leave greater organic species aqueous concentrations in the fog.Only minimal wind during fog events was observed, making advection of new air masses containing differing organic species an unlikely explanation.
Research to demonstrate formation mechanisms of the CHOS fraction of organic aerosol is ongoing (McNeill, 2015).Fog samples at CP contained multiple CHOS and CHNOS species derived from biogenic and anthropogenic molecules, some of which have been observed in fog and aerosol samples previously (Boris et al., 2016) CHNO compounds such as nitrophenols have been recently cited as an important category of chemical species for absorption of sunlight (often cited within the class of "brown carbon" compounds), formed most prominently via aqueous reactions in the atmosphere (Zarzana et al., 2012;Desyaterik et al., 2013;Mohr et al., 2013).These species are of particular atmospheric importance because of their toxicity to animals (Agency for Toxic Substances and Disease Registry, 1992Registry, , 2011) ) as well as plants (Natangelo et al., 1999) Richartz et al., 1990).These nitrophenol species were, however, consistently present above the LOD in the fog samples from CP.There are a variety of possible atmospheric sources of nitrophenols.Pesticide atmospheric degradation is a source of nitrophenols, including formation of 2-methyl-4-nitrophenol from 4chloro-2-methylphenol (Chiron et al., 2009) or formation of various oxidation products of 2-methyl-4,6-dinitrophenol (also called dinitro-ortho-cresol or DNOC), a pesticide that has been observed in atmospheric samples (Asman et al., 2005;Delhomme et al., 2010).Gasoline combustion in vehicle engines is another known source of nitrophenols, including methyl-nitrophenols and di-nitrophenols (Tremp et al., 1993).Primary and secondary biomass burning emissions are also a source of nitrophenols (Harrison et al., 2005), but other signatures of biomass burning were not found at CP in June 2015: [levoglucosan] aq was low (Weber et al., 2007;Sullivan et al., 2008), the correlation between [total nitrophenol] air eq.and [nss-K + ] air eq. was weak (r 2 = -0.005;p = 0.97), and [K + ] aq was similar to that observed in Southern California fog previously (Jacob et al., 1985).Nitrophenols can also be transported to remote areas and/or formed in the atmosphere downwind of pollutant source regions; nitrophenols measured at CP, therefore, plausibly could have been derived from pesticide emissions, motor vehicle exhaust, or secondary reactions with aromatic VOCs such as pesticides, benzene or toluene (Lüttke et al., 1997;Harrison et al., 2005; Fig. S10).

Regional Anthropogenic Emissions of Volatile Organic Compounds
The concentrations of non-polar/slightly polar volatile organic compounds were determined (via gas chromatography/mass spectrometry) within fog samples that had been composited to represent one measurement per event.Polycyclic aromatic hydrocarbons (PAHs), alkanes, and other species such as hopanes were below detectable limits, as were biogenic VOCs previously detected in Southern California (Arey et al., 1995).This was surprising given the location of the sampling sites within densely vegetated hills, and near urban areas.A series of n-alkanoic acids and the pesticide N,N-diethyl-meta-toluamide (DEET) were the only low-polarity VOCs identified in the fog water samples.DEET was identified by mass spectral library match and confirmed using a chemical standard.The origin of the DEET measured in the CP fog water remains unknown; for additional discussion, see the Supplementary Material.
Whole air grab samples were collected near VOC sources potentially impacting CP and at the sampling locations during fog events (Fig. 1).Concentrations of most VOCs were greatest near the site of the Refugio oil spill along I-5.A distinct odor of oil was present at the spill site despite the oil spill having been contained and partially cleaned up by the beginning of the fog sampling period (National Oceanographic and Atmospheric Administration, 2015).A study of the Deepwater Horizon Oil Spill impact on atmospheric chemistry (De Gouw et al., 2011) estimated that emissions of ≤ C 18 oil constituents such as those measured in the present study are theoretically volatilized to the atmosphere within the first day after surfacing of spilled oil.Thus, although gaseous hydrocarbons measured at the oil spill cleanup site included high concentrations of chemical markers for raw O&NG, it is likely that emissions had already subsided, and that the oil spill had little impact on regional air quality during the fog sampling campaign.Measured [VOC]s from grab samples were overall similar between fog samples and events.The enhancement ratio of [iso-pentane] gas to [n-pentane] gas (iC 5 /nC 5 ) ratio can be used to distinguish raw O&NG/extraction emissions from those of vehicles/O&NG processing (Gilman et al., 2013).The iC 5 /nC 5 ratio measured at CP during fog sampling periods was between the ratios indicative of raw O&NG/ extraction and vehicles/O&NG processing (Fig. 4).
No correlation was observed, however, between the concentration of the O&NG extraction tracer propane (Gilman et al., 2013) and the iC 5 /nC 5 ratio (r 2 = 0.13; p < 0.49) among fog period grab samples, consistent with the pentane ratio indication that O&NG extraction emissions did not dominate alkane concentrations during fog collection.A regional VOC signal of O&NG processing (rather than raw O&NG/extraction) was identified during fog sampling periods based on similarities between ratios of measured [VOC]s to those measured near an O&NG processing site (Fig. 4).
The mobile source/O&NG processing signature, characterized by [acetylene] gas and [benzene] gas elevated above concentrations in other samples, was observed (see Supplementary Material; Fig. 5).The concentration of acetylene, which originates mostly from gasoline engines (Harley et al., 1992), was greatest in the sample from the O&NG processing site, at the Serra Cross (located adjacent to a parking lot, above the city of Ventura), and near I-5 in Carpinteria.VOCs from wells along and near the Ojai Valley or offshore platforms cannot be definitively differentiated Fig. 4. Enhancement ratio of the iC 5 /nC 5 ratio in grab samples, which demonstrates distinction between VOCs from raw O&NG/extraction versus vehicles/O&NG processing.Reference ratio from raw/O&NG extraction (Colorado) measured by Gilman et al. (2013); ratio from vehicle gasoline vapors (California) measured by McGaughey et al. (2004); ratio at oil spill site from current study (central dashed line).Fog period samples denoted by circles (darker color, larger marker indicates later sampling date and time), while grab samples collected from possible VOC emissions sources are indicated by circled "×" symbols.from those at the oil spill cleanup site because similar VOCs are emitted from these petroleum and extraction processes (U.S. Environmental Protection Agency, 1994).Simple empirical orthogonal function analysis (similar to principle components analysis) additionally confirmed that O&NG processing emissions and motor vehicle traffic were the most influential VOC emissions sources at CP (see Supplementary Material, Figs.S11 and S12).
The ratios of alkyl nitrates to parent alkanes can be used to determine the atmospheric residence time of VOCs in an air mass since alkyl nitrates are mainly formed from aging of alkanes in the presence of NO x (Bertman et al., 1995;Lee et al., 2014).Ratios of ethyl, 2-propyl n-butyl, and 2pentyl nitrate to parent alkane concentrations were quantified from the whole air grab samples (Fig. S13).No apparent trend was observed in the [ethyl nitrate]/[ethane] ratio, but enhancements were observed before midnight in [2-propyl nitrate]/[2-propane], [n-butyl nitrate]/[butane] and [2-pentyl nitrate]/[2-pentane], which is suggested here to indicate that more aged air before midnight was replaced by fresher emissions as winds shifted from westerly/offshore to onshore (see Supplementary Material for wind measurements).Concentrations of carboxylic acids were also expected to be formed from photochemical aging processes in the atmosphere; however, there was no apparent relationship between concentrations of alkyl nitrates and carboxylic acids (e.g., for [2-propyl nitrate] gas versus [carboxylic acids] air eq., r 2 = -0.13;p = 0.68).This lack of correlation may have resulted because alkyl nitrates are primary oxidation products, while C 1 -C 5 carboxylic acids typically require several oxidation steps for formation.Alternatively, differing atmospheric chemical conditions may be required for formation of carboxylic acids versus alkyl nitrates (e.g., Vione et al., 2005).
The formaldehyde concentration in the region also appears to have decreased between 1985/6 and 2015.The VWM [formaldehyde] aq measured in 2015 (3.4 µM) was approximately half that measured in 1985/6 (7 µM), but within the range of values for remote regions (van Pinxteren et al., 2005).Concentrations of formaldehyde are typically highest near urban areas (Collett et al., 1999).Formaldehyde is mainly produced in the atmosphere via oxidation of hydrocarbons: in remote and marine atmospheres, methane oxidation is the main production pathway, with inputs also from isoprene, while non-methane hydrocarbon oxidation contributes a larger fraction of formaldehyde in areas close to urban emissions (Fortems-Cheiney et al., 2012).Similar Both formate and acetate are suggested to originate from mainly secondary biogenic, but also anthropogenic sources (Paulot et al., 2011;Stavrakou et al., 2011;Bannan et al., 2014).The lack of trend may be a result of complex sources and sinks of these volatile carboxylic acids.
Shipping emissions are estimated to contribute substantially to O 3 and aerosol NO 3 -and SO 4 2-mass along the Southern California coast (Vutukuru and Dabdub, 2008;Coggon et al., 2012), and regional shipping ports have increased in capacity between the 1980s and 2010s (Andrew et al., 2011;US Department of Transportation Maritime Administration, 1985, 2015).Fog constituent concentrations have oppositely decreased, and were thus appear to have been minimally affected by increased shipping.Similarly, air mass transport across the Pacific has also been identified as a substantial source of pollutants to the western U.S. (Park et al., 2004;Heald et al., 2006), but would not explain the observed decreasing trend of pollutants, and especially SO 4 2-.Fog water constituent concentrations in some studies have exhibited a negative trend with atmospheric liquid water content (Elbert et al., 2000), although this is neither a linear nor a strong correlation in some environments (Straub et al., 2012).The effect of dilution, which would cause an inverse relationship between liquid water content and fog on average in 1985/6, but only 80 mg m -3 in 2015), while the 2015 concentrations were also lower; a liquid water effect, if dominant would have yielded higher concentrations in 2015.The 1985/6 study took place later in the summer, when fog frequency and density is typically higher (Johnstone and Dawson, 2010).Coastal fog along the California coast may also be declining over time as a function of the changing boundary layer height and coastal temperatures (Johnstone and Dawson, 2010), heat island effect (Williams et al., 2015), and/or as a result of climatic cycles (Herckes et al., 2015).Additionally, Munger and others have noted that fog frequency in 1985/6 along the California coast was anomalously high (Munger, 1989;Johnstone and Dawson, 2010).However, none of these liquid water content observations can explain the overall decrease in most fog constituent concentrations, which are more consistent with implemented reductions in pollutant emissions over the past three decades.

CONCLUSIONS
Fog chemical composition on the Southern California coast was determined from fog water samples collected at a site in the coastal hills using a variety of analytical techniques.NH 4 NO 3 and marine NaCl dominated fog water chemical composition.Fog samples also contained aged organic species of biogenic and anthropogenic origin, including nitrophenols, several families of carboxylic acids with up to six O atoms, and oxidation products of α-pinene and isoprene.Several CHOS species were identified, including some that were only detected after three to five hours of persistently foggy conditions, suggesting aqueous phase oxidation of organics.A strong correlation observed in three of four fog events between the aqueous concentrations of oxalate, methanesulfonate, and SO 4 2-suggested possible an in-cloud oxidation formation of some fog water constituents.The percOA was only ~20% on average, typical of fog samples containing fresh emissions (Collett et al., 1999;Herckes et al., 2007b;Giulianelli et al., 2014).Regional O&NG processing was determined as a primary contributor to [VOC]s during fog events.Influences of mobile sources and biogenic VOCs were also observed in air samples collected during fog events.
The results of a contrast between this study and a 1985/6 fog study highlight successes in emissions reductions for anthropogenic air pollutants.In contrast to fog water collected at the same location in 1985/6, the concentrations of anthropogenic species including NH 4 + , NO 3 -, and SO 4 2in the 2015 samples were lower by approximately an order of magnitude.The fog was also less acidic in 2015 by approximately two pH units.These differences in major chemical constituent concentrations between the 2015 and 1985/6 fog sampling campaigns can be explained by reduced gaseous and particle phase anthropogenic pollutant emissions due to effective air quality regulations.This work also demonstrates the need for additional research to determine the extent of in-fog formation of CHO, CHOS, CHNO, and CHNOS species.The formation of these species can result in increased organic aerosol mass formation (McNeill et al., 2012) and absorption of ultraviolet and visible light by aerosol (Lin et al., 2015), both of which are important for climate and air quality.The regional O&NG processing VOC emissions observed in this study are likely to be precursors for the CHO, CHOS and CHNOS species.Impacts of these emissions on air quality in the Southern California urban areas as well as natural areas, including the Channel Island National Park, should be considered.

Fig. 5 .Fig. 6 .
Fig. 5. VOCs measured from grab samples during fog sampling periods (solid symbols) and near two VOC sources potentially impacting CP: the oil spill cleanup site on I-5, and an O&NG processing site (empty symbols).
cannot explain the observed trends in chemical concentrations measured during the two campaigns.Measured liquid water content was lower in the 2015 than the 1985/6 study (approximately 170 mg m -3

Table 1 .
VWM, minimum, and maximum concentrations of organic and inorganic species quantified in 2015 CP fog samples.Limits of detection (LODs) calculated from replicate deionized water analyses, and confidence intervals about the multiple analyses of each measurement at the 95 th percentile (95% CI) are tabulated.The VWM pH was calculated as the negative logarithm of the VWM H + concentration.

Table 2 .
Values of fog pH observed at various locations around the world.