In May and June 2023, a scientific cruise was conducted in the SCS by the South China Sea Institute of Oceanology, Chinese Academy of Sciences, onboard the Yuezhanyuke No. 6. This study analyzes the aerosol-meteorology (AM) measurements along the section from the latitude 21°02^′ N to 8°5^′ N and the longitude 110°33^′ E to 115°25^′ E. All AM data were collected from 21 May to 15 June 2023.

To observe actual aerosol loadings in the typical marine environment, it is necessary to exclude aerosol observational data contaminated by ship emissions during the cruise (including data collected in offshore and pelagic regions). To achieve this, we performed quality control during aerosol observational data collection with accompanied WS, WD, rainfall intensity, and NC observational data as follows: i.

Data screening with wind observations. The north arrow of the Vaisala WXT530 was oriented perpendicular to the ship's longitudinal axis, pointing east relative to the vessel's heading. Observed WD thus represented the relative angle between the wind direction and the ship's course. When the ship was sailing, WD sampled ranged from 225 to 315°. We excluded these data because aerosol observations were directly affected by the ship emissions. As depicted in Fig. 2a, a sharp decrease and subsequent increase in the observed aerosol size distributions can be easily identified on 25 May. The underlying cause of abnormal changes in aerosol size distribution could be the removal of coarse aerosol mode by rainfall, as depicted in Fig. 2b. However, this “jump” was located at 07:00 LT (local time; circled in black in Fig. 2a) while the rainfall occurred at 08:00 LT and lasted for approximately 6 h, so it is not the rainfall that influences the data jump in aerosol size distribution. With the help of WD sequence data shown in Fig. 2b (circled in red), we found that the jump in WD curve was also located at 07:20 LT and thus identified the accurate cause of aerosol data jump; it was wind direction rather than rainfall that led to the anomaly in aerosol size distributions.

Figure 3 showed the time series of marine aerosol distributions and meteorological parameters during the observation shipboard in the SCS. Due to the pump of the Model 3321 APS spectrometer failing to work from 4 June to 15 June 2023 during the cruise period, flow rate could not reach the minimum standard. Thus, we only analyzed variations of observational data from 21 May to 3 June 2023. Figure 3a–b presented the trends of aerosol size distribution and comparison of the accumulation (0.5 µm ≤ Dp ≤ 1.981 µm) and coarse (1.981 µm < Dp < 10.37 µm) mode particle NCs. During the shipboard observation period, the summed (Dp < 10.37 µm) NC varied from 18.46 to 89.38 cm⁻³, NC of accumulation mode varied from 17.39 to 87.31 cm⁻³, and NC of coarse mode varied from 0.83 to 2.49 cm⁻³. The NCs exhibited substantial temporal fluctuations. The average values for summed NC, accumulation mode NC, and coarse mode NC were 54.01, 52.35, and 1.66 cm⁻³, respectively. Published observational data of marine aerosol NC relevant to this study were shown in Table 2. The shipboard observational data showed overall average values and standard deviations of maritime aerosol NCs under different temporal and geographical conditions. We used these data to compare with the marine aerosol NCs during this cruise period.

In this study, observed NC of accumulation mode was lower than the observation in the East China Sea (Lin et al., 2007; Ma et al., 2022). Aerosol emissions from the Yangtze River Delta region are higher than those from the Pearl River Delta region (Li et al., 2017). Due to the influence of aerosol transport, a greater amount of continental and anthropogenic aerosols from the Yangtze River Delta were delivered to the East China Sea compared to the amount transported from the Pearl River Delta to the South China Sea. Hence, the accumulation mode NC was significantly lower in the SCS. The SCS is one of the marginal seas of the Western Pacific. The average summed NC observed in this study (54 cm⁻³, 0.5 µm ≤ Dp ≤ 1.98 µm) was slightly lower than the NC reported for the Western Pacific (83 cm⁻³, 0.1 µm ≤ Dp ≤ 1.98 µm) by Flores et al. (2020). Notably, this study focused on particles from 0.5 to 1.98 µm, while Flores et al. (2020) included smaller particles (0.5 µm ≥ Dp ≥ 0.1 µm). Differences in the measured particle size ranges are a potential leading cause of the differences in marine aerosol NCs. Regarding marine aerosol NCs in the same ocean regions, the observations of Cai et al. (2020) and Kong et al. (2016) were significantly higher than those in this study. Although the differences in observation seasons and particle size ranges might influence the average NC observations, these differences may still indicate that the marine aerosol was affected by the continental transport and the anthropogenic activity in the offshore regions, based on the latitude and longitude.

Wet deposition through scavenging by rainfall is a critical sink for aerosols (Atlas and Giam, 1988; Radke et al., 1980). However, intense precipitation events might paradoxically elevate aerosol NCs. Some studies indicate that impaction of liquid droplets on porous surfaces (e.g. the ship surfaces) may generate aerosol particles (Bird et al., 2010; Joung and Buie, 2015; Zhou et al., 2020). By accounting for observation environment and rainfall intensity, short-duration heavy rainfall resulted in numerous raindrops impacting the ocean and ship surfaces, generating aerosol particles. Subsequently, the monitoring instrument captured some of these aerosol particles, ultimately contributing to the increased aerosol particle concentration observed in Fig. 3 (the blue-shaded regions). In addition to the elevated concentrations of marine aerosols resulting from these rain events, the aerosol NC spectrum distributions shown in Fig. 3a demonstrate continuous marine aerosol NC distributions in the size ranges of 523 to 583, 1715 to 1981, and 3786 to 4371 nm during the cruise period. This indicated background characteristics of marine aerosol particle distributions in the marine environment. During the cruise period, comparisons of the time series of NCs between the two aerosol particle modes were made, as shown in Fig. 3b. The temporal trend of the NC of accumulation mode was approximately consistent with the coarse mode. The correlation coefficient between these two modes was approximately r=0.71. However, there were some differences in the NC of the different particle sizes in aerosols caused by the different marine aerosol sources. We also found that the temporal trend of accumulation mode was more variable than that of coarse mode, suggesting the accumulation mode may be more obviously influenced by changes in the marine environment.

For the meteorological parameters, the ship remained in the northeast trade winds during the entire cruise period; therefore, this mainly led to south-westerly and southerly winds, as shown in Fig. 3c. Figure 3d represented air temperature and water temperature. The observed air temperature was consistent with MERRA-2 reanalyzed air temperature (r=0.719), whereas the average T2m and SST over the whole observation period were closer to 29.0 and 29.7 °C, respectively. RH had a negative correlation with the visibility (r=-0.74), and the average RH and VIS were equal to 83.0 % and 45.1 km. Since wind is a major driver of marine aerosol production and transport, we attempted to explain marine aerosol NCs of the accumulation mode and the coarse mode as the functions of the WS and WD (Fig. 4a, b). The RH and the rainfall intensity observations were used to aid analysis (Fig. 4c, d). High NCs (≥ 150 cm⁻³) were observed almost entirely in which the WD were between NW and N that were caused by the high RH accompanied by the rainfall events, and distributions of NCs were uniform when the wind was blowing in the other directions.

Figure 5 showed variations of NCs of two aerosol particle modes with WS, and observational data were binned at 3 m s⁻¹ WS intervals. The variations in NCs with WS observed in this study were consistent with a previous study (Bruch et al., 2023). For example, marine aerosol NCs generated by the bubble bursting process at low WS showed little variation, and the low WS was insufficient to activate spume droplet production (Pietsch et al., 2018; Russell et al., 2023). Consequently, no significant variation in NCs were observed in the 0–6 m s⁻¹ WS range. NCs increased with the increase in WS from 6 to 15 m s⁻¹; however, the increase slowed down when WS exceeded 13 m s⁻¹. Previous study proposed that this phenomenon may be linked to the scavenging of marine aerosols through collision by the larger water drops at high WS (Pant et al., 2009).

This marine scientific research campaign started southward from the harbor of Zhanjiang (21°16^′21.12^′′ N, 110°23^′45.17^′′ E) on 21 May and reached up to the southernmost (8°5^′ N) point of this cruise on 3 June. In different latitudes of the SCS, there were vastly different marine aerosol distribution characteristics, meteorological parameters, and marine aerosol transport sources. Therefore, we assessed the features of marine aerosol distribution at various distances from the coast. We conducted real-time analysis of the 72 h backward trajectories of air masses at the ship's location (Fig. 6a, b). The backward trajectory analysis indicated that the air masses had last passed over continental areas on 22 May 2023, 11:00 LT, at a point 50 km from the coast (the red solid lines in Fig. 6b). Consequently, for all sampling locations within this 50 km boundary, the air masses had directly passed over mainland areas. This meant they carried continental and anthropogenic aerosols that ultimately influenced the aerosol distributions (Braun et al., 2020; Wu and Boor, 2021). For regions more than 50 km from the coast, the backward trajectory results consistently showed that the air masses did not pass over any mainland areas before reaching the sampling site (the blue solid lines in Fig. 6a). The prevailing wind direction was primarily from the southwest (Fig. 3c) in these regions, so aerosols could not be directly transported from the continent to the ship's location. Additionally, continental and anthropogenic aerosols, which were emitted from islands and countries surrounding the SCS, lost their original characteristics through the long-duration (over 72 h) transport. These aerosols likely underwent dry deposition, wet deposition, and aging processes associated with long-range transport. Such processes could have led to the removal of continental aerosols or their gradual dilution and mixing with natural aerosols (Hodshire et al., 2019; Ohata et al., 2016; Xu et al., 2021). Over time, the continental and anthropogenic aerosols may have transformed into or been integrated with the background aerosols. Hence, 50 km from the coast was taken as the boundary distance to distinguish offshore and pelagic regions in this study. We could analyze differences in aerosol transport and production in offshore and pelagic regions. To eliminate the effects of rainfall on aerosol concentrations, we removed aerosol data associated with rainfall intensity greater than 1 mm h⁻¹ during the observation period.

Figure 7 revealed marine aerosol distribution characteristics with different modes in the SCS. From Fig. 7a, we can find that NCs of different aerosol particle modes in offshore regions showed significant differences from those in pelagic regions. The average summed NC in offshore regions during the cruise period was 108.25 cm⁻³, registering a 1.2-fold increase compared to NC in pelagic regions (49.22 cm⁻³). The NC of accumulation mode in offshore regions was 105.57 cm⁻³, and it was 47.65 cm⁻³ in pelagic regions. The average accumulation mode NC in offshore regions was 2.2 times higher than that in pelagic regions, with a statistically difference (p < 0.001). Similarly, the comparison of the coarse mode NC revealed the same result as that for the accumulation mode. The coarse mode NC in the offshore regions (2.68 cm⁻³) was also significantly higher than that in pelagic regions (1.57 cm⁻³), with a statistically significant difference (p < 0.001). Figure 7b showed the number size distribution for marine aerosols of 0.5 to 10 µm diameters in the offshore and pelagic regions, where dN/dlgDp represents the number size distribution. The comparisons in Fig. 7b showed that the number size distributions in the offshore and pelagic regions exhibited a bimodal distribution, and the peak values both occurred at 0.542 and 1.981 µm, which were consistent with the previous studies (Andronache, 2003; Braun et al., 2020). The number size distributions exhibited a relatively stable value in the 0.835–1.981 µm particle size range. Due to aerosol transport, continental air masses may have carried continental and anthropogenic aerosols, which could have ultimately affected aerosol distributions in the 0.5–5.0 µm particle size range. The number size distributions in the offshore regions were obviously higher than in the pelagic regions in the 0.5–5.0 µm particle size range. These findings were consistent with previous studies (Braun et al., 2020; Lorenzo et al., 2023). However, in the 5.0–10 µm particle size range, the number size distributions from offshore and pelagic regions were largely consistent, and this consistency remained robust against instrumentation limitations. This comparability persists despite APS measurements in this range having inherent uncertainties reaching up to 130 % (Pfeifer et al., 2016), primarily due to inefficient particle detection at concentrations approaching 1 cm⁻³. Throughout the cruise, continuous 5 s sampling yielded 64 180 valid samples, which through statistical averaging reduced measurement uncertainty to 0.5 %. At this negligible level, the distribution characteristics and cross-regional correlations are considered reliably preserved.

Figure 7c revealed a decreasing trend in NCs with increasing distance from the coast, and the correlation coefficients between the daily average NCs of accumulation mode, coarse mode, and the summed NCs and the distance from the coast were calculated as r=-0.87, -0.67, and -0.81, respectively. The correlation analysis, based on hourly average NCs of accumulation and coarse modes versus the distance from the coast, yielded r=-0.59 and -0.50 for offshore regions, and r=-0.28 and -0.33 for pelagic regions. The same was true for the summed NC; the correlation coefficient between the hourly average summed NC and the distance was -0.56 in the offshore regions and -0.29 in the pelagic regions. These correlations indicated that the correlation between distance from the coast and NC was mainly reflected in the offshore regions and had almost no significant correlation with marine aerosol in the pelagic regions. In offshore regions, aerosol NC was influenced by continental aerosol transport. This influence diminished with increasing distance from the coast. After the ship entered the pelagic regions, the influence of continental air mass transport almost disappeared. Marine aerosol NCs were lower in pelagic regions compared to offshore regions.

Table 3 revealed differences in meteorological parameters between offshore regions and pelagic regions. For example, the average WS in offshore regions was 10.74 m s⁻¹, slightly higher than that in pelagic regions (8.64 m s⁻¹). Higher WS can enhance marine aerosol production. Therefore, there was a higher production of aerosols in offshore regions. In addition, the offshore regions were relatively close to the coastline (Fig. 6b). Compared to pelagic regions, southwest and west winds in offshore regions could directly transport continental and anthropogenic aerosols to the ship's location from Guangdong and Hainan, China. Therefore, aerosol transport was higher in offshore regions. Figure 7d indicated a difference in the distribution of marine aerosol composition between offshore and pelagic regions. In the offshore regions, the proportions of dust (DUST₁₀; Dp ≤ 10 µm) and sulfate aerosols (SO42-) were 5.04 % and 1.36 %, which were higher than those in the pelagic regions (1.45 % and 0.97 %, respectively). The higher concentrations of dust and sulfate aerosols may further indicate that continental aerosols have influenced the aerosol composition in the offshore regions (Geng et al., 2023; VanCuren, 2003). In the pelagic regions, the proportions of dimethyl sulfide (DMS), organic carbon (OC), and sulfur dioxide (SO₂) were 0.15 %, 1.2 %, and 1.7 %. These proportions were higher than those in the offshore regions (0.1 %, 0.84 %, and 0.56 %, respectively) due to the more frequent marine biological activities (e.g. phytoplankton metabolism) in the pelagic environments. For instance, phytoplankton releases DMS through cellular metabolism and cellular lysis (Saliba et al., 2020); DMS then undergoes atmospheric oxidation to form SO₂ (Kettle and Andreae, 2000). Additionally, phytoplankton produces OC (O'Dowd et al., 2002). These marine biological activities directly contribute to higher proportions of DMS, SO₂, and OC in pelagic regions. The proportion of sea salt aerosol (SEAS₁₀; Dp ≤ 10 µm) in the pelagic regions (94.33 %) was higher than that in the offshore regions (91.9 %), indicating a significant contribution of SEAS₁₀ to the total marine aerosols in the pelagic environments. To sum up, our results indicated that the distance from the coast had a great influence on marine aerosol production and transport. It ultimately led to obvious differences between size distributions of marine aerosols in the offshore and pelagic regions.

The results of Sect. 3.2 showed that distributions and composition of marine aerosols in different regions were influenced by marine production and continental transport. Among them, the degree of impact on marine transport and background aerosols caused by continental transport and marine production differs greatly due to differences in the degree of continental transport and marine biological activities at different distances from the coast. Beyond that, many meteorological parameters, such as WS, T2m, SST, and SST-T2m, might influence concentration and distribution of marine aerosols. It is expected that there are diurnal differences. Therefore, we analyzed the diurnal variations of NCs and the correlations between NCs and different meteorological parameters in this section. In pelagic regions, the sources of 72 h backward trajectory air masses were from the ocean, and observational data were processed to exclude the continental influence. These aerosol data conformed to clean marine periods, which were proposed by Saliba et al. (2019) to extract relatively clean marine aerosol data; these NCs thus distinctly represented characteristics of local production of marine aerosol in these regions.

Figure 8a showed diurnal variations of the mean values of summed NCs in the pelagic region. To better compare the diurnal variation, we divided the data into different periods according to sunrise and sunset times. Consequently, we selected the sunrise and sunset moments, along with the surrounding one-hour interval, as the transition periods to eliminate the effects of day-to-night transitions. Figure 8a showed a clear diurnal variation trend. For accumulation mode, the NCs remained falling trend in the nighttime (00:00–03:00 and 21:00–23:00 LT), and they began steadily rising in the night-to-day transition (NDT) period (04:00–06:00 LT). Then they showed a slight upward trend during the daytime (07:00–17:00 LT), but began to fall steadily during the day-to-night transition (DNT) period (18:00–20:00 LT). We also found that NCs exhibited apparent differences between the daytime and nighttime, and tended to increase or decrease significantly during the transition periods.

Comparisons of size distributions (Fig. 8b) showed that number size distributions exhibit a relatively stable value in the 0.835–1.981 µm particle size range, and subtle differences emerged in this particle range. Quantitatively, peak diameter varied slightly across periods: 0.571 µm in nighttime, 0.567 µm in the NDT period, 0.569 µm in daytime, and 0.570 µm in the DNT period. More notably, the peak value was 147.05 cm⁻³ in nighttime, then rose to 155.87 cm⁻³ in the NDT period, further increased to 165.60 cm⁻³ in daytime, and reached the highest value of 206.79 cm⁻³ in the DNT period, registering a 0.4-fold increase relative to the nighttime baseline. The peak value showed a clear and continuous increasing trend, which may reveal variations in aerosol production. In addition, all size distributions for marine aerosols had the same shape. The consistent shape can be explained by their common marine origin and production mechanisms. The NCs of different aerosol particle modes in different periods were counted in Fig. 8c, and we can find that the average summed NCs were significantly different in different periods. Specifically, the summed NC was 44.03 cm⁻³ in the nighttime, and 47.5 cm⁻³ in the first transition period (i.e. NDT), but it was 49.28 cm⁻³ in the daytime, and reached the highest (59.59 cm⁻³) in the second transition period (i.e. DNT). The differences were all statistically significant (p < 0.01). The comparisons of the accumulation modes were consistent with those of the summed NCs. The NC of the accumulation mode was 42.43 cm⁻³ in the nighttime, 45.90 cm⁻³ in the NDT period, 47.62 cm⁻³ in the daytime, and 57.91 cm⁻³ in the DNT period. The differences were also statistically significant (p < 0.01). There were pronounced diurnal NC variations, with maximum differences of 35 % observed between daytime, nighttime, and transition periods. However, there were no significant differences in the NCs of the coarse modes between different periods, as mentioned above, which were 1.60, 1.58, 1.66, and 1.68 cm⁻³, respectively.

To explore the reasons for these differences, we further analyzed the correlation coefficients between the NCs and the meteorological parameters (Fig. 9a). The correlation coefficients were 0.41 for the coarse mode, 0.57 for the accumulation mode, and 0.58 for summed NCs. This correlation suggested that the NCs of all aerosol particle modes have positive correlations with the WSs. By contrast, negative correlations between SST and NC were found, which were -0.24, -0.45, and -0.47, respectively, and the NCs were low in the time periods with high SST. Compared to the WS and SST, our results showed that SST-T2m has a negative correlation with NCs, with correlation coefficients of -0.30, -0.82, and -0.83. Figure 9b showed that WS was lower during nighttime, and SST and SST-T2m values were higher, resulting in NCs lower than those in daytime. In the NDT period, WS did not change significantly, but an obvious decrease in the SST and SST-T2m was found, ultimately resulting in a noticeable increase in NCs. In this period, SST and SST-T2m were the dominant factors. However, in the DNT period, NCs were highest due to the lowest values of SST and SST-T2m and the highest WS. In this period, the significant reduction in the WS led to a decrease in NCs. From the above analysis, the meteorological parameters have a joint impact on the production and distribution of marine aerosols.