Sea surface salinity (SSS) and temperature (SST) can serve as proxies to detect freshwater fluxes at the sea surface during precipitation. Widely unknown, however, is the impact of precipitation (droplet sizes and velocities) and wind speed on the sea surface microlayer (SML), the <1 mm boundary layer between atmosphere and ocean. We used the autonomous surface vehicle HALOBATES in the German Bight, North Sea, to collect SSS and SST data with high vertical resolution at 7 depths in the top meter, spaced 10–30 cm apart, including the SML. During two rain events with similar maximum rainfall intensity, the magnitude and persistence of the resulting salinity anomaly (between SML and 100 cm depth) was dependent on wind speed. The maximum rate of change of salinity in the SML was 4 times faster with high wind speeds: 0.009 g kg min−1 at low wind speeds, 0.037 g kg min−1 at high wind speeds. A threshold of about 5 m s−1 determined whether stratification occurred in the near-surface layer or freshwater mixed with the underlying layers during precipitation. Strong winds still caused salinity changes in the near-surface layer and SML, but the water mixed rapidly with the underlying water masses. The SML salinity was affected by droplet distributions with smaller droplets, as small droplets stay at the surface. Salinity and temperature changes in the SML were >9 times higher than at 100 cm depth (−0.037 g kg min−1 vs. −0.004 g kg min−1) and still detectable during very high wind speeds. Overall, these results contribute to a better understanding of the vertical distribution of freshwater in the surface ocean and its dependence on rain intensity, wind speed, and droplet properties, helping to improve understanding of the fate of freshwater in a changing ocean due to climate warming.
1. Introduction
In recent decades, measurements of sea surface salinity (SSS) and temperature (SST) have been of great relevance as key indicators of climate variability, ocean circulation, water masses, and the hydrological cycle (Durack, 2015; Schmitt, 2018). The global hydrological cycle determines the distribution of freshwater, the most crucial life resource, and is strongly affected by climate change (Zhang et al., 2007; Ingram, 2016). This cycle is mainly driven by evaporation, precipitation, and freshwater fluxes (i.e., evaporation minus precipitation). Extreme weather phenomena occur and, in regions where precipitation is predominant, the duration and intensity of these events are increasing (Allen and Ingram, 2002; Held and Soden, 2006; Collins et al., 2013; Durack, 2015). Global warming causes an exponential increase in the atmospheric capacity to hold water vapor; consequently, freshwater fluxes are altered (Manabe and Wetherald, 1980; Trenberth et al., 2003; Durack, 2015). As the ocean is the largest water resource on Earth, about 80% of the evaporation and precipitation of the global water cycle takes place over the ocean (Baumgartner and Reichel, 1975). In this context, precipitation lowers the SSS and SST, as rainfall temperature is generally 1–2°C cooler than air temperature (Gosnell et al., 1995). Conversely, evaporation leads to an increase in SSS and a decrease in SST due to the removal of latent heat (Katsaros, 1980).
These changes, induced by precipitation and evaporation, occur first in the skin layer of the ocean. This boundary layer between the atmosphere and the ocean, often called the sea surface microlayer (SML), is less than 1 mm thick. Freshwater can accumulate on the surface and build up freshwater lenses in the upper few meters with very pronounced SST and SSS anomalies (Wijesekera et al., 1999; Reverdin et al., 2012; Drushka et al., 2016; Iyer and Drushka, 2021b), but they have also been observed at depths less than 1 m (Wurl et al., 2019; Gassen et al., 2023). Freshwater lenses are typically detected at depths of several meters using ship-based measurements, moorings or Argo-floats and in the upper few centimeters using remotely sensed data (Boutin et al., 2016). Therefore, measurements of SSS and SST at the skin layer and the near-surface layer (NSL; i.e., less than 1 m depth) are scarce because ship-based measurements are challenging. These challenges may lead to underestimates of SSS and SST anomalies in the sea surface microlayer. The project called Salinity Processes in the Upper-Ocean Regional Study Part 2 (SPURS-2) aimed to understand atmosphere-ocean fluxes with associated salinity changes in the eastern tropical Pacific, a precipitation-dominated region (Lindstrom et al., 2015; Bingham et al., 2019). During SPURS-2, Drushka et al. (2019) used a surface salinity profiler to measure SSS and SST anomalies at depths of 0.05–1.10 m and found that accumulated precipitation and wind speed significantly influenced freshwater lens formation, but not at wind speeds higher than 7 m s−1. In the Indian Ocean Moulin et al. (2021) observed that the majority of freshwater lenses were present at wind speeds of less than 5 m s−1, with no lenses present at wind speeds over 10 m s−1.
There is a need to understand the thermohaline features of the NSL, including the SML, beyond the previous measurements in tropical regions. Here we provide the first vertically resolved data on thermohaline features in the NSL during precipitation events in the North Sea. SSS and SST anomalies in the SML and NSL are addressed along with data on wind speed, rain intensity, and droplet properties, such as droplet sizes and fall velocities. We aim to better understand freshwater fluxes and how precipitation determines SSS and SST anomalies in the SML and NSL at mid-latitudes.
2. Materials and methods
Data from SSS and SST were collected in the North Sea during expeditions with the research vessel Heincke for the project Sea-surface variability of essential Climate variables in the North Sea (SCANS; Figure 1). We collected a complete seasonal dataset of SSS and SST to improve our understanding of their variability in the SML and NSL. The SSS (computed from conductivity and temperature according to TEOS10; McDougall and Barker, 2011) and SST data were collected using the autonomous surface vehicle HALOBATES. Density (σT) was calculated from the determined temperature and salinity minus 1000 kg m−3. We recorded SSS and SST in two expeditions during 11 precipitation events at six stations. During the first expedition HE598 (May 4–23, 2022), precipitation occurred at Station 8 on May 10, 2022, and at Station 4 on May 19, 2022. The second SCANS expedition HE609 (October 4–26, 2022) took place with precipitation events on October 10, 2022 (Station 7), October 17, 2022 (Station 7), October 21, 2022 (Station 12), and October 22, 2022 (Station 2). HALOBATES has six rotating glass discs on its front, partially immersed in water and skimming the SML with a thickness of about 80 µm (Shinki et al., 2012; Ribas-Ribas et al., 2017; Wurl et al., 2024).
HALOBATES measures SSS and SST in the NSL with tubes fixed at depths of 30, 40, 50, 60, 85, and 100 cm. Seawater is pumped through conductivity, temperature, and depth (CTD) sensors (Ocean Seven 310, Idronaut, Italy), with an accuracy of 0.0015 mS cm−1 and 0.0015°C for conductivity and temperature, respectively, according to the manufacturer’s data sheet. Salinity measurements were corrected with a discrete water sample taken at every deployment. HALOBATES can pump seawater for all CTDs at the same depth to correct possible biases between the CTDs. Due to technical issues, not all CTDs were operational at every station. At higher sea states (e.g., October 10, 2022, at Station 7), spikes occurred due to air bubbles and some spikes were removed later, with a salinity threshold defined as the difference of 0.015 g kg−1 from the previous value. Irregularly occurring air bubbles still caused a decrease in salinity, especially at a depth of 30 cm (see Discussion section). Two weather stations were installed on HALOBATES (Clima Sensor US and DLU data logger, Thies Clima, Germany; MetSENS550 and CR100X data logger, Campbell Scientific Ltd., UK). HALOBATES has the capability of autonomous operation—that is, following a course of pre-defined waypoints (Wurl et al., 2024). However, we decided to leave the surface vehicle drifting for this study to reduce spatial variations in the measurements, as experiments showed that, while drifting, HALOBATES stays in the same water mass and has a nearly Lagrangian drifting behavior.
At the upper deck of the RV Heincke, an optical laser disdrometer (Thies Clima, 5.4110.xx.x00, Germany) and an acoustic disdrometer (IAV Technologies SARL, RainFlow RF4, Switzerland) were installed to measure rainfall intensity, total accumulated rain, and drop size distribution. Precipitation droplets of the optical disdrometer were categorized into 22 sizes and 20 velocities according to the manufacturer’s default settings (0.16 to >8 mm Ø and 0.2 to 20 m s−1). During the observations, the ship remained at a distance (a few hundred meters) from HALOBATES to avoid disturbing the measurements.
The fractional change in salinity was calculated from salinity at the beginning of the decrease (Sbegin) and salinity at the end of the decrease (Send). Rainwater is expected to have a salinity of 0 (Srain; Katsaros and Buettner, 1969).
The amount of freshwater in milliliters on the surface was calculated from the SML salinity at the beginning (Sbegin) and at the end of the slope (Send) as follows (Schlüssel et al., 1997):
The monotonic relationship between two variables based on the rank of the data was analyzed with the Spearman correlation. The results were considered significant when p ≤ 0.05, with a 95% confidence level.
3. Results and discussion
3.1. Salinity, temperature, and density changes during precipitation
Precipitation occurred at six stations during the HALOBATES deployment. The mean rain intensity, total accumulated rain, wind speed, and air temperature are summarized in Table 1. Data from two selected stations of three rain events with the most intensive precipitation are shown as examples in Figures 2 and 3. Rain events with a maximum intensity of approximately 10 mm h−1 occurred at the two stations. However, weather conditions, especially wind speed and raindrop distributions, varied considerably, resulting in a varied distribution of freshwater across the SML and NSL (Figures 2 and 3, Table 1).
Station, Date (2022) . | Rain Event . | Time (min) . | Mean Rain Intensitya (mm h−1) . | Mean Rain Intensity (mm min−1) . | Total Accumulated Rain (mm) . | Mean Wind Speeda (m s−1) . | Mean Air Temperaturea (°C) . | SML Salinity Change (g kg min−1) . | SML Temperature Change (°C min−1) . | Max. Salinity Anomalyb (g kg−1) . | Max. Temperature Anomalyb (°C) . |
---|---|---|---|---|---|---|---|---|---|---|---|
Station 8, May 10 | 1. | 29 | 0.11 ± 0.13 | 0.002 ± 0.002 | 0.06 | 5.55 ± 0.92 | 12.68 ± 0.09 | −0.024 | −0.037 | —c | −1.263 |
2. | 120 | 0.50 ± 0.91 | 0.008 ± 0.015 | 1.00 | 5.16 ± 0.86 | 12.28 ± 0.15 | −0.006 | −0.001 | —c | −0.564 | |
3. | 38 | 1.28 ± 0.49 | 0.021 ± 0.008 | 1.21 | 5.56 ± 1.01 | 12.63 ± 0.15 | −0.008 | −0.002 | 0.033 | −0.043 | |
4. | 55 | 0.04 ± 0.01 | 0.001 ± 0.000 | 0.04 | 6.17 ± 0.80 | 12.73 ± 0.16 | −0.005 | −0.001 | 0.034 | −0.027 | |
Station 4, May 19 | 1. | 147 | 1.50 ± 1.64 | 0.025 ± 0.027 | 3.92 | 3.71 ± 1.53 | 13.41 ± 0.13 | −0.009 | −0.010 | −0.300 | −0.035 |
2. | 91 | 0.42 ± 0.39 | 0.007 ± 0.006 | 0.71 | 2.87 ± 1.23 | 13.31 ± 0.13 | −0.002 | −0.001 | −0.018 | −0.016 | |
Station 7, Oct 10 | 1. | 94 | 0.54 ± 1.33 | 0.009 ± 0.022 | 0.97 | 9.89 ± 1.12 | 13.56 ± 0.67 | −0.037 | −0.007 | −0.139 | −0.126 |
Station 7, Oct 17 | 1. | 68 | 0.78 ± 0.58 | 0.013 ± 0.010 | 0.87 | 5.86 ± 1.68 | 16.08 ± 0.30 | −0.025 | −0.001 | −0.058 | −0.014 |
Station 12, Oct 21 | 1. | 42 | 0.09 ± 0.11 | 0.001 ± 0.002 | 0.06 | 5.36 ± 0.84 | 13.20 ± 0.02 | −0.001 | −0.006 | −0.006 | −0.077 |
2. | 14 | 0.06 ± 0.09 | 0.001 ± 0.001 | 0.02 | 2.86 ± 0.49 | 13.78 ± 0.06 | −0.001 | —c | −0.003 | −0.042 | |
Station 2, Oct 22 | 1. | 10 | 0.21 ± 0.22 | 0.003 ± 0.004 | 0.04 | 7.24 ± 0.56 | 15.02 ± 0.04 | −0.033 | −0.009 | −0.162 | −0.069 |
Station, Date (2022) . | Rain Event . | Time (min) . | Mean Rain Intensitya (mm h−1) . | Mean Rain Intensity (mm min−1) . | Total Accumulated Rain (mm) . | Mean Wind Speeda (m s−1) . | Mean Air Temperaturea (°C) . | SML Salinity Change (g kg min−1) . | SML Temperature Change (°C min−1) . | Max. Salinity Anomalyb (g kg−1) . | Max. Temperature Anomalyb (°C) . |
---|---|---|---|---|---|---|---|---|---|---|---|
Station 8, May 10 | 1. | 29 | 0.11 ± 0.13 | 0.002 ± 0.002 | 0.06 | 5.55 ± 0.92 | 12.68 ± 0.09 | −0.024 | −0.037 | —c | −1.263 |
2. | 120 | 0.50 ± 0.91 | 0.008 ± 0.015 | 1.00 | 5.16 ± 0.86 | 12.28 ± 0.15 | −0.006 | −0.001 | —c | −0.564 | |
3. | 38 | 1.28 ± 0.49 | 0.021 ± 0.008 | 1.21 | 5.56 ± 1.01 | 12.63 ± 0.15 | −0.008 | −0.002 | 0.033 | −0.043 | |
4. | 55 | 0.04 ± 0.01 | 0.001 ± 0.000 | 0.04 | 6.17 ± 0.80 | 12.73 ± 0.16 | −0.005 | −0.001 | 0.034 | −0.027 | |
Station 4, May 19 | 1. | 147 | 1.50 ± 1.64 | 0.025 ± 0.027 | 3.92 | 3.71 ± 1.53 | 13.41 ± 0.13 | −0.009 | −0.010 | −0.300 | −0.035 |
2. | 91 | 0.42 ± 0.39 | 0.007 ± 0.006 | 0.71 | 2.87 ± 1.23 | 13.31 ± 0.13 | −0.002 | −0.001 | −0.018 | −0.016 | |
Station 7, Oct 10 | 1. | 94 | 0.54 ± 1.33 | 0.009 ± 0.022 | 0.97 | 9.89 ± 1.12 | 13.56 ± 0.67 | −0.037 | −0.007 | −0.139 | −0.126 |
Station 7, Oct 17 | 1. | 68 | 0.78 ± 0.58 | 0.013 ± 0.010 | 0.87 | 5.86 ± 1.68 | 16.08 ± 0.30 | −0.025 | −0.001 | −0.058 | −0.014 |
Station 12, Oct 21 | 1. | 42 | 0.09 ± 0.11 | 0.001 ± 0.002 | 0.06 | 5.36 ± 0.84 | 13.20 ± 0.02 | −0.001 | −0.006 | −0.006 | −0.077 |
2. | 14 | 0.06 ± 0.09 | 0.001 ± 0.001 | 0.02 | 2.86 ± 0.49 | 13.78 ± 0.06 | −0.001 | —c | −0.003 | −0.042 | |
Station 2, Oct 22 | 1. | 10 | 0.21 ± 0.22 | 0.003 ± 0.004 | 0.04 | 7.24 ± 0.56 | 15.02 ± 0.04 | −0.033 | −0.009 | −0.162 | −0.069 |
aThe standard deviation (square root of the variance of the mean) is shown for rain intensity, wind speed, and air temperature.
bAnomaly = SML − 100 cm, with negative anomalies indicating a higher salinity at 100 cm depth.
cData not available.
Figure 2 shows the salinity, temperature, and σT on May 19, 2022, at Station 4 when two rain events occurred. Temperature and conductivity were measured in the SML and at 30, 40, and 100 cm depths. In the first rain event (06:45–09:31 UTC), we observed precipitation with a maximum rain intensity of 9.57 mm h−1 and a total accumulated rain of 4.16 mm (Figure 2). From 06:45 to 07:42 UTC, the mean wind speed was 2.28 m s−1. At this time, salinity reached maximum anomalies (compared with before precipitation) of −0.301 g kg−1 in the SML, and −0.153 g kg−1 at a 40 cm depth. Even at a depth of 100 cm, the salinity anomaly was −0.204 g kg−1. Near-surface stratification occurred during lower wind speed regimes. From 07:42 to 08:02 UTC, the wind speed exceeded 5 m s−1 and removed the stratification, but salinity in the SML remained slightly lower than at other depths. During this time, the mean wind speed was 4.54 m s−1, and the mean salinity anomaly (SML − 100 cm) was −0.040 g kg−1. Between 08:04 and 08:53 UTC, wind speed decreased to a mean of 3.98 m s−1, and freshwater did not mix rapidly with the underlying water masses, as indicated by a mean SML anomaly (SML − 100 cm) of −0.015 g kg−1. From 08:54 UTC to the end of this rainfall event (09:31 UTC), the wind speed increased to a mean of 5.06 m s−1, and SML anomalies occasionally turned positive due to the mixing of saltier water from below (mean SML salinity anomaly of −0.001 g kg−1).
At higher wind speeds, the SML anomalies turned positive, and the freshwater lens on the surface was mixed with the underlying water masses. Our results confirmed the observations by Iyer and Drushka (2021a), who found that freshwater lenses could occur for hours with wind speeds below 5 m s−1 and mix immediately with the underlying layers at higher wind speeds. Thompson et al. (2019) found a threshold of 8 m s−1 in the tropics; however, because rainfall is typically less intense in the North Sea, fresh lenses do not persist at wind speeds >5 m s−1 in this region. σT of the first rainfall event followed the salinity pattern and showed a decrease of −0.222 g L−1 in the density at the SML (Figure 2).
Unlike salinity and σT, temperature did not show a clear signal during the first precipitation event. Although precipitation temperature in the tropics is known to be colder than air temperature (Gosnell et al., 1995), very little is known about the difference between surface temperature and rain temperature in the mid-latitudes. We observed only minor temperature anomalies. The maximum negative temperature anomalies were −0.034°C in the SML, −0.026°C at a 30 cm depth, and −0.014°C at a 40 cm depth (all in reference to 100 cm). Air temperatures during the first precipitation event were only slightly lower (mean 13.18°C) than surface temperatures (mean of all depths: 13.41°C). The low differences between the air and surface temperatures explain why precipitation did not significantly affect the SST. At the end of the precipitation, the intensity decreased below 1 mm h−1, and the air temperature increased. At the same time, the SST of all depths down to 100 cm increased, and the temperature of the SML changed from the lowest (13.13°C) to the highest temperature (13.27°C) compared with the other depths. After the first precipitation event, the air temperature increased further, and the surface was well stratified, with the highest temperatures in the SML. For example, between 09:03 and 13:43 UTC, the air temperature increased by 0.8°C, which caused an increase in SML temperature by 0.401°C and 0.282°C at a 100 cm depth.
Lower rainfall intensities occurred during the second precipitation event at Station 4. In the beginning, wind speeds occasionally exceeded 5 m s−1. After 13:55 UTC, average wind speeds remained below 5 m s−1, and salinity anomalies (SML − 100 cm) of −0.018 g kg−1 occurred in the SML. Seawater temperature at all depths was not affected by precipitation, suggesting that rain temperature was not significantly different from surface temperature.
Figure 3 shows the salinity, temperature, and σT changes at Station 7 on October 10, 2022. The weather variables were very different from the rain events in May (Figure 2). Between 13:00 and 13:20 UTC, the mean wind speed was between 8 and 10 m s−1. At the beginning of the precipitation event at 13:30 UTC, the surface water masses were mixed, as indicated by an initially homogeneous salinity. Salinity slightly decreased with low rain intensities at the beginning of the precipitation event. At 14:12 UTC, rain intensity increased until it reached a maximum intensity of 10.57 mm h−1 at 14:15 UTC. Three minutes later, the SML salinity reached its maximal anomaly relative to 100 cm of −0.139 g kg−1. At the other depths, salinity decreased by less than 0.02 g kg−1. Temperature also decreased during the rain event and followed the air temperature pattern. When rain intensity reached its maximum, the air temperature dropped by 1.10°C, causing a decrease in surface temperatures, which were most pronounced in the SML at 0.21°C. The surface temperature (i.e., SML) remained lower than the other depths afterward. When the wind speed reached the maximum of 15 m s−1, the underlying water was mixed, and the SML temperature anomaly was the lowest at −0.126°C. The lower temperature in the SML is most likely caused by evaporation fluxes which are most pronounced in the SML. The high wind speeds cause the removal of latent heat from the surface and promote cooling (Katsaros, 1980). The temperature and salinity at 60 cm depth was lowest before the rain event, which was probably caused by shear flows close to the surface.
The influence of wind speed, rain intensity, and accumulated total rain on the SML salinity of all rain events at all stations (Table 1) was analyzed (see Figure 4). A significant negative correlation between maximum SML salinity anomalies and maximum rainfall intensity was only found at wind speeds less than 5 m s−1 (R = −1, p = 0.044). The relationships between maximum SML salinity anomalies and mean rainfall intensity and total accumulated rainfall were also linear, but not significant. Significant correlations and linear relationships were only observed at wind speeds below 5 m s−1. This finding confirms that a threshold value of about 5 m s−1 determines whether a linear relationship between the SML salinity anomalies and rainfall intensity occurs. The correlation analysis was performed with all data points from Table 1. However, only 3 data points were related to wind speeds of less than 5 m s−1, which emphasizes the need for more field data at mid-latitudes to obtain a mechanistic understanding of the relationship between rainfall characteristics and salinity anomalies in the SML.
3.2. Dynamics of salinity and temperature changes due to precipitation
To emphasize the different dynamics of freshwater distribution at different depths, the rates (changes in relation to time) of the salinity and temperature decrease were calculated (Figure 5). At Station 4, with a mean rainfall intensity of 1.50 mm h−1 and total accumulated rain of 3.92 mm (Table 1), the SML had the most significant negative change in salinity of −0.009 g kg min−1 among all depths, followed by −0.006 g kg min−1 at a depth of 30 cm, −0.005 g kg min−1 at a 40 cm depth, and −0.004 g kg min−1 at a 100 cm depth (Figure 5a). These results show that salinity in the SML decreased more than twice as rapidly as at the 100 cm depth. The most rapid change in salinity of the SML was observed 2 min after the occurrence of the highest rain intensity of 9.57 mm h−1 (Figure 2). After 86 min, salinity at all depths reached the same value of 32.320 g kg−1 at 09:02 UTC and was not further influenced by precipitation.
The temperature dropped at the same rate of −0.010°C min−1 at all depths and at the same time as salinity. The lowest temperatures occurred in the SML due to evaporative cooling and the highest at 100 cm before and after the temperature decrease. During that time, the air temperature decreased with a similar slope of −0.015°C min−1.
Figure 5c and d shows the rates of change for salinity and temperature, respectively, during the second rainfall event at Station 4. The mean rainfall intensity and mean total accumulated rain were, respectively, with 0.42 mm h−1 and 0.71 mm lower than during the first rain event (see Table 1). At 30 cm the salinity dropped intermittently before the salinity at the other depths decreased. These drops were most likely caused by air bubbles in the sensor caused by the movements of HALOBATES. After 14 min, the salinity of the SML decreased at a rate of −0.002 g kg min−1. The salinity at 40 cm depth decreased at −0.001 g kg min−1, while the rate at 100 cm was unaffected. The temperature decrease in the second rainfall event (Figure 5d) was 10 times slower than in the first one (Figure 5b), at a rate of −0.001°C min−1. Compared to the first event, the air temperature was constant but 0.16°C lower than the surface temperature.
The salinity changing rates from the rain event at Station 7 on October 10, 2022, are shown in Figure 5e. The mean rain intensity was 0.54 mm h−1, and the total amount was 0.97 mm (Table 1). The SML showed the most rapid decrease of −0.037 g kg min−1 (14:07 to 14:11 UTC). At depths of 50, 60, and 100 cm, salinity decreased by −0.004 g kg min−1, −0.007 g kg min−1, and −0.004 g kg min−1, respectively. At 14:15 UTC, rain intensity reached its maximum at 10.57 mm h−1, and 15 min later, a maximum wind speed of 15 m s−1 was observed. Compared with the changing rates of the first rain event at Station 4, the SML salinity changes occured more rapidly at Station 7. Despite the strong wind speeds at Station 7, the rate of decrease in salinity in the SML was more than nine times higher than at 100 cm depth. The high freshwater input at the surface and probably a droplet distribution with smaller sizes and lower impact velocities (Section 3.3) resulted in the freshwater having the greatest influence on the SML. At Station 7, the decrease in salinity in the SML during the rain event was four times faster than in the SML at Station 4. Recovery to a non-stratified NSL with the same salinity at all depths and without anomalies took 86 min at Station 4. The recovery time after the rain event at Station 7 was 10 min. The SML temperature slope at Station 7 of −0.007°C min−1 differed greatly from those at all other depths (−0.002°C min−1; Figure 5f). A high average wind speed of 9.5 m s−1 presumably caused a cooling of the SML due to the loss of latent heat (Katsaros, 1980; Cronin et al., 2019).
The mean wind speed of the rain event at Station 7 was more than twice as high as the wind speed of the event at Station 4. After precipitation, ongoing mixing caused a quick recovery to the initial condition—that is, dispersion of the freshwater lens. Drushka et al. (2019) observed that temperature and salinity anomalies (0.05–1.10 m) showed no dependence on precipitation, with wind speeds exceeding 7 m s−1. Iyer and Drushka (2021a) reported that freshwater lenses quickly disperse at wind speeds exceeding 8 m s−1. Measurements by their sea surface profiler were close to the sea surface but did not measure the SML. Thus, changes between the ocean-atmosphere layer and a 0.05 m depth could have been underestimated, as indicated by earlier laboratory (see figure 1 in Katsaros and Buettner, 1969) and theoretical studies (see figure 3 in Schlüssel et al., 1997). Moulin et al. (2021) analyzed 28 freshwater lenses based on remote sensing and found most of them (88%) at wind speeds of <5 m s−1. Our results confirm the existence and the quick dispersion of very shallow (<30 cm) freshwater lenses at high sea states (Figure 5c), which are undetectable by remote sensing techniques.
3.3. Droplet properties at the greatest rates of salinity changes
Droplet properties were measured during the rainfall events using a laser precipitation monitor, which categorized the droplets into 22 sizes and 20 falling velocities. The droplet sizes and velocities of the first and second rainfall event during the time with the greatest rates of salinity change at Station 4 on May 19, 2022, are shown in Figures 6 and 7, respectively, as binned scatter plots. A binned scatter plot for the rain event at Station 7 on October 10, 2022, is shown in Figure 8.
Figure 6 shows that most droplets were counted at Station 4 on May 19, 2022, during the first rain event, as it had the longest period of decreasing salinity at 38 min compared to the other rain events (see also Figure 5). Most droplets could be classified into small diameters of 0.125–1.000 mm and low falling velocities of 1.0–1.8 m s−1. Additionally some droplets with larger diameters and higher speeds, up to 5 mm and 10 m s−1, respectively, were counted and caused a rapid change in salinity at depths of 30–100 cm (Figures 2 and 5a).
The droplet plot for the second rain event at Station 4 on May 19, 2022, shows that smaller droplets fell on the sea surface during the 12 min time when SML salinity dropped the most (Figure 5c). During this period, the rain properties were characterized by smaller droplet diameters and slower velocities than during the first rain event (Figure 7). The precipitation intensity of the second event was a third of that of the first event, with the lower impact of the smaller droplets and lower droplet velocities explaining why only the SML salinity was noticeably affected by the precipitation between 14:37 and 14:48 UTC.
The size and falling velocity of the raindrops during precipitation determine their penetration depth. In general smaller droplets have a lower impact velocity and penetrate less. If small enough, they can accumulate on the SML and do not break through it (Katsaros and Buettner, 1969; Schlüssel et al., 1997). Larger droplets have a higher vertical velocity and can break through the surface. As the size of the droplets increases, the final velocity and momentum increase (Rodriguez and Mesler, 1988). In previous studies we conducted rainfall experiments in a mesocosm tank with different droplet properties and showed that the SML salinity is more affected by droplets with a higher kinetic energy, that is, higher size and velocity (Gassen et al., 2024c). The first rain event at Station 4 and the rain event at Station 7 had similar maximum rain intensities of about 10 mm h−1 but showed different rain droplet distributions during the greatest rates of salinity change in the SML (Figures 6 and 8). At Station 4, the rain event was characterized by many droplets of smaller size (mean droplet size of 0.30 mm in diameter), but also by some large droplets with a high vertical velocity of up to 10 m s−1. At Station 7, the maximum falling velocities of the droplets were similar but contained no droplets larger than 2.5 mm (mean droplet size of 0.53 mm in diameter). The occurrence of smaller droplets explain why, despite wind speeds of up to 15 m s−1, distinct thermohaline features with much higher salinity anomalies in the SML compared to the NSL were observed. The thermohaline features were still detectable in the surface mixed layer despite high wind speeds (refer to Figures 3 and 5e), indicating that the rate of freshwater input was higher than the mixing rate with the underlying water.
3.4. Fractional change in salinity due to precipitation
Katsaros and Buettner (1969) conducted rain experiments with tanks filled with seawater and calculated the fractional change in salinity (see Methods section) to show the salinity changes caused by rain at different depths. Figure 9 shows the fractional change in salinity of the three rain events described in the previous sections. The fractional change in salinity was most pronounced in the SML with higher rain intensities, as shown in Figures 2, 3, and 5. Compared with the results of Katsaros and Buettner (1969), our fractional change values were relatively high, that is, lower salinity changes (the highest value of 1 indicates no change in salinity). They conducted their experiments with two different rainfall intensities (4.2 mm h−1 and 17 mm h−1) and droplet sizes (1.2 mm and 3 mm) and observed lower fractional changes of up to 0.3 at the lower rain intensity and smaller droplet size, but they did not include any mixing processes in their experiments.
Schlüssel et al. (1997) described a surface renewal model to investigate the influence of precipitation in the SML. They calculated the fraction of volume that did not submerge into deeper ocean layers as a function of rain intensity with a span of 0–50 mm h−1. The Schlüssel et al. (1997) model predicts a substantial fraction of freshwater that remains on the surface during low rain intensities. Only the fraction of the first rain event at Station 4 nearly agreed with the model, which predicted a fraction of 45% for an intensity of 1.50 mm h−1 for smaller droplets of <1 mm in diameter. The fraction calculated for the first rain event at Station 4 was 51.8%, and the event was characterized by relatively low wind speeds (i.e., less surface layer mixing). Wurl et al. (2019) observed a fraction of about 7.7% freshwater remaining on the surface during a higher-intensity rain event of about 47 mm h−1. Compared with the calculations of Schlüssel et al. (1997), the difference was only 2.7%. With higher rain intensity, the proportion of freshwater staying at the surface decreases (Schlüssel et al., 1997).
The approaches of Katsaros and Buettner (1969) and Schlüssel et al. (1997) did not include mixing processes due to wind. Furthermore, our results show field data and emphasize that in situ conditions are more complex. Some extensive studies have been done during the SPURS-1 and SPURS-2 campaigns, but they addressed lower latitude regions in the central Atlantic and Southern Pacific (Lindstrom et al., 2015; Bingham et al., 2019). More data are needed on rainfall events from the mid-latitudes to better compare and understand how much impact wind has on freshwater fluxes at the ocean-atmosphere boundary layer.
4. Conclusions
We have provided novel data on the thermohaline features in the NSL during precipitation events in the German Bight, including the SML as the upper boundary layer. This study is the first to describe freshwater lenses in the North Sea. A wind speed threshold of 5 m s−1 appears to be critical in determining whether a freshwater lens with distinct stratification can form for an extended period of time or disperse quickly with the underlying water through wind-induced mixing in this area. Higher wind speeds do not exclude the existence of salinity and temperature anomalies caused by rain but depend on the input rate of freshwater, its mixing with the underlying NSL, and the droplet distribution. However, freshwater remains on the surface for a much shorter time at higher wind speeds, reaching a rate of salinity change (−0.037 g kg min−1) that is nine times higher compared to the rate at a depth of 100 cm (−0.004 g kg min−1). We compared our results with those of Katsaros and Buettner (1969) and Schlüssel et al. (1997) and find that wind speed plays a crucial role in the dispersion of freshwater during precipitation events. Because the properties of precipitation can vary greatly between different rainfall events, this study provides first insights into how droplet size and fall velocity are also crucial for understanding the effects of precipitation on SSS and SST anomalies in the SML.
With the comparison of two different rainfall events, we have shown that the balance between rainfall intensity, wind speed, and droplet properties determines the SSS anomalies at the SML. More comprehensive datasets are required for a detailed mechanistic understanding of the fate of freshwater in the ocean, including freshwater fluxes and the formation of freshwater lenses at mid-latitudes. As freshwater lenses at the surface are known to influence gas and heat fluxes (Liss et al., 1997; Ho et al., 2004), knowledge of the occurrence and magnitude of freshwater fluxes is crucial. This study can also provide a better understanding of the global water cycle and the usefulness of SSS as a proxy for freshwater fluxes over the ocean.
Data accessibility statement
All the data generated or analyzed during this study is available in this published article and on PANGAEA (Gassen et al., 2024a; Gassen et al., 2024b).
Acknowledgments
We thank the crew and captain of RV Heincke for their support. We would like to thank the team of ICBM-WHV workshop for their support and maintenance of HALOBATES. We also thank R. Henkel for conducting the salinometer measurements and the Alfred Wegner Institute for providing the salinometer.
Funding
Lisa Gassen thanks the University of Oldenburg for the internal funding of her position in the framework of “Promotion of Young Scientists.” The work on the RV Heincke was funded by the Lower Saxony Ministry of Science and Culture within the project The North Sea from Space: Using explainable artificial intelligence to improve satellite observations of climate change (NorthSat-X)—project number VWZN3680. The evaluations and the writing of the publication were carried out within Biogeochemical Processes and Air–sea Exchange in the Sea-Surface Microlayer (BASS), SP 2.3—project numbers 427614800 and 451574234.
Competing interests
The authors have declared that no competing interests exist.
Author contributions
Contributed to conception and design: LG, OW.
Contributed to acquisition of data: LG, OW, SMA.
Contributed to analysis and interpretation of data: LG, OW, MR-R, THB.
Drafted and/or revised the article: All authors.
Approved the submitted version for publication: All authors.
References
How to cite this article: Gassen, L, Ayim, SM, Badewien, TH, Ribas-Ribas, M, Wurl, O. 2024. Wind speed effects on rainfall-induced salinity and temperature anomalies at the sea surface microlayer at mid-latitudes. Elementa: Science of the Anthropocene 12(1). DOI: https://doi.org/10.1525/elementa.2024.00004
Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA
Associate Editor: Lisa A. Miller, Institute of Ocean Sciences, Fisheries and Oceans, Sidney, BC, Canada
Knowledge Domain: Ocean Science