Which Processes Cause Surface Seawater To Increase In Density?

Abstract

Under conditions with a large solar flux and low wind speed, a stably stratified warm layer forms at the ocean surface. Evaporation can then lead to an increase in salinity in the warm layer. A large temperature gradient will decrease density enough to counter the density increase caused by the salinity increase, forming a stable positive salinity anomaly at the surface. If these positive salinity anomalies are large in terms of the change in salinity from surface to the base of the gradient, if their areal coverage is a significant fraction of the satellite footprint, and if they persist long enough to be in the satellite field of view, they could be relevant for calibration and validation of L-band microwave salinity measurements. A towed, surface-following profiler was deployed from the N/O Thalassa during the Subtropical Atlantic Surface Salinity Experiment (STRASSE). The profiler measured temperature and conductivity in the surface ocean at depths of 10, 50, and 100 cm. The measurements show that positive salinity anomalies are common at the ocean surface for wind speeds less than 4 m s⁻¹ when the average daily insolation is >300 W m⁻² and the sea-to-air latent heat flux is greater than zero. A semiempirical model predicts the observed dependence of measured anomalies on environmental conditions. However, the model results and the field data suggest that these ocean surface salinity anomalies are not large enough in terms of the salinity difference to significantly affect microwave radiometric measurements of salinity.

1. Introduction

Salinity is a fundamental state variable for ocean waters and can be used as a tracer of ocean circulation and the freshwater flux [Fedorov et al., 2007; Heffner et al., 2008; Maes, 2008]. Additionally, understanding regional trends of ocean salinity is a key component of monitoring the response of the hydrological cycle to global warming and natural climate variability [Boyer et al., 2007]. The NASA Aquarius and ESA SMOS missions have the goal using L-band microwave radiometry to provide global maps of sea surface salinity with a precision of 0.1 psu and a spatial resolution on the order of 100 km at time scales on the order of 30 days. The success of these missions represents a major advance in the ability to measure ocean surface salinity on global and regional scales. Furthermore, they will advance the understanding of satellite instrumental technologies required for space-based long-term monitoring of sea surface salinity with the accuracy and precision required for climatic and oceanographic objectives.

L-band salinity measurements rely on the known dependence of sea surface microwave emissivity on the conductivity of seawater [Klein and Swift, 1977], which is a function of salinity and temperature. However, L-band radiometers measure salinity at depths of less than a centimeter [Swift, 1980] which is considerably less than the typical 5 m measurement depth for sensors deployed on platforms like ARGO drifters and thermosalinographs (TSGs) mounted on ships. Salinity gradients can form in the surface due to precipitation [Boutin and Martin, 2006; Boutin et al., 2013; Soloviev and Lukas, 1996] and evaporation [Saunders, 1967; Soloviev and Lukas, 1997]. In the case of precipitation, these negative salinity anomalies (i.e., salinity decreasing with decreasing depth) can extend for several kilometers [Soloviev and Lukas, 1996] and can persist for many hours under low wind conditions [Boutin and Martin, 2006].

Although negative salinity anomalies have been observed in the ocean, less is known concerning the behavior of positive anomalies (i.e., increasing salinity with decreasing depth) that form through evaporation. When the shortwave solar flux is high at low wind speeds, the top few meters of the ocean form a warm layer where the temperature increases toward the surface. If these conditions persist for many hours, evaporation from the ocean surface can lead to an increase in salinity. If the temperature gradient is large enough, the decrease in density due to the increase in temperature will counter the increase in density due to the increase in salinity, and a stable, positive salinity anomaly at the surface can form. If these positive salinity anomalies are large in terms of the change in salinity from surface to the base of the gradient, if their areal coverage is large, and if they persist long enough to be in the satellite field of view, they could also be relevant for calibration and validation of L-band microwave salinity measurements.

Determining the conditions under which very near-surface salinity anomalies will be present, their magnitude, and the spatial and temporal extents, is a critical step in understanding if they are affecting microwave radiometric salinity measurements. This paper presents results from measurements made in August and September of 2012 aboard the N/O Thalassa during the Subtropical Atlantic Surface Salinity Experiment (STRASSE). A towed, surface-following profiler was used to measure temperature and salinity in the top meter of the ocean with submeter resolution in both the vertical and the horizontal directions. Under conditions with low wind and high insolation, positive salinity anomalies were observed. The formation mechanisms of these gradients are discussed in terms of the evaporation rate, wind-induced mixing at the surface, and downwelling shortwave flux.

2. A Semiempirical Model of Positive Salinity Anomaly Formation

The formation of a positive salinity anomaly requires the presence of a surface warm layer, where the decrease in density with temperature compensates for the increase in density as salinity increases. This situation is shown graphically in the inset to Figure 1, which shows how the decrease in density due to the surface warming must be larger than the increase in density due to evaporation in order for a positive surface salinity anomaly to form. Formation of a diurnal warm layer in turn requires a low to moderate wind speed to minimize wind-generated mixing and a high downwelling shortwave flux to provide the warming. As wind speed increases, the magnitude of the diurnal warm layer and density stratification decreases [Gentemann et al., 2003; Ward, 2006]. Therefore, it might be expected that the maximum salinity increase would be found at the lowest wind speeds.

The diurnal cycle of the absolute value of the density difference, Δρ (kg m⁻³), between density at a reference depth of 1 m and the density at the sea surface due to an increase in sea-surface temperature (SST) and due to evaporation of water from the surface for a wind speeds, U, of 2 m s⁻¹. The diurnal SST cycle was predicted by the model of Gentemann et al. [2003]. The details of the calculation for Δρ for salinity are provided in section 2. The inset in the upper left shows typical density profiles due to temperature and salinity at the maximum value of the diurnal temperature signal.

Salinity increases at the ocean surface due to evaporation (i.e., the latent heat flux), which increases with wind speed. Therefore, although low wind speeds and high insolation result in a large stable density stratification that could in principle support a large increase in salinity, it is possible that the evaporation rate would be too small to allow a significant increase in salinity over the lifetime of the diurnal warm layer. Formation of positive salinity anomalies requires a balance between wind speeds low enough to allow formation of a diurnal warm layer and wind speeds high enough to generate a significant latent heat flux.

The starting point for the semiempirical model used here to explore the relationship between the temperature and salinity gradients is the parameterization of the diurnal warm layer evolution developed by Gentemann et al. [2003]. This relation predicts ΔSST, the difference in sea surface temperature (SST) from the Reynolds Optimum Interpolated (OI) SST [Reynolds and Smith, 1994] to the subskin temperature measured by microwave radiometry using the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) [Wentz et al., 2000], through a diurnal cycle as a function of wind speed, U (m s⁻¹), and average daily insolation, Q _DW (W m⁻²) from

(1)

for k = 0.53, Q _DW >Q ₀ (where Q ₀ = 132 W m⁻²), and f (t) defined as

(2)

where ω = 0.2668 h⁻¹. Given a bulk water temperature, T _W, ΔSST, and salinity, S, the density stratification due to the diurnal warm layer can be calculated using standard relations [IOC, SCOR, and IAPSO, 2010]. This stable stratification can then be compared to the density increase over some specified time period that would be expected due to the evaporation of water from the surface. When the density stratification from ΔSST exceeds the increase in density due to evaporation, then it is possible for positive salinity anomalies to form.

The evaporation of water from the ocean surface is defined by the latent heat flux, Q _L (W m⁻²). The heat flux can be estimated from the bulk formula for water vapor transport in terms of U and the specific humidity difference across the air-sea interface as

(3)

where C _E is the water vapor transfer coefficient and has the value of 1.2 × 10⁻³ [Kraus and Businger, 1994], ΔH _VAP is the heat of vaporization of water (J kg⁻¹) (which can be calculated as a function of T _W and S using standard relations such as those found in the Gibbs Seawater Oceanographic Toolbox of TEOS-10 [IOC, SCOR, and IAPSO, 2010] or published by Sharqawya et al. [2010]), q _A is the specific humidity in the air at the ocean surface (kg m⁻³), and q _S is the saturation specific humidity (kg m⁻³) for water at temperature T _W. Q _L can be converted into the equivalent volume of water lost over a given time period by dividing by ΔH _VAP and the density of seawater, ρ_SW (kg m⁻³), which is known from S and T _W.

In order to relate the volume of water lost due to evaporation to an increase in S at the surface, a near-surface profile in temperature and salinity and the depth to which the salinity and temperature gradient extend must be defined. This is equivalent to assuming that the volume of water lost will come from a shallow layer of depth h (m) near the surface, with the profile defining the fraction of the total volume lost from a specific depth. Here for computational convenience a linear profile of salinity with depth will be used [Fairall et al., 1996] (note that the cool-skin layer included in the original model will be neglected here since the volume contained in the skin layer is negligible compared to the total volume contained in the diurnal warm layer). If a linear gradient in salinity from the surface, s _s, to the well-mixed bulk later, s _b, over a distance from the surface to a depth h is assumed, then the total volume lost due to evaporation must also be distributed linearly over this salinity gradient. Therefore, the volume of water lost at a given depth z where 0 ≤z ≤h is given by

(4)

where the volume change at depth z given by ΔV(z) is the fraction of water lost from a unit volume of one cubic meter, s _b is in units of kilograms of salt per cubic meter (here s will be used to denote salinity in units of kilograms of salt per cubic meter and S will be used for salinity in units of per mille), and s(z) is defined as

(5)

following the linear gradient model of Fairall et al. [1996] assuming the change in salinity in the skin layer is negligible. Equation 4 can be integrated for z = 0 to z =h to yield an expression for the total volume of water lost, V _T, in terms of h, s _b, and s _s. Since the former two are known and V _T is given by Q _L (which is a function of U and T _W), the expected increase in s _s can be predicted from U, T _W, and s _b for a given value of h.

The depth to which s(z) extends, h, is critical in determining the maximum surface salinity since a deeper layer means the volume lost due to evaporation is a smaller fraction of the total volume in the layer. Here it is assumed that h is equal to the depth of the diurnal warm layer, D _T, which is a function of the momentum flux across the air-sea interface and the relative balances of the air-sea heat fluxes, including Q _DW, Q _L, the sensible heat flux, and the upwelling longwave radiative flux [Fairall et al., 1996]. The model proposed by Fairall et al. [1996] for estimating D _T shows that h scales in proportion with U ³, although in extreme cases the dependence on the heat fluxes can complicate this relation.

Detailed study of the dependence of D _T on environmental conditions is beyond the scope of this paper. Instead, several empirically based methods were used to estimate h: (1) h constant with a value chosen based on profiles measured by the salinity profiler; (2) h increasing linearly with U; and (3) h increasing in proportion with U ³ as proposed by Fairall et al. [1996]. For both the linear relationship and the cubic relationship, the proportionality constants and boundary conditions were chosen by fitting the model predictions to the field data from the profiler. The effect of the parameterization of h on S will be discussed below in the context of the field data.

Once the expected increase in temperature at the sea surface is known from equation 1, the density gradient due solely to temperature can be calculated by assuming s is constant throughout the diurnal warm layer (with value equal to s _b). This stable density gradient can be compared to the increase in density caused by the increase in s at the surface due to evaporation. So long as the stable stratification caused by surface warming is smaller than the unstable stratification caused by evaporation, positive salinity anomalies will not form since the total density stratification will be unstable and the surface will overturn and mix downward.

Figure 1 gives a graphical interpretation of the proposed model, where the dashed line shows a time series of the diurnal cycle in the density difference, Δρ (kg m⁻³), from bulk phase to the surface due to formation of a diurnal warm layer for U = 2 m s⁻¹. The magnitude of ΔSST as a function of time and U was predicted by (1) with Δρ calculated as the difference between the density at a depth of 1 m and the density at the surface. Boundary conditions for the model were derived primarily from the field measurements (which are discussed in detail below). Specifically, the bulk temperature, T _B, was 26.0°C, the salinity at the base of the warm layer, S _B, was 37.4‰, the average daily insolation was 300 W m⁻², and the relative humidity, RH, was 80%. The solid line represents the density increase at the surface due to the increase in salinity from evaporation over a period of 1 h that would be expected given a linear gradient in salinity from the surface to a depth of 0.5 m, Q _L given by (3) where q _S is determined by the vapor pressure of seawater at T _W and q _A is defined by RH, T _A, and the atmospheric pressure assumed to be 1 atm (note that Δρ for the evaporative flux is actually opposite in sign from Δρ for the surface warming, and the absolute value of each is plotted in the figure).

The model output in Figure 1 shows that from approximately 0100 until 0900, the thermal stratification is either itself negative, or smaller than the density increase due to evaporation. These conditions lead to an unstable density profile, and the surface mixes by convective overturning so that salinity does not increase at the surface. After 0900, the thermal stratification is large enough that the water lost from the surface to evaporation is now taken only from the layer defined by h. This causes an increase in surface salinity at a rate defined by Q _L. Once the solar flux decreases in the afternoon, the density stratification due to temperature begins to decrease as the surface cools. Once the density lines intersect at approximately 1700, the density increase due to evaporation ceases since the surface is now unstably stratified and overturns. However, the total increase in density between onset and cessation of the capped layer defines the maximum increase in salinity that is possible for these conditions. Specifically, the density increase given in Figure 1 corresponds to an increase in S at the surface of 0.068‰.

Once a surface warm layer depth has been specified, the semiempirical method presented above allows the bulk-surface salinity anomaly to be calculated given a set of environmental conditions. These calculated values can be compared to salinity anomalies measured in the field.

3. Field Measurements and Methods

The measurements made here used a towed surface salinity profiler (SSP) that mounted a vertical array of four Seabird SBE-49 FastCAT CTDs (Seabird Electronics, Bellevue, Washington) mounted to a 1 m deep rigid keel. The keel was attached to a 2.5 m surfboard (ACS Magnum, BiC Sport, Vannes, France). The SBE-49s were mounted at nominal depths of 10, 20, 50, and 100 cm. Data were logged locally using a Wetlabs DH-4 (Wetlabs, Grants Pass, Oregon) operating at a sample rate of 6 Hz for each SBE-49. Given a nominal tow speed of 2 m s⁻¹ and a temperature and conductivity response time constant of 0.085 s, the horizontal spatial resolution of the SSP was on order of 0.3 m. Position of the SSP was determined from an on-board GPS that was also logged by the DH-4 at a rate of one sample per 60 s. The GPS data were linearly interpolated to provide positions at each data point from the four SBE-49s. A photograph of the instrumented keel of the profiler is shown in Figure 2 (top right).

(top left) The surface salinity profiler with the outrigger floats used for the deployments STR-1, STR-2, and STR-3 (SSP). (bottom left) The surface salinity profiler in the catamaran configuration used for the deployments STR-4, STR-5, STR-6, and STR-7 (SSPC). (right) Instrumented keel used on both the SSP and SSPC showing the locations of the four SBE-49 CTDs mounted at depths of 10, 20, 50, and 100 cm.

Two configurations of the SSP were used. The first used two cantilevered outrigger pontoons as shown in Figure 2 (top left). This design worked for tow speeds at or below 1.5 m s⁻¹, but above this speed it became unstable and would invert and dive underwater. The second configuration substituted a second surfboard in place of the two outrigger pontoons and provided additional lateral bracing for the keel holding the instruments (Figure 2, bottom). This design, called the SSP-Catamaran (SSPC), was stable at speeds up to 3.5 m s⁻¹, which was the maximum tow speed attempted. The instrument package was identical for the SSP and SSPC and has been described above.

The difference in performance between the SSP and SSPC is evident from the increase in stable towing speed. One explanation for the increase in stability is provided by comparison of power spectra of the time series of depth recorded by the SBE-49 mounted at 1 m on the SSP and SSPC shown in Figure 3. The SSP and SSPC are essentially surface following, with the standard deviation in depth measured by the CTD mounted at 10 cm being ±3 cm. However, the standard deviation of the pressure signal from the CTD at 1 m for the SSP was approximately double that the value. The power spectra of the pressure record from the 1 m CTD on the SSPC shows a peak at 0.14 Hz which is likely due to small changes in depth as the SSPC rides on the swell. In contrast, the power spectrum for 1 m CTD on the SSP shows considerably more variance at both higher and lower frequencies. It is hypothesized that this represents vibration and flexing of the SSP keel, which was greatly reduced in the SSPC by the addition of cross-bracing of the keel. This flexing likely increased as tow speed increased, leading to instability of the SSP.

Power spectra of the variations in depth measured by the SBE-49 mounted at a depth of 100 cm on the keel of the SSP or SSPC.

The tow bridle of both the SSP and SSPC was a three-point harness mounted to the base of the keel, bow, and stern of the surfboard. Combined with a vane on the keel, the three-point harness forced the profiler to point away from the ship. The profiler then tracked outboard until the force outward and perpendicular to the tow direction from the keel was balanced by the force inward and perpendicular to the tow direction from the tow line. When viewed from above the angle formed by the tow line and the centerline of the keel was 40° as measured forward from the centerline of the keel. With 55 m of tow line, the SSP and SSPC towed outboard of the ship by 35 m. Observations of the ship wake and the SSP or SSPC when under tow showed this was outside of the ship wake so that each sampled undisturbed water.

The time series for water temperature, T _W (C), conductivity, C (mmho/cm), depth, d (m), and salinity, S (‰), from each SBE-49 were low-pass digitally filtered at a frequency of 1 Hz. At a tow speed of 2 m s⁻¹ this gives a spatial resolution of 4 m. It was found that the temperature sensor on the SBE-49 mounted at a depth of 20 cm failed, although the conductivity probe functioned normally. Attempts were made to interpolate temperature from the SBE-49s at 10 and 50 cm to retrieve an estimate of salinity at 20 cm, but this did not produce a usable value. Seawater density, ρ (kg m⁻³), at 10, 50, and 100 cm was calculated from T _W and S using the Gibbs Seawater Oceanographic Toolbox of TEOS-10 [IOC, SCOR, and IAPSO, 2010].

Air temperature, T _A (°C), relative humidity, RH (%), atmospheric pressure, P _A (Pa), wind speed, U (m s⁻¹), and the instantaneous downwelling shortwave radiative flux, Q _SW (W m⁻²), were provided by the onboard data system from meteorological instruments on the Thalassa. These data were available as 30 s averages for the duration of the cruise. The data were used as reported by the data system with the exception of Q _SW which was multiplied by a scale factor of 0.7 to compensate for miscalibration of the sensor. This calibration factor was determined by comparing the measured output of the sensor at local noon on cloud-free days with the expected value based on the known top-of-the-atmosphere irradiance and the ship's latitude.

4. Results and Discussion

Figure 4 shows the locations of the seven deployments of the surface profiler made during STRASSE. There were three deployments of the SSP (STR-01, STR-02, and STR-03) and four of the SSPC (STR-04, STR-05, STR-06, and STR-07). Of these, STR-01, STR-02, STR-06, and STR-07 had a diurnal warm layer and a measurable positive salinity anomaly. There was no indication of stratification in either temperature or salinity in the remaining three. Table 1 lists Q _DW, which was calculated as the daily average of Q _SW, and RH, U, and T _A averaged over the duration of each deployment. Also shown in Table 1 are the average values of T _W and S measured by the SSP or SSPC at a depth of 100 cm, and the difference between T _W and S at depths of 10 and 100 cm, ΔT or ΔS, respectively.

Map showing locations of the three deployments of the SSP (STR-1, STR-2, and STR-3) and the four deployments of the SSPC (STR-4, STR-5, STR-6, and STR-7). The insets show the courses for the tows of the SSP and SSPC for the deployments when positive salinity anomalies were detected. On all four inset plots, the tick marks represent 0.005° of longitude or latitude.

Table 1. Deployment Details, Environmental Conditions, and Average Salinity and Temperature Anomalies Measured During STRASSE

No.	Versiona a Configuration of the Surface Salinity Profiler (SSP): see Figure 2 (top left). SSPC: see Figure 2 (bottom left).	Locationb b Location at the start of the deployment.	Timec c Time at the start of the deployment. (UTC)	T Ad d Air temperature as measured by the ship's meteorological instruments. (°C)	Ue e Wind speed as measured by the ship's anemometer. (m s−1)	RHf f Relative humidity as measured by the ship's meteorological instruments. (%)	Q DWg g Daily averaged downwelling shortwave radiation. (W m−2)	T Wh h Average water temperature at a depth of 1 m as measured by the SSP/SSPC. (°C) (at 1 m)	Si i Average salinity at a depth of 1 m as measured by the SSP/SPPC. (‰) (at 1 m)	ΔTj j Average water temperature difference between 0.1 and 1 m as measured by the SSP/SSPC. (°C)	ΔSk k Average salinity difference between 0.1 and 1 m as measured by the SSP/SSPC. (‰)
1	SSP	26.969°N	1730	25.8	1.1	84.	283	25.93	37.26	2.54	0.050
		24.100°W
2	SSP	26.479°N	1615	27.2	2.0	74.	277	27.55	37.54	0.97	0.024
		35.001°W
3	SSP	26.004°N	1740	27.0	8.4	79.	295	27.59	37.71	n.d.l l n.d. = not detected.	n.d.
		35.697°W
4	SSPC	25.491°N	1540	26.6	4.2	79.	288	27.63	37.74	n.d.	n.d.
		35.762°W
5	SSPC	25.449°N	1540	26.4	4.6	71.	296	27.43	37.59	n.d.	n.d.
		35.684°W
6	SSPC	26.318°N	1645	26.4	4.0	68.	298	27.37	37.58	0.005	0.006
		35.141°W
7	SSPC	29.515°N	1740	26.1	2.9	75.	252	27.35	37.20	0.15	0.004
		32.716°W

• ^a Configuration of the Surface Salinity Profiler (SSP): see Figure 2 (top left). SSPC: see Figure 2 (bottom left).
• ^b Location at the start of the deployment.
• ^c Time at the start of the deployment.
• ^d Air temperature as measured by the ship's meteorological instruments.
• ^e Wind speed as measured by the ship's anemometer.
• ^f Relative humidity as measured by the ship's meteorological instruments.
• ^g Daily averaged downwelling shortwave radiation.
• ^h Average water temperature at a depth of 1 m as measured by the SSP/SSPC.
• ⁱ Average salinity at a depth of 1 m as measured by the SSP/SPPC.
• ^j Average water temperature difference between 0.1 and 1 m as measured by the SSP/SSPC.
• ^k Average salinity difference between 0.1 and 1 m as measured by the SSP/SSPC.
• ^l n.d. = not detected.

A time series of the meteorological conditions present during STR-1 is shown in Figure 5. This data set was collected under conditions of very low wind speed, low relative humidity, high insolation, and a large air-sea temperature difference. The plot of SST in Figure 6 (top) shows that these conditions led to a temperature stratification of nearly 3°C between a depth of 100 and 10 cm, the topmost CTD on the SSP. This large temperature stratification, combined with the low RH and nonzero wind speed, resulted in enough evaporation to produce the large positive salinity anomaly between 100 and 10 cm shown in Figure 6 (middle). The plot of conductivity at all four depths shown in Figure 6 (bottom) demonstrates that the temperature and salinity gradients measured at 100, 50, and 10 cm are consistent with the conductivity gradient at all four depths.

Time series of the instantaneous downwelling shortwave solar radiation, Q _SW (W m⁻²), relative humidity, RH (%), wind speed, U (m s⁻¹), and air temperature, T _A (°C), measured during the SSP deployment STR-1 on 18 August 2012 during STRASSE. The heavy horizontal bar in the bottom plot shows the times and tow distances over which the SSP provided usable data.

(top) Sea-surface temperature at depths of 10, 50, and 100 cm as measured by the Surface Salinity Profiler (SSP) plotted as a function of tow distance during the deployment STR-1. The data key is shown in the figure and shows the three measurement depths. (middle) Sea-surface salinity at depths of 10, 50, and 100 cm as measured by the Surface Salinity Profiler (SSP) plotted as a function of tow distance during the deployment STR-1. The data key is shown in the figure and shows the three measurement depths. (bottom) Time series of conductivity measured by the four SBE-49 CTDs on the SSP at depths of 10, 20, 50, and 100 cm. The data key is shown in the figure and shows the four measurement depths.

Figure 7 shows the salinity, temperature, and density profiles measured in the top meter of the ocean at four locations on the STR-1 track profile. Salinity and temperature at a depth of 3 m averaged over the tow distances 2.0–3.5 km from the thermosalinograph on the Thalassa are also shown. The salinity profiles on the left show that over the 4 km distance sampled by the SSP, ΔS was approximately 0.05‰ with a minimum of 0.01‰. The density profiles plotted in Figure 7 (right) show that the stratification was largely independent of the ΔS, suggesting that it is the temperature stratification that controls ocean surface density under these conditions. This is not surprising as the downwelling shortwave energy flux creating the temperature stratification is 2 orders of magnitude larger than the latent heat flux responsible for the salinity stratification. Furthermore, comparing S and T _W from the SSP to the ship data shows that h, the depth over which the evaporative water flux mixes downward, is approximately 0.5 m.

(left) The salinity, S, profile as a function of depth at four track distances for STR-1 measured by the SSP. (middle) The water temperature, T _W, profile measured by the SSP at those same locations. (right) The ocean surface density, ρ, profile calculated from S and T _W at the same locations. The vertical dashed lines in the salinity and temperature plots show S and T _W measured by the ship's thermosalinograph at a depth of 3 m averaged over the interval from 2.0 to 3.5 km. The data key lists the four tow distances at which the gradients were measured.

Although not shown, the second straight line tow for STR-2 also measured vertical salinity anomalies over the depths of 10–100 cm. The vertical and spatial structures of these salinity anomalies were consistent in their general features to that shown above for STR-1.

Figure 8 is a time series of meteorological conditions for STR-7 similar to that shown for STR-1 in Figure 5. The main differences between the 2 days are U was higher at 2.9 m s⁻¹ and RH was lower at 75%. These conditions led to a higher evaporation rate, so that a measurable salinity anomaly could form despite the smaller temperature stratification. The impact of the higher wind speed can be seen in the plots of temperature and salinity at all three measured depths as a function of track distance shown in Figure 9. In contrast to the data from STR-1, the temperature data for STR-7 show that over most of the track T _W is well mixed to a depth of 50 cm, with a small decrease in temperature from 50 to 100 cm. However, there are also short sections where the ocean surface is well mixed to a depth of 100 cm (e.g., 8 km).

Time series of the instantaneous downwelling shortwave solar radiation, Q _SW (W m⁻²), relative humidity, RH (%), wind speed, U (m s⁻¹), and air temperature, T _A (°C), measured during the SSP deployment STR-7 on 10 September 2012 during STRASSE. The heavy horizontal bar in the bottom plot shows the times and tow distances over which the SSP provided usable data.

(top) Sea-surface temperature at depths of 10, 50, and 100 cm as measured by the SSPC during the STR-7 deployment plotted as a function of tow distance. The data key is shown in the figure and shows the three measurement depths. (bottom) Sea-surface salinity at depths of 10, 50, and 100 cm as measured by the SSPC during the STR-7 deployment plotted as a function of tow distance. The data key is shown in the figure and shows the three measurement depths.

The SSPC was towed in a circle pattern (see Figure 4) for STR-7, and Figure 10 shows a contour plot of T _W at a depth of 50 cm. Figure 10 reveals there is a strip of higher temperature water running from the northwest to the southeast.

Contour plot of sea surface temperature, T _W (°C), measured at a depth of 50 cm by the SSPC during STR-7. The solid white line shows the track of the profiler under tow by the N/O Thalassa where the SSPC approached the circle pattern from the northwest. The diameter of the circle is 1.8 km.

Salinity stratification was not as pronounced for STR-7 as seen in STR-1, and the plot of the salinity as a function of depth in Figure 9 shows that salinity was well mixed in the upper meter over large distances of the track (e.g., from 4 to 6 km, from 10 to 12 km). Like temperature, S shows a pronounced correlation with position on the circle pattern, and the contour plot of salinity measured at 50 cm depth shown in Figure 11 also has the same NW-SE strip of water seen in Figure 10. Contours for ρ at 50 cm calculated from T _W and S are shown in Figure 12. In agreement with previous measurements of temperature and salinity fronts in the ocean, the warmer saltier water is seen to be less dense than the cooler fresher water [Rudnick and Martin, 2002].

Contour plot of sea surface salinity measured at a depth of 50 cm by the SSPC during STR-7. The solid white line shows the track of the profiler under tow by the N/O Thalassa where the SSPC approached the circle pattern from the northwest. The diameter of the circle is 1.8 km.

Contour plot of sea density, ρ (kg m⁻³), calculated at a depth of 50 cm from the temperature and salinity data in Figures 10 and 11. The solid white line shows the track of the profiler under tow by the N/O Thalassa where the SSPC approached the circle pattern from the northwest. The diameter of the circle is 1.8 km.

Figure 13 shows a close-up of T _W and S from STR-7 from 4 to 8 km. The data for T _W show the presence of a diurnal warm layer over the entire segment, although the surface layer is well mixed down to a depth of 50 cm. In contrast, S has segments where there is a salinity anomaly (e.g., from 6.5 to 7.5 km) and also sections where salinity is well mixed (e.g., from 5 to 6 km). Furthermore, the salinity anomalies measured during STR-7 are an order of magnitude smaller on average than those measured for STR-1 (see Table 1).

(top) A close-up plot of sea-surface temperature at depths of 10, 50, and 100 cm for a 4 km long section of the data shown in Figure 9 as measured by the SSPC during the STR-7 deployment plotted as a function of tow distance. The data key is shown in the figure and shows the three measurement depths. (bottom) A close-up plot of sea-surface salinity at depths of 10, 50, and 100 cm for a 4 km long section of the data shown in Figure 9 as measured by the SSPC during the STR-7 deployment plotted as a function of tow distance. The data key is shown in the figure and shows the three measurement depths.

The magnitude of the observed salinity anomaly is small, and it might be possible these salinity differences reflect noise or drift in the SBE-49s rather than an actual salinity increase at the surface. This was tested by comparing the spatial distributions of T _W, S, ρ, and the corresponding values for ΔT, ΔS, and Δρ to see if these features were correlated spatially (here ΔT is defined as the difference between T _W at 10 cm and T _W at 100 cm so that a positive value denotes a surface warm layer, ΔS is defined as the difference between S at 10 cm and S at 100 cm so that a positive value implies a positive salinity anomaly, and Δρ is defined as the difference between ρ at 10 cm and ρ at 100 cm so that a negative value implies a stably stratified surface layer). Figure 14 shows a contour plot of ΔT, Figure 15 is a contour plot of ΔS, and Figure 16 shows a contour plot Δρ.

Contour plot of the vertical sea surface temperature anomaly, ΔT (°C), as measured by the SSPC during STR-7. ΔT has been calculated as the temperature at a depth of 10 cm minus the temperature at 100 cm so that a positive value denotes the existence of a diurnal warm layer. The solid white line shows the track of the profiler under tow by the N/O Thalassa where the SSPC approached the circle pattern from the northwest. The diameter of the circle is 1.8 km.

Contour plot of the vertical sea surface salinity anomaly, ΔS (‰), as measured by the SSPC during STR-7. ΔS has been calculated as the salinity at a depth of 10 cm minus the salinity at 100 cm so that a postive value reflects salinity increasing toward the surface and a negative value is salinity decreasing toward the surface. The solid white line shows the track of the profiler under tow by the N/O Thalassa where the SSPC approached the circle pattern from the northwest. The diameter of the circle is 1.8 km.

Contour plot of the vertical sea surface density anomaly, Δρ (kg m⁻³), as measured by the SSPC during STR-7. Δρ has been calculated as the density at a depth of 10 cm minus the density at 100 cm so that a negative value reflects a stably stratified surface layer. The solid black line shows the track of the profiler under tow by the N/O Thalassa where the SSPC approached the circle pattern from the northwest. The diameter of the circle is 1.8 km.

Comparing Figures 10 and 14 shows ΔT is largest in the region where the surface temperature is coolest. In contrast, the warmest regions were those where the top meter showed no temperature stratification. In similarity with T _W, S is also seen to have little spatial correlation with ΔS, with the region of high salinity water lying to the southwest of the region where ΔS is largest. Not surprisingly, comparison of the spatial distribution of ΔT in Figure 14 with ΔS in Figure 15 shows a high level of spatial correlation between the two. The spatial distribution of Δρ shows that areas with the most stable density stratification are found where both ΔT (stably stratifying) and ΔS (unstably stratifying) are largest. This shows that in the vertical direction, ΔT is more important in controlling Δρ than ΔS, a finding that supports previous observations that the salinity ratio, R (which quantifies the relative importance of T _W or S in determining the variability of ρ [Rudnick and Martin, 2002]), is found to be approximately 2 in the ocean.

The difference in spatial patterns for T _W and ΔT and for S and ΔS, combined with the high degree of spatial correlation between ΔT, ΔS, and Δρ provide evidence that the salinity gradients are being formed by evaporation. If the gradients were being formed by high salinity water moving horizontally at the surface to overlay lower salinity water underneath (possibly due to slumping of the front [Rudnick and Martin, 2002]), that would require the high salinity water to cool to match the observed surface temperature in regions where positive salinity anomalies were measured. However, given the high insolation during the measurement period, surface cooling would not be expected.

Figure 17 shows ΔS plotted as a function of U where ΔS was calculated as the average over track distances of length from 0.2 to 1 km for the STR-1, STR-2, STR-6, and STR-7 data sets. There is a clear correlation of ΔS with U, with ΔS decreasing rapidly with increasing U. One feature of the data is that near-surface salinity anomalies were never observed for U > 4 m s⁻¹, and when salinity anomalies were observed at U = 4.0 m s⁻¹ during STR-6, RH was 68%, the lowest value measured during the SSP deployments. The low RH led to a large evaporation rate and a measurable ΔS despite the higher wind speed.

The salinity difference between a depth of 100 and 10 cm, ΔS (‰), plotted as a function of U as measured during STR-1, STR-2, STR-6, and STR-7 (see Table 1). The values for ΔS were calculated as averages of 1 km track segments for periods when a measurable salinity anomaly was detected. Also shown in the figure is ΔS calculated from the model discussed in section 2 using three different parameterizations for the depth of the diurnal warm layer, h, as a function of U, h constant with U, h increasing linearly with U, and h increasing as a cubic function of U (details of these relationships are found in the text). All model runs assumed that relative humidity was constant with a value of 80% and that the average daily downwelling solar flux was constant at 300 W m⁻². The data key is shown in the figure.

Following the discussion in section 2, the magnitude of a positive near-surface salinity anomaly depends on a balance between evaporation and surface mixing. If the evaporation rate is too small, then the salinity anomaly will be small regardless of the value of the surface layer depth h. Similarly, if h is large, then even for high evaporations rates the anomaly will be small since the increase in salt concentration is mixed into a larger volume, diluting the salinity signal. The decrease of ΔS with increasing U in Figure 17 suggests that h is a critical parameter in formation of positive salinity anomalies. In order to test this hypothesis, the semiempirical model described in section 2 was applied to the conditions present for STRASSE.

The model for predicting ΔS presented in section 2 was initialized using Q _DW = 300 W m⁻², T _W = 26.4°C, S = 37.4‰, and RH = 80% and ΔS was calculated as a function of U for three different parameterizations of h: h constant with the depth equal to 0.5 m as measured by the SSP during STR-1; h increasing linearly with a slope defined by assuming h = 0.1 m at U = 0 m s⁻¹ and h = 0.5 m at U = 1 m s⁻¹ (where the lower limit of h = 0.1 m was defined by the diffusive length scale of heat over a time period of 1 h and h = 0.5 m at U = 1 m s⁻¹ is equal to that measured during STR-1); and h increasing in proportion with U^{− 3} following the functional form described in detail below. Results from all three model runs are shown in Figure 17.

The large dashed line in Figure 17 shows ΔS calculated assuming h was constant. Under these conditions ΔS increases to a maximum of 0.09‰ for U = 2 m s⁻¹ and then decreases to zero at a wind speed of 7 m s⁻¹. In this case, the shape of the curve for ΔS is determined mainly by the behavior of the diurnal warm layer defined by (1) and to a lesser extent Q _L. Although the predicted magnitude of the anomaly is approximately correct, the measured ΔS values decrease much faster than the model output.

The steep decrease of the measured ΔS with increasing U implies that h increases with wind speed, since mixing the high salinity water downward into a larger volume would cause a decrease in ΔS. The short-dashed line in Figure 17 shows the model results calculated assuming that h increased linearly as described above. Allowing h to increase with increasing U yields a larger maximum value of ΔS since at lower wind speeds the water lost due to evaporation is taken from a smaller total volume. The overall shape of the curve is more similar to the field data than assuming a constant h. For the linearly increasing h, the maximum value of ΔS is a more sharply defined, and it occurs at a lower wind speed compared to when h was held constant. However, even a linear increase in h is unable to produce as rapid a decrease in ΔS as was observed.

The solid line shows h increasing as the cube of U following the functionality of the depth of the diurnal warm layer as a function of wind speed proposed by Fairall et al. [1996]. Here rather than determine h from the wind stress as specified in the model, the relation between h and U has been parameterized solely in terms of wind speed. It was assumed that h could be estimated from

(6)

where the minimum depth of 0.13 m was chosen to be consistent with the minimum depth used in the linear model. The coefficient 0.47 was chosen to minimize ε, the total absolute difference between the measured and modeled values for ΔS defined as

(7)

where the summation is over all the measured data points shown in Figure 17. Use of (6) to estimate h does an excellent job reproducing both the magnitude of ΔS and its dependence on U.

Clearly, the dynamical processes that lead to the formation of positive near-surface salinity anomalies are more complicated than have been described in section 2, but the agreement between the solid curve and the field data suggests that the simple model developed here captures most of the essential physics.

The SSPC data point at U = 4 m s⁻¹ from STR-6 (see Table 1) in Figure 17 is interesting since ΔS is larger than that predicted by the cubic model. This disagreement was most likely caused by the drier air present on this deployment (RH = 68%). Drier air would lead to a larger Q _L, more evaporation, and a larger salinity anomaly. In fact, it is possible that some of the scatter between the cubic model and the experimental data might be due to environmental variability in RH, T _W, T _A, and Q _DW. This hypothesis was tested by initializing the empirical model on a case-by-case basis using the environmental data provided in Table 1 and comparing the predicted values for ΔS with the averaged measured values.

Figure 18 shows ΔS calculated using the model initialized with the set of constant environmental conditions used in Figure 17 plotted versus the measured values of ΔS. As suggested by Figure 17, there is good agreement between the model predictions and measured data. Also shown in Figure 18 is ΔS calculated by the model when initialized on a case-by-case basis using the environmental data in Table 1. When the model is run taking environmental variability into account, the agreement between the measured and modeled salinity anomalies is much better. Furthermore, there is also evidence that the model is systematically underpredicting ΔS. The reasons for this are not clear at present, but might be related to the model failing to capture the effects of gustiness on Q _L or overestimation of the salinity mixing depth.

The salinity anomaly between a depth of 100 and 10 cm as calculated by the model presented in section 2, ΔS _Modeled (‰), plotted versus the salinity anomaly between a depth of 100 and 10 cm measured by the SSP/SSPC, ΔS _Measured (‰). The model runs were initialized using two sets of boundary conditions: the open squares show ΔS _Modeled calculated assuming relative humidty, RH, was constant at 80%, bulk water temperature, T _W, was constant at 26°C, air temperature, T _A, was constant at 28°C, and the average daily downwelling shortwave flux, Q _DW, was constant at 300 W m⁻² with wind speed, U, taken from Table 1. the solid circles show ΔS _Modeled calculated using RH, T _W, T _A, and Q _DW were taken from Table 1. The data key is shown in the figure.

5. Conclusions

The measurements made during STRASSE with the surface salinity profiler demonstrate that positive salinity anomalies can exist and persist at the ocean surface at wind speeds up to 4 m s⁻¹. They likely represent a ubiquitous feature of an ocean surface with high insolation and low wind speed. However, based on the data and model results shown in Figure 17, the model developed here suggests that ΔS for these features is likely to be small, with a maximum value on order of 0.1‰ for the range of relative humidity typically found in the marine atmosphere. If the spatial inhomogeneity of these features seen in Figure 15 is typical of their surface expression, it is unlikely that positive salinity anomalies would result in significant biases between radiometrically measured salinities and salinities measured at depths of a few meters when area averaged over the footprint of a satellite-mounted instrument with a length scale on order of 100 km.

The semiempirical model developed here is instructive in understanding the processes that control near-surface positive salinity anomalies. A more sophisticated model that determines the bulk-skin temperature gradient and its mixing depth from first principles would provide more detail on this process. Although such a modeling study is beyond the scope of this paper, the simple model presented here illustrates how the competing effects of wind-generated mixing and wind-forced evaporation combine with the near-surface temperature gradient to limit the magnitude of the salinity anomaly.

Finally, the data also show that a surface-following platform such as the SSP or SSPC provide useful measurements of salinity and temperature very near the ocean surface, allowing the processes that control near-surface gradients to be studied. The relatively high speed of a towed profiler allows it to sample relatively large areas of the ocean and measure variability on spatial and temporal scales seen by satellite instruments. This complements measurements from slower moving platforms such as wavegliders and seagliders that can make measurements over much longer time periods.

6 Acknowledgments

We extend our thanks to the captain and crew of the N/O Thalassa for their assistance in deploying and operating the surface salinity profiler. We would also like to thank the Chief Scientist of STRASSE, Gilles Reverdin, for providing the opportunity to participate in STRASSE and also for his invaluable help aboard the Thalassa during the cruise. These measurements would not have been possible without his support, patience, and enthusiasm. This research was funded by the NASA Aquarius Mission under grant NNX09AU73G and the data will be available through the NASA Aquarius Mission website.

References

• Boutin, J., and N. Martin (2006), ARGO upper salinity measurements: Perspectives for L-band radiometers calibration and retrieved sea surface salinity validation, IEEE Geosci. Remote Sens. Lett., 3(2), 202– 206, doi:10.1109/LGRS.2005.861930.
• Boutin, J., N. Martin, G. Reverdin, X. Yin, and F. Gaillard (2013), Sea surface freshening inferred from SMOS and ARGO salinity: Impact of rain, Ocean Sci., 9(1), 183– 192, doi:10.5194/os-9-183-2013.
• Boyer, T., J. Carton, Y. Chao, A. Gordon, G. Johnson, G. Lagerloef, B. Large, S. Riser, and R. Schmitt (2007), What's next for salinity?, Oceanography, 21, 82– 85.
• Fairall, C. W., E. F. Bradley, J. S. Godfrey, G. A. Wick, J. B. Edson, and G. S. Young (1996), Cool-skin and warm-layer effects on sea surface temperature, J. Geophys. Res., 101(C1), 1295– 1308.
• Fedorov, A., M. Barreiro, G. Boccaletti, R. Pacanowski, and S. G. Philander (2007), The freshening of surface waters in high latitudes: Effects on the thermohaline and wind-driven circulations, J. Phys. Oceanogr., 37(4), 896– 907, doi:10.1175/jpo3033.1.
• Gentemann, C. L., C. J. Donlon, A. Stuart-Menteth, and F. J. Wentz (2003), Diurnal signals in satellite sea surface temperature measurements, Geophys. Res. Lett., 30(3), 1140, doi:10.1029/2002GL016291.
• Heffner, D. M., B. Subrahmanyam, and J. F. Shriver (2008), Indian Ocean Rossby waves detected in HYCOM sea surface salinity, Geophys. Res. Lett., 35, L03605, doi:10.1029/2007GL032760.
• Klein, L. A., and C. T. Swift (1977), An improved model for the dielectric constant of sea water at microwave frequencies, IEEE J. Oceanic Eng., 25, 104– 111.
• Kraus, E. B., and J. A. Businger (1994), Atmosphere-Ocean Interactions, 2nd ed., 362 pp., Oxford Univ. Press, N. Y.
• Maes, C. (2008), On the ocean salinity stratification observed at the eastern edge of the equatorial Pacific warm pool, J. Geophys. Res., 113, C03027, doi:10.1029/2007jc004297.
• IOC, SCOR, and IAPSO (2010), The international thermodynamic equation of seawater – 2010: Calculation and use of thermodynamic properties, Intergovernmental Oceanographic Commission, Manuals and Guides No. 56, UNESCO (English), 196 pp.
• Reynolds, R. W., and T. M. Smith (1994), Improved global sea surface temperature analyses using optimum interpolation, J. Clim., 7, 929– 948.
• Rudnick, D. L., and J. P. Martin (2002), On the horizontal density ratio in the upper ocean, Dyn. Atmos. Oceans, 36(1–3), 3– 21, doi:10.1016/s0377-0265(02)00022-2.
• Saunders, P. M. (1967), The temperature at the ocean-air interface, J. Atmos. Sci., 24, 269– 273.
• Sharqawya, M. H., J. H. Lienhard V, and S. M. Zubair (2010), Thermophysical properties of seawater: A review of existing correlations and data, Desalin. Water Treat., 16, 354– 380, doi:10.5004/dwt.2010.1079.
• Soloviev, A., and R. Lukas (1996), Observation of spatial variability of diurnal thermocline and rain-formed halocline in the western Pacific warm pool, J. Phys. Oceanogr., 26(11), 2529– 2538.
• Soloviev, A., and R. Lukas (1997), Observation of large diurnal warming events in the near-surface layer of the western equatorial Pacific warm pool, Deep Sea Res., Part I, 44(6), 1055– 1076.
• Swift, C. T. (1980), Passive microwave remote-sensing of the ocean—Review, Boundary Layer Meteorol., 18, 25– 54.
• Ward, B. (2006), Near-surface ocean temperature, J. Geophys. Res., 111, C02005, doi:10.1029/2004JC002689.
• Wentz, F. J., C. L. Gentemann, D. K. Smith, and D. B. Chelton (2000), Satellite measurements of sea-surface temperature through clouds, Science, 288(5467), 847– 850, doi:10.1126/science.288.5467.847.

Citing Literature

Which Processes Cause Surface Seawater To Increase In Density?

Source: https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2014JC009808

Posted by: vanarsdalegagainfoute.blogspot.com

Share this post