Drone-Based Measurement of the Size Distribution and Concentration of Marine Aerosols above the Great Barrier Reef

Abstract

Marine aerosol particles can act as cloud condensation nuclei and influence the atmospheric boundary layer by scattering solar radiation. The interaction of ocean waves and coral reefs may affect the distribution and size of marine aerosol particles. Measuring this effect has proven challenging. Here, we tested the hypothesis that the distribution and size of marine aerosol particles would vary over three distinct zones (i.e., coral lagoon, surf break, and open water) near One Tree Island in the Great Barrier Reef, which is approximately 85 km off the east coast of Australia. We used a modified DJI Agras T30 drone fitted with a miniaturised scanning electrical mobility sizer and advanced mixing condensation particle counter to collect data on aerosol size distribution between 30 and 300 nm at 20 m above the water surface. We conducted 30 flights over ten days during the Austral summer/autumn of 2023. The fitted bimodal lognormal curves indicate that the number concentrations for aerosols below 85 nm diameter are more than 16% higher over the lagoon than over open water. The average mean mode diameters remained constant across the different zones, indicating no significant influence of breaking waves on the detected aerosol size modes. The most influential explanatory variable for aerosol size distribution was the difference between air temperature and the underlying sea surface, explaining around 40% of the variability. Salinity also exhibited a significant influence, explaining around 12% of the measured variability in the number concentration of aerosols throughout the campaign. A calculated wind stress magnitude did not reveal significant variation in the measured marine aerosol concentrations. Overall, our drone-based aerosol measurements near the water surface effectively characterise the dynamics of background marine aerosols around One Tree Island Reef, illustrating the value of drone-based systems for providing size-dependent aerosol information in difficult-to-access and environmentally sensitive areas.

1. Introduction

Marine aerosols represent one of the most important natural aerosol systems globally [1,2]. They directly impact the energy balance in the marine atmospheric boundary layer by scattering solar radiation [3,4,5] and indirectly by acting as cloud condensation nuclei [6,7,8,9,10]. Marine aerosols consist of a combination of natural and anthropogenic elements deriving from sea spray aerosols (SSAs), the long-range transport of aerosols from land, secondary marine aerosols (SMAs), and particles emitted from ship exhausts [11]. In remote areas such as the Great Barrier Reef (Figure 1), marine aerosols primarily originate from SSAs and SMAs [12].

SMAs form through the conversion of gas-phase oxidation products from ocean-emitted precursor gases, such as dimethyl sulphide (DMS) and other volatile organic compounds (VOCs) (Figure 2). These gases are released due to the biological activity of marine phytoplankton, algae, and endosymbiont zooxanthellae found in reef-building corals [1,13,14,15]. DMS remains the dominant natural source of atmospheric sulphur [16,17,18]. However, its role in regulating climate [13], particularly the role of oxidation products of DMS acting as active cloud condensation nuclei, is a topic of debate [17,18].

The second source of marine aerosols is a mechanically driven process initiated by air–sea interactions (Figure 2). SSAs often dominate as the primary atmospheric aerosols over the ocean, particularly in remote areas with minimal human and continental influence [8,19,20]. The annual global emission flux of sulphates (all natural primary and secondary sources) ranges from 107 to 374 Tg y⁻¹, and that of sea salt ranges from 3000 to 20,000 Tg y⁻¹ [21]. SSAs encompass dry inorganic sea salt particles and organic components bound in seawater drops [9,22,23]. They mostly form when wind stress interacts with breaking waves, entraining air into the seawater surface (Figure 2). These air bubbles burst at the ocean surface, releasing film droplets that evaporate into SSAs, mainly consisting of sea salt, with a diameter smaller than 300 nm. Larger drops from bursting jets rise several centimetres into the air (jet drops) with some of them evaporating (Figure 2). This process leaves SSAs larger than two micrometres in diameter in the atmosphere, containing mainly sea salt, approximately 10% organic compounds, and bacteria from the ocean surface [1,21,24]. The largest SSAs are generated when wind speeds exceed ~10 m/s, as higher winds physically shear seawater spume droplets from the surface of breaking waves [24,25] (Figure 2). SSAs are also formed when ocean waves break onto beaches and reefs [26]. Anthropogenic aerosol sources, including fuel combustion, industrial processes, nonindustrial fugitive sources (like roadway dust from both paved and unpaved roads, wind erosion of cropland, construction activities, etc.), and transportation sources (such as cars, ships, airplanes, etc.), collectively contribute approximately 20% to global particulate emissions. Higher concentrations of these aerosols are predominantly observed in industrialised regions across the globe [27].

Figure 2. (A) Secondary marine aerosols are produced through the gas-to-particle conversion of oxidation products from gas-phase species emitted from the ocean due to biological activity. (B) The schematic illustrates the phases of film droplet and jet drop formation as an air bubble bursts at the sea surface: (a) ascent of the air bubble; (b) reaching the surface; (c) initiation of seawater film breaking; (d) formation of 5000 to 30,000 nm droplets from the upper part of the bubble film; (e) evaporation of film droplets, releasing sea-salt particles into the air; and (f) the burst results in 1 to 5 large drops (approximately 15% of the air bubble diameter) breaking away from the jet with a time span of approximately 2 ms among phases (c,f). (C) At elevated wind speeds (>~10 m/s), the tearing or breaking of wave crests leads to the release or injection of large droplets (figure adapted from Mayer et al. (2020) [12], Tomasi and Lupi (2017) [21]; Photo: M. Goolmeer).

Figure 2. (A) Secondary marine aerosols are produced through the gas-to-particle conversion of oxidation products from gas-phase species emitted from the ocean due to biological activity. (B) The schematic illustrates the phases of film droplet and jet drop formation as an air bubble bursts at the sea surface: (a) ascent of the air bubble; (b) reaching the surface; (c) initiation of seawater film breaking; (d) formation of 5000 to 30,000 nm droplets from the upper part of the bubble film; (e) evaporation of film droplets, releasing sea-salt particles into the air; and (f) the burst results in 1 to 5 large drops (approximately 15% of the air bubble diameter) breaking away from the jet with a time span of approximately 2 ms among phases (c,f). (C) At elevated wind speeds (>~10 m/s), the tearing or breaking of wave crests leads to the release or injection of large droplets (figure adapted from Mayer et al. (2020) [12], Tomasi and Lupi (2017) [21]; Photo: M. Goolmeer).

Size is one of the key determinants of aerosol behaviour and its impact on atmospheric and geophysical processes, influencing concentration, production rates, intrinsic properties, residence time, and mixing heights [23,27]. Aerosol particle size can refer to its radius [23,28] or diameter [21,27]. Size is usually assessed based on the size of an ideal spherical aerosol, which is equivalent to the volume of the actual aerosol particle [21]. Aerosols, especially SSAs, alter their equilibrium water content and size in response to atmospheric conditions primarily driven by ambient relative humidity [23]. The SSA size can be expressed as the dry size, assuming 0% relative humidity, the normalised size at 80% relative humidity (typical value over oceans), or the size at formation, representing the size in an environment with equilibrium relative humidity around 97% to 98% just above seawater [1,23,28]. A rule of thumb states that the size at 80% relative humidity is approximately twice the dry size, and the size at formation is roughly four times the dry size [23,28]. The same physical assumption is often made for aerosols other than SSAs within the marine boundary layer [23].

The residence time of SSAs in the atmosphere varies from seconds to days, depending on their size. Larger aerosols, particularly those exceeding 5000 to 10,000 nm in diameter, have shorter lifetimes, settling back to the surface more effectively with increasing size. Aerosols in the accumulation-mode size range (100 nm ⪅ D_dry ⪅ 1000 nm) have the most extended atmospheric lifetime, and clouds and precipitation predominantly remove them. With decreasing size, diffusivity increases, making the smallest aerosols more susceptible to removal through coagulation with other aerosols and dry deposition [19,24,29,30].

When considering the process leading to the formation of SMAs and SSAs and their subsequent upward entrainment into the atmosphere, the production rate is affected by various meteorological and environmental factors that influence the surface properties of the ocean. One key meteorological factor is wind, which causes waves to break, forming whitecaps and tearing droplets from the wave crest [15,23] (Figure 2). Wind also affects the vertical and horizontal entrainment of aerosols [23,31]. The vertical motion of aerosols within the marine boundary layer depends on atmospheric stability, which is determined by the vertical profiles of wind speed, temperature, and relative humidity near the sea’s surface. Atmosphere–ocean interactions can be described by the total wind stress magnitude variable, which depends on wind speed and atmospheric stability. When trying to understand the size-dependent marine aerosol composition at a specific location, it is important also to consider other factors affecting aerosols’ entrainment, advection, and removal at prior times and locations, as governed by the size-dependent atmospheric residence time [23] and history of the airmass. Sea temperature determines seawater’s kinematic viscosity, affecting the bubble rise velocities and the gas exchange between a bubble and the surrounding fluid and, therefore, the interfacial production flux [23]. The difference between the air temperature at 10 m above the surface and the water temperature is a key factor that influences the convective and mechanical mixing above the ocean. The salinity content of ocean water affects the properties of seawater, impacting the density, buoyancy, and behaviour of dissolved gases [23]. The mentioned factors represent only a selection of key parameters that possibly influence the size-dependent composition of marine aerosols at a specific location within the complex ocean ecosystem.

Various methods have been employed for measuring natural aerosols in maritime environments with each having advantages and drawbacks. Land-based lidar systems [32,33] provide detailed data but have significant costs and are location-bound. Aerosol counters and spectrometer probes on poles [34] offer valuable information but are typically confined to onshore locations and specific altitudes. Buoy-mounted aerosol probes [35] allow in situ measurements in wave-breaking zones but are severely impacted by wave height and tide conditions and have altitude restrictions. Shipborne measures [36,37] offer mobility but are prone to ship-induced contamination and cannot access shallow reef areas where waves break. Kite platforms [38] provide environmentally friendly sampling but depend on favourable wind conditions and have limited payload capacity. Aircraft-based measurements [39] provide good spatial coverage but have altitude restrictions close to the water surface and involve high costs and logistical complexities. Multirotor drones have emerged as a promising solution for investigating marine aerosol production globally [40]. These drones provide flexibility, mobility, and hover capabilities and also operate without emissions, making them well suited for remote, hard-to-reach, and environmentally sensitive areas.

This study used a large hexacopter drone, the Agras T30 (T30) (DJI, Shenzhen, China) (Figure 3), which is equipped with particle sizing and counting sensors to collect aerosol size distribution profiles at an altitude of 20 m above three distinct environmental zones in the Great Barrier Reef (Figure 1). The research-grade instruments used had a detection limit capable of discerning natural background aerosol number concentrations above the Great Barrier Reef. The T30 drone was chosen for its ability to carry the instrumentation despite an overall payload weight of ~8 kg. Using this drone, the primary objective of our study was to investigate variations in submicron aerosol distribution influenced by breaking waves around an offshore coral reef. The study focused on three specific zones: a lagoon, the surf break, and open water around One Tree Island. Each zone was chosen for its unique environmental dynamics: the lagoon, characterised by sheltered conditions with limited wave action; the surf break, known for intense wave activity; and the open water, serving as a control with continuous exposure to oceanic conditions.

2. Materials and Methods

2.1. Sampling Methodology

A modified DJI hexacopter Agras T30 drone was used in March 2023 to sample submicron aerosol size distributions above three environmentally different zones (lagoon, surf break, and open water) near One Tree Island (Figure 1). This small coral cay is located approximately 85 km off the east coast of Australia (23.50833° S, 152.09167° E) and is a protected research zone within the Great Barrier Reef Marine Park.

Sampling sets consisted of three consecutive drone flights each hovering for seven minutes at 20 m above the target zone. The order of the three flights was randomised to ensure systematic biases were avoided. The influence of breaking waves on marine aerosols, likely to persist in the atmosphere, becomes detectable when measuring size-resolved aerosol number concentrations at 20 m above the three designated locations [32,41,42]. Throughout the field campaign, we conducted 10 sets of 3 consecutive drone flights, primarily in fair weather, occasionally with partly cloudy or cloudy skies. The general sampling zones were positioned to the north and northwest of One Tree Island (Figure 1). Due to the large size of the drone (Figure 3), we chose the compacted coral spit northeast of the research station as the take-off and landing area (Figure 1). This area is accessible during mid-tide or lower, ensuring safe take-off and landing procedures for the large drone. This constraint maintains a relatively constant influence of water depth on wave characteristics and DMS emissions throughout the campaign [43].

2.2. Setup Hardware

The aerosol sampling equipment was mounted on a modified agricultural T30 spraying drone (Figure 3). The drone’s spraying system tank and nozzles were removed to create space for the instrumentation housing. With roughly 8 kg of payload from the aerosol sampling setup, the drone completed a 12-minute flight 20 m above the ocean and returned home with approximately 30% of the battery charge remaining. The impact of rotor-wash on the sampling air stream when hovering is typically negligible at a propeller length distance of 0.4 to 0.5 m above the drone (<25 kg) [40,44]. Therefore, on the larger T30, the air inlet was positioned about 0.9 m vertically above the drone’s centre to avoid the effects of local mixing produced by the rotors. We used a custom-built aluminium frame to secure the inlet tube and prevent vibration-induced resonance. We also designed a separate aluminium structure to hold the instrumentation box securely, replacing the removed liquid tank. The instrumentation’s power supply was connected to the 29,000 mAh, 51.8 V drone battery through an electrical step-down converter. The T30, designed for spraying fertilisers and pesticides, is rated IP67, and it is washable and corrosion-resistant [45]. It is, therefore, suitable for operations in the harsh environmental conditions above the Great Barrier Reef. A specially designed carbon fibre box housed the aerosol sampling setup. The integrated instruments included a model 9403 advanced mixing condensation particle counter (aMCPC) and a model 9404 miniaturised scanning electrical mobility sizer (mSEMS) (both Brechtel, Hayward, CA, USA) (Figure 3). It also featured a model 9002 XRC-05 soft X-ray charger (HCTM, Icheon-si, Republic of Korea) (XRC-05) and an ACC sample flow dryer (Brechtel, Hayward, CA, USA) (ACC-Dryer) (Figure 3). The aMCPC, with a 180 ms response time, detects particles from 7 to 2,000 nm. The unit was modified during a previous project’s field campaign [46] and now operates at 5 Hz with onboard data storage for efficient use on mobile platforms [47]. It can operate within ambient temperatures ranging from −20 to 38 °C, weighs 2.3 kg, and has a recommended concentration limit of 10⁵ cm⁻³ with an associated uncertainty of ±8% [47].

The mSEMS is a compact module for sizing particles that is equipped with a miniature differential mobility analyser column for selecting particles based on their size. The sizing module must be used with an aMCPC to measure particle concentration. This combination offers scan times as short as 5 seconds and can select and count aerosol particles between 5 and 375 nm in diameter (at a 2.0 lpm sheath flow) [48]. In this setup, specific scan settings included a sheath flow rate of 2.5 lpm and a scan size range from 30 to 305 nm, divided into 30 bins, each with a one-second scan time. The total scan time of 30 seconds allows for up to 14 completed scans per sampling event. The mSEMS has an operating temperature range from −20 to 35 °C and weighs 1.6 kg [48]. The XRC-05 uses soft X-rays, rather than radioactive materials, to charge-neutralise aerosol samples before they enter the mSEMS. It operates within a flow rate range of 0.3 to 4.0 lpm and can handle a maximum input concentration of 10⁷ aerosols/cm³. The device itself weighs 1.1 kg [49]. The ACC-Dryer has Nafion tubing running through a chamber filled with desiccant to absorb moisture in the sample flow [50]. When filled with silica beads, the unit weighs approximately 0.3 kg and is necessary for sample operation in regions with high relative humidity. Air temperatures during the Austral summer/autumn in the subtropical area around One Tree Island approached the upper environmental operating limits of the instruments, necessitating the integration of six 60 mm × 60 mm 12 V cooling fans into the carbon fibre housing.

2.3. Modelling Data and Analysis

To gain insight into the geospatial pathway taken by the relevant air parcel sampled in this experiment, the National Oceanic and Atmospheric Administration’s Air Resources Laboratory’s Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT) [51] was used to model 48 h backwards trajectories for each sampling run. The input data for each trajectory included the average timestamp of three consecutive sample runs, and the final altitude of the air parcel at the end of the trajectory was set to 20 m. The modelled values (i.e., ambient air temperature, mixing height, and relative humidity) were exported for each calculated way-point of the back-trajectory. To obtain a broader picture of the underlying state of the ocean, the trajectory was merged with information from the eReefs hydrodynamic model, which was part of the eReefs project, a collaboration between CSIRO, the Australian Institute of Marine Science, the Bureau of Meteorology, the Government of Queensland, and the Great Barrier Reef Foundation [52,53]. The one-hour hydrodynamic model run employed a 1 km and 4 km grid size each with different spatial coverages. The chosen modelled variables included water temperature (set to −0.5 m water depths), wind speed and direction at 10 m (derived from the Bureau of Meteorology, Australian Community Climate and Earth-System Simulator weather models [54]), and salinity. The retrieval of the variables followed a spatial crop of the hourly way-points of the backward trajectory, which were constrained by the boundary of the 1 km grid eReefs extend. The remaining trajectory was then cropped to the boundary of the 4 km eReefs model. The subsets of the back-trajectory were subsequently merged with the eReefs data using the timestamps provided in both datasets.

The total wind stress magnitude τ [Pa] transmitted on the ocean by wind was estimated by the following formula:

$$\mathsf{\tau }={\mathrm{C}}_{\mathrm{D}}\times {\mathsf{\rho }}_{\mathrm{a}\mathrm{i}\mathrm{r}}\times {\mathsf{\upsilon }}^2$$

(1)

where C_D is a dimensionless wind stress coefficient (set to 0.0013) [23,55], ρ_air [kg m³] is the density of air, and υ [m/s] is the wind velocity at 10 m above the ocean surface. The density of air ρ_air was calculated using a combined form of the ideal gas law:

$${\mathsf{\rho }}_{\mathrm{a}\mathrm{i}\mathrm{r}}=\frac{{\mathrm{p}}_{\mathrm{d}}}{{\mathrm{R}}_{\mathrm{d}}\times {\mathrm{T}}_{\mathrm{K}}}+\frac{{\mathrm{p}}_{\mathrm{v}}}{{\mathrm{R}}_{\mathrm{v}}\times {\mathrm{T}}_{\mathrm{K}}}$$

(2)

where p_d [Pa] is the pressure of dry air, R_d [J/kg K] is the specific gas constant for dry air (287.058 J/kg K), T_K [K] is temperature, p_v [Pa] is the actual water vapour pressure and R_v [J/kg K] is the specific gas constant for water vapour (461.495 J/kg K). The actual vapour pressure p_v is a product of the saturation vapour pressure p_s and the relative humidity RH [%], which was transformed from a percentage to a decimal:

$${\mathrm{p}}_{\mathrm{v}}={\mathrm{p}}_{\mathrm{s}}\times \frac{\mathrm{R}\mathrm{H}}{100}$$

(3)

Multiple formulas for calculating the saturation vapour pressure p_s [hPa] exist [56,57,58]. Here, we used the improved magnus form approximation for moist air above a plane surface of water [59]:

$${\mathrm{p}}_{\mathrm{s}}=1.00071\times {\mathrm{e}}^{0.0000045 \times \mathrm{p}}\times 6.1094\times {\mathrm{e}}^{17.625 \times {\mathrm{T}}_{°\mathrm{C}}/(243.04 + {\mathrm{T}}_{°\mathrm{C}})}$$

(4)

where p [hPa] is the ambient air pressure, T_°C [°C] is temperature, and there are five empirically fitted constants. The pressure of dry air p_d [Pa] is calculated by

$${\mathrm{p}}_{\mathrm{d}}=\mathrm{p}\times 100-{\mathrm{p}}_{\mathrm{v}}$$

(5)

where p [hPa] is the ambient pressure, and p_v [Pa] is the actual vapour pressure. Furthermore, the difference between the temperature of the air at 10 m above the surface and the temperature of the underlying ocean at 0.5 m below the surface was calculated (temp_diff = temp_air − temp_water). The mixing depth of the atmospheric boundary layer was obtained from the extended meteorological data of the back-trajectory, and the ocean’s salinity was provided by the eReefs model and expressed as a conductivity ratio with no units (practical salinity scale). All environmental factors were categorised based on the trajectory hours into three groups: 0 to −5 h, −5 to −10 h, and beyond −10 h. Within these categories, the mean for each factor was calculated. The average duration, the HYSPLIT backward trajectory that aligns with the eReefs data, is 29 h. The approximate atmospheric residence time for SSAs is 12 h [11,60]; for SMAs, it is 24 h [61]. After assessing the collinearity of environmental factors for distance-based linear modelling, we narrowed the independent predictor variables to (i) total wind stress magnitude, (ii) air–water temperature, and (iii) salinity. These independent predictor variables were then averaged to generate a single value for each throughout the back-trajectory when the spatial path aligns with the boundaries of the eReefs model.

The inversion of the raw counts into number concentration versus aerosol diameter data was performed using an algorithm provided by the manufacturer of the sampling instruments [62]. As no other instrument was used for a further investigation of the chemical composition of the sampled marine aerosols (i.e., single-particle Raman spectroscopy [63]), a generalised adjustment for the relative humidity was employed. According to tests conducted by the instrument manufacturer, the integrated ACC-Dryer can reduce an input ambient relative humidity of up to 90% to an output relative humidity of 30% for at least 5 h of operational time. These results are achieved with a flow rate of 0.36 lpm and by using 1.5 to 3 mm granular silica gel as a desiccant, as employed during this experiment [50]. Regarding the hygroscopic growth of particles with varying organic content, we apply a factor of 0.9 (for 25% to 30% relative humidity in the sample flow) to the measured size ranges [64,65]. Subsequently, we refer to the aerosol size in terms of D_dry simply as D.

The loss related to the sampling and transport of aerosols with an inlet tube system was calculated using the Particle Loss Calculator software (Max-Planck Institute for Chemistry, Leipzig, Germany) [66] within the IGOR Pro 9 environment (WaveMetrics, Portland, OR, USA). For each of the 30 size bins, a mean aerosol diameter was calculated from all scans, and the corresponding loss percentage was retrieved using the BinarySearchInterp function. The transport loss decreases from 12.32% in bin 1, with a mean diameter of 31 nm, to 1.19% in bin 30, with a mean diameter of 291 nm.

Each sampling run involved a seven-minute hover above each of the three zones (lagoon, surf break, and open water) with a fixed scanning time of 30 seconds per scan to examine the short-term impact of location on aerosol number concentration. Approximately 14 measurements were accumulated during each size bin during these runs (invalid measurements, indicating no count results from the connected aMCPC, were excluded from the scan data before analysis). A mean value (μ) was computed for every size bin and scan to consider sizing variability. Subsequently, all measured aerosol number concentrations were normalised to this mean value, providing a centred representation of the standard deviation for each size bin. In total, 11,039 measurements were considered (Figure 4).

We parameterised the number distributions into two lognormal modes (ultrafine mean diameter < 85 nm and fine mean diameter > 85 nm) to further evaluate the longer-term influence of breaking waves on aerosol number concentrations. The amplitude dN/dlogD [cm⁻³] and mean-mode aerosol diameter of the bimodal fitted curves are referred to as height and centre. The parameterisation process involved a median split of each sampling run based on the calculated total scan number concentrations to take the variability of the 7-minute hover into account for further analyses. A bimodal lognormal curve fitting was employed for both averages of each sampling run, ‘low’ and ‘high’ number concentrations. The fitting procedure used the Multipeak Fit package V3 within Igor Pro 9. Two lognormal peak functions with a baseline set to 0 and a maximum width set to 0.5 were used as constraints for the fitting. The Levenberg–Marquardt algorithm, integrated into the Multipeak Fit package, iteratively adjusts the fit parameters to minimise the sum of squares of deviations, starting from its initial guesses.

We used permutational analysis of variance (PERMANOVA) [67] to test hypotheses about the significant influence of zones on the size distribution of aerosols 20 m above the ocean surface. These analyses treated the ‘zone’ factor as an orthogonally fixed variable with three levels (open sea, surf break, and lagoon). At the same time, the ‘day’ of sampling was considered a random factor. Each size median split number concentration, characterised by its two fitted modes, was treated as an individual replicate in the PERMANOVA analyses, which were based on Euclidean distance and involved 9999 permutations. In case of a significant main effect, pairwise post hoc tests were used to establish differences among zones. Additionally, we explored the potential impact of the average values of the three most uncorrelated meteorological and environmental factors derived from HYSPLIT and eReefs model data (total wind stress magnitude, the temperature difference between the air at 10 m above the surface and the water 0.5 m below the surface, and salinity) using a Distance-Based Linear Model (DistLM) with a selected model structure, guided by the Corrected Akaike’s Information Criterion selection criterion and a best-fit selection procedure. PERMANOVAs, post hoc tests and DistLM analyses were performed using the PRIMER 7 (Version 7.0.21) statistical software with the PERMANOVA+ add-on (PRIMER-e, Auckland, New Zealand).

3. Results

A total of 30 valid seven-minute scanning periods conducted over eight monitoring days resulted in 11,039 measurements of aerosol number concentrations, which were segmented into 30 size bins. Subtracting the calculated mean values for each bin in a scanning period, the aligned overall standard deviations reveal significantly greater short-term variability above the surf break and even more so above the lagoon in the larger aerosol diameter bins (>85 nm) of the sampled size range (Figure 4). The standard deviation reaches its peak over the lagoon in size bin 23 (D ≈ 144 nm ± 5 nm) with 325 dN/dlogD [cm⁻³], and it achieves its minimum of 87 dN/dlogD [cm⁻³] over open water in size bin 29 (D ≈ 265 nm ± 5 nm) (Figure 4). The mean standard deviation over all scans and locations is 151 dN/dlogD [cm⁻³]. The most considerable difference appears between the measurements over the lagoon and the open water in size bin 30 (D ≈ 290 nm ± 5 nm) with a 220% difference. The mean difference among the standard deviations is 59% between lagoon and open water, 30% between lagoon and surf break, and 31% between surf break and open water (Figure 4).

After the median split, the averaging and parameterisation process of the inverted raw scan data, the identified bimodal distributions (ultrafine < 85 nm and fine > 85 nm) in terms of amplitude of the number concentrations (dN/dlogD [cm⁻³]) (referred to as height), and mean-mode-aerosol diameters (referred to as centre), exhibit variability across three distinct locations: lagoon, surf break, and open water (Figure 5). Concerning heights, percentual differences ((fine – ultrafine/fine) × 100) reveal nuanced distinctions between the two fitted modes in each zone (Figure 5). Over the lagoon, the results indicate a slightly lower mean height (−0.42%) compared to the surf break (−1.83%) and a more pronounced difference with open water (11.83%) (Figure 5). The most substantial difference among zones, within the same size mode, is observed in the ultrafine mode with 14.06% more aerosols measured above the lagoon (Figure 5). Consistently, the lagoon presents higher mean heights for ultrafine aerosols at 711 dN/dlogD [cm⁻³] and fine aerosols at 708 dN/dlogD [cm⁻³]. In contrast, surf break and open water exhibit lower mean heights, with open water ultrafine aerosols having the lowest mean height at 611 dN/dlogD [cm⁻³]. The largest interquartile range (Q1_25%–Q3_75%) was calculated for fine particles over the lagoon with a 543 dN/dlogD [cm⁻³] value. The lowest range is observed for ultrafine aerosols above the surf break with a value of 202 dN/dlogD [cm⁻³] (Figure 5). Regarding particle centres, the percent differences between sampling locations within size modes only show slight variations (centre_min to centre_max: ultrafine = 49 to 51 nm, fine = 144 to 147 nm). The maximum difference between the open water and surf break appears within the ultrafine mode: just above a 3.9% difference (Figure 5).

The aerosol number concentration heights of the bimodal fitted curves for all three sampling locations and all considered environmental factors (total wind stress magnitude, air–water temperature difference, salinity, mixing depth and relative humidity) varied among sampling times (Figure 6,

Table 1. Ranges of considered environmental factors, along with the modelled trajectories, number concentration amplitudes (height), and mean mode diameters (centre) of the bimodal fitted curves. The presented values encompass the analysis of 120 results obtained from fitting curves over ten sampling days and 385 modelled parameters for each environmental factor. PERMANOVAs were performed with 9999 permutations using Type III SS and a calculated Euclidean dissimilarity matrix. The Distance-Based Linear Model was constructed with a Corrected Akaike’s Information Criterion selection criterion and a best-fit selection procedure. p values are obtained using permutation.

	MIN	MAX	MEAN	±SD
Total wind stress magnitude [Pa]	0.003	0.158	0.050	0.037
Air–water temperature [°C]	−3.1	1.2	−1.0	0.9
Salinity	34.95	35.60	35.20	0.13
Mixing depth [m]	250	936	446	212
Relative humidity [%]	70	94	83	5
Height [dN/dlogD [cm⁻3]]	ultrafine|fine
Lagoon	392|267	1498|1625	711|708	251|408
Surf break	377|240	1123|1329	668|656	175|328
Open water	284|235	935|1519	611|693	196|422
Centre [nm]
Lagoon	42|121	60|175	50|144	5|13
Surf break	42|121	70|184	51|147	7|16
Open water	40|110	58|201	49|145	5|20

). Although fine aerosol number concentrations (PERMANOVA, p < 0.05) did not differ significantly among zones, there was a significant zone effect for ultrafine aerosol particles (PERMANOVA, p < 0.05,

Table 2. Results of PERMANOVA, pairwise tests, and distance-based redundancy analyses.

PERMANOVA (Ultrafine|Fine)
	df	Pseudo-F	p
Location (Fixed)	2|2	3.237|1.832	0.037|0.183
Day (Random)	9|9	12.373|115.57	0.0001|0.0001
PAIRWISE TESTS (ultrafine)
		Pseudo-t	p
Lagoon–Surf break		0.974	0.3546
Lagoon–Open water		2.444	0.0146
Surf Break–Open water		2.088	0.0423
DISTANCE BASED LINEAR MODEL (ultrafine|fine)
MARGINAL TESTS		R2	p
Total wind stress magnitude		0.03|0.00	0.1955|0.8012
Air–water temperature		0.34|0.40	0.0001|0.0001
Salinity		0.12|0.14	0.0056|0.0044
BEST SOLUTIONS (ultrafine)		R2	AICc
Air–water temperature		0.34	620.14
Air–water temperature, salinity		0.34	622.24
BEST SOLUTIONS (fine)
Air–water temperature		0.40	685.92
Total wind stress magnitude, air–water temperature		0.42	686.48

). The maximum magnitude of ultrafine particles differed significantly between the lagoon and open ocean (post hoc tests, p < 0.05) and between the surf break and open water (post hoc tests, p < 0.05, Table 2). As a predictor variable, the total wind stress did not explain the significant variation for ultrafine or fine aerosols (p > 0.05, Table 2). In contrast, air–water temperature and salinity explained significant variation in fine and ultrafine aerosol concentrations (p < 0.05 for all tests, Table 2). The difference in air–water temperature was the best single variable solution, with R² values of 0.339 and 0.401 for ultrafine and fine aerosols, respectively. However, when all predictor variables were included in the models, the most variation was explained by the combination of air–water temperature and salinity for ultrafine aerosols and air–water temperature and total wind stress for fine aerosols (Table 2).

4. Discussion

Laboratory-generated studies on sea spray production, focusing on the submicrometre size distribution structure, have revealed distinct size modes at approximately 100 to 110 nm and a secondary mode at around 40 to 45 nm in dry diameter [1]. This observation aligns with the minimum values recorded over open water (Table 1). After examining the overall mean values across the three zones, two distinct modes were identified at 50 nm and 145 nm. These slightly elevated values may be attributed to the regional-specific influences or the impact of humidity on the sampled aerosols. For our study, we assumed a relative humidity in the sample flow below 30% and applied a correction factor of 0.9 based on the manufacturer’s recommendations. For future measurements, acquiring humidity measurements of the sample flow is advisable to ensure compliance with these specified values. After applying the bimodal curve fitting routine, the presented values effectively capture the background aerosol measurements. Upon examining the mean centre values, breaking waves have no discernible influence on aerosol size modes. Hence, the aerosol spectrum between 30 and 300 nm above the lagoon and surf break seems to resemble the spectrum observed above the open water [68]. The amplitudes of the fitted bimodal lognormal curves for ultrafine aerosols at an altitude of 20 m over the lagoon exhibit a noticeable location effect (Figure 5, Table 1). Breaking waves have been observed to elevate the concentration of aerosols in proximity to the water surface by approximately 0.7 to 1 order of magnitude, as referenced [68]. Although our results corroborate this phenomenon, the magnitude differs (Table 1).

An increase in sea surface temperature reduces surface tension and decreases kinematic viscosity in water, thereby enhancing bubble rising rates and gas exchange with the surrounding fluid [23,26,28]. The atmospheric stability and the convective and mechanical mixing in the atmosphere above the ocean are primarily determined by the difference between air and sea surface temperatures [23]. Consequently, this temperature difference plays a role in the upward entrainment of SSAs [23]. Parameterisation attempts for DMS emission from the ocean surface have often used sea surface temperature as an influencing factor [69,70,71]. The specific roles of sea surface temperature and the difference between air and sea surface temperature as predictors of marine aerosols are not fully understood [24,30,72]. The most influential environmental factor affecting the aerosol concentrations in our study was the difference between air temperature and water temperature at a depth of −0.5 m, explaining around 40% of the variability (Table 2). This finding aligns with the assumption that the air–sea temperature difference plays a crucial role in parameterising the flux of SSAs [26] and SMAs [17]. After averaging this variable along the back-trajectory, a consequent PERMANOVA analysis revealed that all sampling days exhibited a negative heat flux, varying in strength from −1.68 to −0.10 °C. Our data indicate that within the boundary of the eReefs model, the measured air parcels primarily traversed during the night, devoid of solar radiation, leading to elevated water temperatures compared to the surrounding air. The moderate negative correlation between the air–water temperature difference and the number concentrations of the sampled marine aerosols implies that a more pronounced negative heat flux along the air parcel trajectory is likely linked to lower marine aerosol number concentrations around One Tree Island.

Salinity has a substantial influence on various properties of seawater [17] and is thought to influence the production of SSAs [11] and the dynamics of SMAs [17]. However, despite its potential significance, salinity has received relatively scant attention as an environmental factor predicting production fluxes [73]. While it had been suggested that on a global scale, salinity usually does not play a significant role in sea spray aerosol production, mainly due to the observed uniformity in salinity levels across the world’s oceans [30], our data suggest salinity is significant predictor at local scales on the GBR. When modelling a pathway and considering spatial and temporal variations in salinity levels, we observed a notable influence of salinity on aerosol number concentrations. Our data suggest a trend wherein increasing salinity augments ultrafine and fine marine aerosols around One Tree Island. This observation may imply the predominant involvement of SSAs in the measured marine aerosols [74,75]. However, a more comprehensive understanding of the composition and differentiation between SSAs and SMAs should be pursued through additional instrument integration in subsequent field campaigns.

Removing the spray equipment from a readily available agricultural drone facilitated sampling of the size distribution and concentration of marine aerosols with a ~8 kg instrument payload. Previous drone aerosol research included less payload capacity, using only an optical particle counter [76,77] for sizing and counting. However, this approach had limitations, such as a size range limitation of >300 nm. Alternatively, researchers utilised multiple drones with split payloads flying simultaneously [78] or employed purpose-built multi-copter drones [76,79]. With the proposed setup, we achieved 30 sampling flights measuring size-resolved aerosol number concentrations of background marine aerosol (<300 nm) in the hot, humid environment of the southern GBR. The lifting capacity of the employed T30 drone leaves the potential to integrate a further 10 kg of instrumentation payload and still achieve ⪆10 min flight time [45]. An additional optical particle counter or potable optical particle spectrometer could enhance the size range up to around 3000 nm aerosol diameter [76,80,81]. Amperometric gas sensors, metal oxide semiconductor sensors, non-dispersive infrared sensors, and photo-ionisation detectors would enhance the capability of the setup for air quality assessment and the potential identification of anthropogenic aerosol sources influencing the measured marine aerosol load [40,82]. Redundant aerosol sizing and counting instruments could be implemented to determine differences between SSAs and SMAs based on a detailed determination of the hygroscopic growth [83]. Around One Tree Island, our findings indicate that wind stress does not substantially influence marine aerosol size and concentration. This conclusion diverges from earlier research in other locations, which reported a notable impact of wind stress, particularly on SSAs abundance [21,24,84]. A strong correlation between aerosol number concentration and wind speed often relies on the specified conditions prevailing at the time and location of the measurements [26]. The maximum tolerable wind speed the drone can handle is 8 m/s [45], so strong winds were not incorporated into our analysis. The importance of wind in aerosol formation is likely to be greatest when the winds are strong, and the lack of a significant wind effect in our data could result from a truncated data set. Overall, the current generation of multirotor drones generally limits the collection of environmental and atmospheric data in relatively benign weather conditions. Drone technology is, however, rapidly improving, and in the future, it may be possible to operate multirotor drones safely in more extreme weather conditions.

5. Conclusions

This paper highlights the efficacy of using a multirotor drone with advanced aerosol sizing and counting instruments to measure fine and ultrafine marine aerosols in challenging environments, such as surf zones and sheltered coral lagoons. We observed horizontal variability in size-resolved aerosol number counts (ranging from 30 to 300 nm, categorised into 30 bins) above distinct environmental zones (lagoon, surf break, and open water). Our findings reveal significant horizontal variability in aerosol concentrations influenced by local environmental dynamics. Temperature differences between air and water along the aerosol back-trajectory notably impacted marine aerosol concentrations. Over the lagoon, there were more than 16% additional ultrafine aerosols than over open water. Future efforts to parameterise marine aerosol production flux around coral reefs should incorporate additional instrumentation to distinguish between SSAs and SMAs. This approach could provide deeper insights into the composition and characteristics of background marine aerosols. In summary, our study highlights the advantages of employing advanced drone technology to explore complex aerosol dynamics in marine environments previously inaccessible for direct measurement. The spatial and size-resolved aerosol in situ measurements presented here contribute significantly to understanding ocean–atmosphere interactions and aim to refine the modelling of background marine aerosols.