Airborne transmission pathway for coastal water pollution

Tracer dye as a water pollution mimic

The CSIDE study simulated the impacts of coastal water pollution events on coastal air quality by releasing rhodamine WT (RWT) dye into the surfzone at Imperial Beach, CA, USA (Fig. 1) (Grimes et al., 2020). Rhodamine WT (molecular weight 566.99 g/mol) is a water soluble, non-toxic, fluorescent dye used as a semi-conservative water mass tracer, having no known oceanic sources and a minor sink in the form of photodegradation (Smart & Laidlaw, 1977; Wilson, Cobb & Kilpatrick, 1986; Suijlen & Buyse, 1994; Clark et al., 2009). It is a non-volatile, solid compound at room temperature with maximum fluorescence at excitation/emission wavelengths of 558/582 nm (Wilson, Cobb & Kilpatrick, 1986; Mackay & Shiu, 1981). We simulated three pollution events by releasing roughly 30 gallons of concentrated RWT solution into the surfzone on September 23 (early morning, 3.2 h release), October 8 (early morning, 3.8 h release), and October 12 (mid morning, 1.8 h release into estuary outflow on ebb tide), 2015 (Fig. 1). Sea surface concentrations of the fluorescent dye (Fig. 1, Table 1) were measured with a hyperspectral camera mounted on an airplane that flew over the study site multiple times each hour on the first 2 days of each dye release (DR) (Melville et al., 2016). The hyperspectral camera was calibrated with in-situ near-surface dye measurements to provide dye concentrations in surface waters with high spatial resolution (2 m × 2 m) (Grimes et al., 2020; Melville et al., 2016; Clark et al., 2014; Feddersen et al., 2016). This method is limited to daylight hours, and therefore no nighttime sea surface dye concentrations are reported.

Aerosol sampling

Atmospheric aerosols were sampled at up to 3 sites over 2–4 days following each DR, at a total of 9 locations (Fig. 1 and Table 1). Aerosol sampling began the morning of each DR and continued into the afternoon, when samples were recovered and new sampling periods began, yielding ~6 h daytime and ~16 h overnight collection periods. Aerosol sampling locations were selected each morning based on the current dye observations (aerial and in-situ) in an effort to sample downwind of the ocean dye. The oceanic dye plume continued to evolve over time such that the aerosol sampling sites did not always remain downwind, and the downwind distance varied. Aerosols were collected using a PAS450-10 SpinCon I (heretofore “SpinCon I”; see Supplemental Materials for more information), which utilizes a swirling technique to transfer aerosols and water soluble compounds from the air into a fixed volume of sterile water (Michaud et al., 2018; Yooseph et al., 2013). The SpinCon I samples air at 450 lpm and can sample continuously for hours, greatly concentrating airborne constituents into a 10 ml liquid sample. Internal surfaces of the SpinCon I were cleaned with 70% ethanol and tubing was flushed with 10% bleach followed by sodium thiosulfate to neutralize the bleach. In between samples, rinse cycles, in which the instrument is run for 30 s and then the sample is discarded, were used to flush surfaces. Field blanks were collected on October 7–8, between the first and second DRs, after the dye had dissipated in the coastal waters for 2 weeks.

RWT measurement in aerosol samples

Samples were analyzed for RWT using fluorescence spectroscopy, normalizing to the water Raman peak (Lawaetz & Stedmon, 2009; Murphy, 2011). The fluorescence measurements were made the day the samples were recovered, using the Horiba Aqualog fluorescence spectrometer (Aqualog UV 800C, Horiba Ltd., Kyoto, Japan), at 425–625 nm excitation wavelengths, at 5 nm increments, and reading at emission wavelengths of 248–827 nm, at ~4.5 nm increments. A calibration curve was generated to quantitatively relate fluorescence to RWT concentration, which also provided the RWT fluorescence pattern in the excitation-emission matrix, which we used to identify RWT in aerosol samples (Fig. S1). Positive detection of RWT in the aerosol samples are those that showed the RWT fluorescence pattern and a RWT fluorescence intensity above that of our limit of detection derived from the field blanks (Eq. 1) (Armbruster & Pry, 2008):

(1) $$\mathrm{l}\mathrm{i}\mathrm{m}\mathrm{i}\mathrm{t}\mathrm{o}\mathrm{f}\mathrm{d}\mathrm{e}\mathrm{t}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}=\alpha -3.29\beta$$where α is the mean of the field blanks and β is the standard deviation of the field blanks. The RWT fluorescence pattern was not detected in any aerosol field blanks nor any sampler rinses.

Atmospheric dye concentrations

The average concentration of RWT in the air during each sampling period was determined by dividing the amount of RWT in the SpinCon I liquid sample by the total volume of air sampled. Figure S2 shows fluorescence spectra from a RWT standard (Fig. S2A), from aerosol samples with strong and weak RWT signatures (Figs. S2B & S2C), and from a field blank (Fig. S2D), as well as the peaks used for quantitation and the region for background correction. See Supplemental Information. RWT concentrations are reported for all air samples and those which met both criteria for RWT detection are noted (Table 1).

Dye concentrations in upwind waters

To determine sea surface dye concentrations upwind of the aerosol samplers ([dye]_sea (ppb); Table 1; x-axis in Fig. 3), wind vectors were used from a local meteorological station (KNRS, Fig. 1) and interpolated to the time intervals of aerial sea surface dye measurements. No lag between source and wind times was imposed because the observed ~5 m/s winds advected material from the surfzone to the furthest sampler (~1.5 km) in ~5 min, much shorter than the multi-hour sampling intervals. The intersection of the interpolated upwind vector with the tidally varying shoreline location, estimated using the 2012 NOAA Tsunami DEM and local water-level (including tides, but neglecting wave set-up), provided the center location for the upwind dye source in the ocean. A 200 m alongshore by 100 m offshore sea surface dye concentration window was extracted for each sea surface dye measurement taken during each aerosol sampling period. This box was limited to 100 m in the offshore direction because SSA was mainly generated in the surfzone region of depth-limited breaking waves (Van Eijk et al., 2011; de Leeuw et al., 2000). The spatial and temporal average of the sea surface dye concentration in those windows was taken for each aerosol sampling period ([dye]_sea (ppb)). This allowed us to precisely define the RWT aerosol source waters and quantitatively relate RWT concentrations in the ocean and atmosphere.

RWT transport by ocean currents

The three DRs demonstrated different ocean transport pathways of the RWT, driven by the coastal ocean currents. Figures 1A–1C show the maximum sea surface dye concentration at each point in the ocean on the first day of each dye release as measured from the air with the hyperspectral camera (Melville et al., 2016; Clark et al., 2014). Although temporal information such as persistence/transience is obscured in these figures, they provide a broad visual comparison of the major dye distribution for each DR. At the broadest level, the three DRs showed that the pollution mimic did not immediately disperse offshore, but instead remained within 1 km of the shoreline during the first day of each release, consistent with multiple studies showing surfzone trapping (Boehm, 2003; Grant et al., 2005; Clark, Feddersen & Guza, 2010; Rodriguez, Giddings & Kumar, 2018). For DRs 1 and 3 (Figs. 1A & 1C), the dye remained within ~2 km alongshore of the release location. For DR2 (Fig. 1B) a strong, wave-driven, alongshore current rapidly advected the dye ~10 km northward along the coast (Grimes et al., 2020; Wu et al., 2020). Thus for DR2, although high sea surface dye concentrations are seen along the coast, they only persisted at any given location for a short time due to the rapid northward advection. Sea surface dye concentrations are overall much lower for DR3, likely a result of localized mixing and, to a lesser extent, missing aerial measurements due to flight access limitations in the first few hours following the DR. Combined, the three DRs show the variability in water and pollution transport expected for the coastal ocean and the motivation for the CSIDE study. The ocean transport aspects of the CSIDE study are discussed in greater depth in other publications (Grimes et al., 2020; Wu et al., 2020).

RWT detected in coastal aerosols

Dye was detected in coastal aerosols on days 1 and 2 of DRs 1 and 3 in four total aerosol samples (Figs. 1A & 1C, Table 1). Dye was detected as far as 668 m inland and 720 m downwind from source waters. Wind rose plots (Figs. 1D–1F) for daytime winds show that onshore winds, typical of coastal meteorology patterns, were common. Wind data link the oceanic source waters with locations of atmospheric dye detection over land. Figure 1 broadly shows that atmospheric dye was detected downwind of the dye plume. The well characterized distribution of RWT in coastal waters allows quantitative comparison between dye concentrations in the air and in the source waters.

Comparing ocean and atmospheric observations

Repeated measurements of dye in the coastal waters allow us to examine dye concentrations in the source waters during the atmospheric sampling periods. Figure 2 shows the comparison of ocean dye data from two exemplary samples for when dye was and was not detected in the air. The mean sea surface dye concentration in the upwind waters during the aerosol sampling period is presented relative to the aerosol sampling location on land. The inset dotted box shows a 200 m × 100 m region centered where the mean wind direction during the aerosol sampling period intersected the coastal waters. Sample 2 (left panel) was collected on the first day of DR1 and sample 4 (right panel) was collected the day after, i.e. day 2 of DR1. Dye was detected in sample 2, which was collected downwind of waters with high dye concentrations. Dye was not detected in sample 4, which was collected downwind of waters with little to no detectable dye. Analogous plots for all air samples with coupled ocean measurements are provided in the Supplemental Materials (Fig. S3).

Atmospheric transfer of RWT in SSA

Surfzone production of SSA transferred RWT from the ocean to the atmosphere. Rhodamine WT’s transfer in SSA is consistent with the detection of other chemical species in SSA (Fleming et al., 2011; Cheng et al., 2005; Hawkins & Russell, 2010; Cochran et al., 2017; Kuznetsova, Lee & Aller, 2005) and larger entities including viruses and bacteria (Patterson et al., 2016; Baylor et al., 1977; Michaud et al., 2018; Aller et al., 2005). Dissolved RWT has a low surface activity, therefore it acts as a tracer of bulk seawater, which transfers to SSA primarily in jet drops. Sea spray aerosol production, specifically from depth-limited waves breaking in the surfzone, must have been the dominant RWT transfer mechanism because surfzone production of SSA has been shown to be dominant for low to moderate wind speeds (Van Eijk et al., 2011; de Leeuw et al., 2000). Dye transfer outside the surfzone was negligible because whitecap conditions were not common during the study period, due to low wind speeds often below 6 m/s.

Atmospheric transport of RWT SSA

We understand that the major factors influencing atmospheric concentrations of RWT SSA include: dye concentrations in aerosol source waters, which depends on advection and dispersion of the dye in the ocean, driven by waves, regional pressure gradients, buoyant plumes, winds, tides, and internal tides; ocean-to-atmosphere RWT SSA flux, which occurred primarily through SSA production by surfzone depth-limited wave breaking; downwind advection and horizontal and vertical dispersion in the atmosphere, driven by coastal winds and atmospheric boundary layer dynamics; and dry deposition (gravitational settling) of RWT SSA. We understand that dye was not detected in most aerosol samples due to high variability in ocean and air transport patterns. Photobleaching of RWT is negligible on our <4 d timescales (Suijlen & Buyse, 1994). We compare dye concentrations in the ocean and atmosphere to test whether source water dye concentrations influenced airborne dye concentrations (Fig. 3). Dye was not always detected when upwind waters contained elevated dye concentrations, due to downwind distance, 3D atmospheric dispersion, and the lack of a vertical sampling array. When RWT dye was detected in the aerosol (Fig. 3, magenta points), the data suggest dye concentrations in the air are proportional to dye concentrations in source waters. Importantly, there were no false positive atmospheric dye values because higher RWT concentrations in the air correspond to higher RWT concentrations in the ocean; there are no data points high in air RWT and low in ocean RWT (Fig. 3). More experiments with sampling at higher temporal and vertical/horizontal spatial resolution are needed to confirm a robust, quantitative relationship between SSA and upwind source waters, and to better understand the role of atmospheric dispersion.

Implications

Our results demonstrate the ability to trace SSA back to its source waters with high spatial and temporal resolution, and to quantitatively relate dye concentrations in source waters and air masses. More importantly, this study confirms coastal water pollution can be transferred from the surf into the atmosphere by releasing a tracer dye into the ocean and detecting it in the air inland from the beach. These observations highlight an under-appreciated airborne exposure pathway to coastal water pollution. We expand on the pioneering work of Baylor et al. (1977) by simulating large-scale artificial water pollution events, measuring sea surface dye concentrations at high spatio-temporal resolution (2 m × 2 m, hourly), tightly constraining RWT SSA source waters, and making quantitative ocean-atmosphere comparisons that extend inland beyond the beach. Whereas many atmospheric aerosol studies detect particles from distant sources, and aerosol dispersion models provide insight into particle origin at regional level resolution (~100 km), we were able to pinpoint SSA source waters on the <1 km scale. Although we were unable to detect RWT SSA beyond 668 m inland (Fig. 1, Table 1), it is well documented that SSA can travel much further (Smith et al., 2013; Bondy et al., 2017; Prospero et al., 2005; Smith et al., 2012). A 2 µm SSA particle has a predicted dry deposition residence time of ~1.5 weeks with an estimated travel distance of 10,000 km (Lewis & Schwartz, 2004). SSA particles this size have been detected hundreds to thousands of km from their oceanic source region (Bondy et al., 2017). Bacteria can remain viable after travelling thousands of km in the atmosphere (Prospero et al., 2005; Smith et al., 2012). Our findings emphasize that species transfer from the ocean to the atmosphere includes not just natural seawater components but also pollutants. This study also demonstrates that coastal water pollution reaches people beyond the shoreline through atmospheric transport. These results detail a promising approach for linking pollution in source waters, including the surfzone, estuaries, and other outflows, to atmospheric pollutants and health impacts downwind.