Transport at the Air-Sea Interface

Significant advances have been made over the last decade in estimating air-sea CO₂ fluxes over the ocean by the bulk formulation that expresses the flux as the product of the gas transfer velocity and the concentration difference of aqueous CO₂ over the liquid boundary layer. This has resulted in a believable global monthly climatology of air-sea CO₂ fluxes over the ocean on a 4^o by 5^o grid [38]. It is shown here that the global air-sea CO₂ fluxes are very sensitive to estimates of gas transfer velocity and the parameterization of gas transfer with wind. Wind speeds can now be resolved at sufficient temporal and spatial resolution that they should not limit the estimates, but the absolute magnitudes of winds for different wind products differ significantly. It is recommended to use satellite-derived wind products that have the appropriate resolution instead of assimilated products that often do not appropriately resolve variability on sub-daily and sub-25-km space scales. Parameterizations of gas exchange with wind differ in functional form and magnitude but the difference between the most-used quadratic relationships is about 15%. Based on current estimates of uncertainty of the air-water CO₂ concentration differences, the winds, and the gas exchange-wind speed parameterization, each parameter contributes similarly to the overall uncertainty in the flux that is estimated at 25%.

The gas transfer process across the air-water interface induced by far-field homogeneous turbulence generated in the water phase was investigated experimentally. The measurements were performed in a grid-stirred tank using a combined Particle Image Velocimetry (PIV) - Laser Induced Fluorescence (LIF) (PIV-LIF) technique, which enables simultaneous and spatially synoptic measurements of velocity and gas concentration fields. The techniques allowed visualization of the velocity and concentration fields with good spatial and temporal resolution and thus provided good insight into the gas transfer mechanisms. Detailed quantification of the gas concentration distribution within the thin aqueous boundary layer as well as the near surface hydrodynamics were obtained. With the combined PIV-LIF technique, the turbulent mass flux covariance term c′w′ was quantified directly. Comparing the turbulent mass flux with the total mean mass flux determined from the reaeration (bulk) measurements, it could be shown that the contribution of c′w′ is indeed significant.

We present laboratory measurements of simultaneous velocity and concentration fields for the transfer of CO₂ across a free surface. The interface is subject to the effects of free shear turbulence generated far beneath the surface, exhibiting low mean flow and excellent homogeneity. From measurements of the spatio-temporal mass flux we examine coherent structures below the free surface, as well as one-point statistics to better understand the fundamental physics of turbulent transport at a free surface in the absence of mean shear. We observe surface penetration events caused by bulk fluid impacting the interface from below, as well as downwelling events in which the near-surface fluid is injected into the bulk in narrow filaments. Both types of events contribute to the turbulent mass flux, and we measure that downwelling events are responsible for at least as much mass transfer as the upwellings on which existing models are based. Our measurements indicate that the dominant length and time scales are different for upwellings and downwellings; the quantification of these will be important to modeling efforts.

Laser-Induced Fluorescence (LIF) is applied to observe directly the mechanism of gas exchange in the aqueous viscous boundary layer at a free water surface. In order to make dissolved oxygen (DO) visible, a new class of dyes with a long phosphorescent lifetime in the order of microseconds is used. This property makes the quenching constant for DO sufficiently high for sensitive measurements. Depth profiles of the O₂ concentration near the water surface are obtained by a vertical laser light sheet at a rate of 185 frames per second. This technique is capable of visualising a measurement window of some centimetres down from the water surface with a resolution in the order of 50–100 µm. For a small circular wind-wave facility a correlation between wind speed and gas-transfer velocities calculated from the extracted mean boundary-layer thickness are presented and compared to the results of parallel measurements with a mass balance method for other gases with given Schmidt numbers.

An experimental study designed to determine free-surface characteristics of a turbulent flow, generated by jets that emerge from the bottom of the pool, is described. Particle Tracking Velocimetry at the free-surface was used to measure the velocity field, vorticity field and two-dimensional divergence (Hanratty’s β). While the high magnitudes of β did occur at similar locations to high surface vorticity, the scale of these maximal and minimal are approximately 1/3 of the scales of surface vorticity. This indicates that the scales of vorticity and β may not be correlated, or at least are correlated at a substantial multiplicative reduction. The relationship between β and liquid film coefficient is discussed. For flows where transfer is dominated by all frequencies of β, K _L ∼ (Dβ _rms )^1/2 is sufficient. However, there are flows where the large eddies are more important, and a spectra of β needs to be computed, according to McCready et al’s [12] relation K _L ∼ (DS _βmax )^1/2. In this case, a methodology is needed for transferring wave number spectra of β to frequency spectra in tanks without a significant flow, in order to estimate K _L . In our experiments, 1% of the turbulent kinetic energy was chosen to simulate similar results in a flume.

Transport processes at gas-liquid interfaces play a central role in many industrial and environmental systems. For example, such processes at the air-water interface impact evaporation of water from reservoirs, control of regional climate, evasion of carbon in tropical river systems, and the aeration of hypoxic water, to name a few. There has also been recent intense interest in this area associated with the transport of greenhouse gases and moisture between the atmosphere and terrestrial water bodies. But despite all this activity, recent estimates of oceanic carbon dioxide uptake still vary by factors ∼2. Since carbon dioxide is thought to be sequestered in the oceans at a rate equivalent to ∼40% of its generation rate from man-made sources, such uncertainties have obvious policy implications.

This poor state of knowledge arises primarily due to interfacial scalar exchange being controlled by near-surface turbulence, which is difficult to measure and simulate as interfaces move, deform, and sometimes break. However, recent advances in particle imaging velocimetry and numerical approaches have led to some progress in this area, which is discussed. In low winds, the large-scale turbulence structure on the liquid side of the surface is found to be quasi two-dimensional, consisting primarily of attached spiral eddies (whirlpools). With increasing wind, streaks and bursts appear, much like in wall turbulence, but with detailed differences in structure, e.g. interface-parallel intensities peak right at the surface. At even higher wind speeds, short length, ∼ \( \mathcal{O} \) (10 cm), interfacial waves start to break, qualitatively enhancing turbulence and scalar exchange rates. It is shown that over this whole range of conditions the mean surface-velocity divergence field is expected, for theoretical reasons, to be highly correlated with scalar transfer rates. Direct numerical simulations and experimental measurements appear to support this hypothesis. It is shown that the surface divergence field is related to the mean square wave slope, which may be remotely observed. Such remote measurements may then serve as scalar exchange surrogates, enabling estimation of reliable regional and global budgets.

The laminar-turbulent transition of the water surface boundary layer generated by a steady wind at the entrance of a wind-wave tank is investigated experimentally. Observations of the velocity field in water were made both by flow visualization techniques and laser Doppler velocimeter measurements. Two stages in the development of the perturbations have been clearly identified. First, the slow growth of streamwise longitudinal vortices embedded into the laminar flow is followed by a rapid development of secondary instabilities resulting in the pattern breakdown. The picture looks similar to the by-pass transition to turbulence in the boundary layers over rigid plate. At the second stage, peculiar to this free surface flow, an explosive deepening of the boundary layer and a fast development of inflexional instabilities occur inside localized areas. This phenomenon leads to an intense vertical mixing, which differs dramatically from the rigid plate scenario.

Over the last few years, compelling evidence has emerged that the exchange of low-solubility gases across air-water interfaces is strongly enhanced by microscale breaking (e.g. Jähne and Haußecker [12], Zappa et al. [28]). Jähne and Haußecker [12] observe that low-solubility gas flux rates are enhanced by up to a factor of 5 in the presence of small scale waves. Investigations using surface infrared imagery [10, 22, 27, 28] have demonstrated a strong correlation between total flux and a proportional area of surface with a high infra-red radiation emission associated with the passage of microscale breaking waves. The mechanisms causing this significant enhancement in exchange rate remain unclear. Zappa et al. [28] proposed that thinning of the aqueous diffusion sublayer by subsurface turbulence in the vicinity of the high infra-red emission region was primarily responsible for this enhancement. Alternate to this is a relationship between the air-water surface exchange rate and the passage rate of wind-forced microscale breaking waves proposed by Peirson and Banner [21]. They have suggested that subduction of the aqueous diffusion sublayer by the microscale wave spilling regions coupled with a weak surface divergence on the upwind faces of the waves primarily determines the microscale-breaking associated flux rate. We have completed a sequence of precise oxygen re-aeration measurements with the specific objective of testing the findings of Peirson and Banner [21]. Specifically, we have compared the flux rates of wind-forced, flat water surfaces in the absence of waves with those in the presence of wind-forced, steep, unbroken waves and wind-forced, microscale breaking waves. With the introduction of steep, unbroken micro-scale waves the surface exchange rate is enhanced by a factor of approximately 2.5. The transition from incipient breaking of the waves to the microscale breaking state induces a significant increase in the associated wind stress [1]. The observed rapid increase in flux rate is approximately proportional to the increase in the wind stress. For the microscale-breaking state, the observed flux rates show good agreement with the predictions of Peirson and Banner [21].

Breaking waves dissipate energy in the oceanic surface layer (top few meters) and also support the air-sea momentum flux. Spectrally resolved energy dissipation and momentum flux are extracted from open ocean observations of the breaking crest length distribution Λ(c). This concept, first introduced by Duncan and Phillips more than 2 decades ago, includes an unknown proportionality factor b. Independent estimates of direct turbulence measurements are used to evaluate this proportionality factor.

In order for air-sea exchange processes to be estimated in a dynamically consistent manner in the coupled atmosphere-ocean boundary-layer system, it is necessary to account for the dynamics of surface waves and other movements of the air-water interface. This is also necessary for the interpretation of turbulent flux observations made within the boundary layers, particularly those made from non-stationary measurement platforms. In recent years, considerable progress has been made in observational technology for the direct determination of the vertical flux of momentum, heat, and mass by eddy covariance techniques.

We present an approach to the study of atmosphere-ocean boundary-layer fluxes which employs a general time-dependent coordinate formulation. To provide a uniform treatment both above and below the water surface it is advantageous to use the instantaneous sea surface as a coordinate surface. Reynolds covariances for turbulent flux are then replaced by more complex expressions, and we explore the implications for the design of flux measurement systems and modelling of the coupled interfacial/boundary-layer system. We show that for the flux of trace substances, and other scalar quantities such as heat, measurements of averaged fluxes from moving measurement platforms are subject to biases which are proportional to the square of the amplitude of the platform displacement or of the wave slope, and we indicate how such biases may be corrected or allowed for.

We present results from numerical simulation of an aqueous turbulent boundary layer underneath a dynamic air-water interface and driven by wind stress and pressure. The simulation results reveal distinct surface and flow structures of micro-breaking wind waves, including a bore-like crest preceded by parasitic capillary waves riding along the forward face and elongated streamwise velocity streaks in the backward face, and confirm the observations in the laboratory and field experiments. The results also highlight the potential impacts caused by the short-wavelength capillaries on the gravity dominant free-surface flows, and consequently reveal the necessity in incorporating such microscale processes in the parameterizations of fluxes across the air-sea interface.

The effects of impinging raindrops on both turbulence below the airwater interface and CO₂ transfer across the air-water interface are discussed using laboratory measurements by Takagaki and Komori [1]. The measurements of CO₂ absorption rate and turbulence quantities in an open-channel flow show that impinging raindrops enhance both turbulent mixing near the free surface on the liquid side and CO₂ transfer across the air-water interface, and that the mass transfer velocity due to impinging raindrops is well correlated with the mean vertical momentum flux of raindrops. The reason why the mass transfer velocity is well correlated by the mean vertical momentum flux is explained by showing the instantaneous velocity vectors induced by a falling single droplet. Further, in order to clarify the effects of rainfall on the global and local CO₂ transfer across the air-sea interface, the mean annual net air-sea CO₂ flux was estimated using both the daily precipitation data set and the empirical correlation [1] between the mass transfer velocity and mean vertical momentum flux. The rainfall effects are also compared with wind shear effects. The results show that rainfall effects are significant for the local CO₂ budget between atmosphere and ocean in equatorial and mid-latitude regions, but are not so important for global budget, compared to the wind shear effect.

Studies deploying atmospheric flux-profile techniques in laboratory wind-wave tanks have been performed to demonstrate and verify the use of airside turbulent transport models and micrometeorological approaches to accurately determine air-water gas transfer velocities. Air-water gas transfer velocities have been estimated using the CO₂ atmospheric flux-profile technique in laboratory wind-wave tanks both at the NASA Wallops Flight Facility, USA and Kyoto University, Japan. Gas fluxes using the flux-profile and the waterside mass balance techniques have been reconciled. Air-water fluxes of H₂O and momentum were also measured simultaneously in a linear wind-wave tank. The waterside mass balances used the evasion of SF₆. The CO₂, H₂O, and momentum fluxes were calculated using the atmospheric flux-profile technique over a wind speed range of 1 to 14 m s⁻¹. The CO₂ and H₂O atmospheric profile model uses airside turbulent diffusivities derived from momentum fluxes. These studies demonstrate that the quantification of air-water CO₂ fluxes using the atmospheric flux-profile technique can be implemented in the laboratory. The profile technique can be used to measure an air-water flux in much less time than a mass balance. Effects of surfactants, wind speed, and wind stress on air-water transfer are also explored using the flux-profile technique. Validation of the air-water CO₂ gas exchange in laboratory wind-wave tanks provides evidence and support that this technique may be used in field studies.

Measurements at the air-water interface were made with a high spatial and temporal resolution microthermometer in the Air-Sea Interaction Saltwater Tank (ASIST). ASIST is a linear wind-wave tank of dimensions 15×1×1 m, and is constructed from transparent acrylic sheets. During the experiments, the air-water heat flux was varied by changing the air-water temperature difference (ΔT ^AW ), and the wind speed was varied with the ASIST fan. The microthermometers were mounted on a J-shaped support, which was attached to a linear servo motor which was driven in the vertical direction. This allowed profiles to be made from a depth of about 13 cm to the surface. Data was acquired for three ΔT ^AW regimes (−5, +10, and +15°C) with wind speeds from 1 to 9 ms⁻¹. Data for a total of 8 runs were acquired and provided estimates of the boundary layer thickness (δ _c ) and the skin-bulk temperature difference (ΔT ^SD ). We found that the relationship between ΔT ^SD and wind depended on the ΔT ^AW regime. There was a clear relationship between δ _c and wind (μ), and we derived the empirical expression δ _c = 0.35 + 4.9 _e ^−u based on the data. Comparison with previous estimates of δ _c based on measurements and rigid wall boundary layer theory showed that our values were lower by a factor of 2 at low wind speeds, but were in good agreement at 10 ms⁻¹.

The thermal structure of an air-water interface is investigated by examining thermal imagery obtained from a high resolution infrared (IR) sensor. The experiments were performed at the ASIST facility at the University of Miami for wind speeds ranging from approximately 2 ms⁻¹ to 10 ms⁻¹ and for flux based Richardson numbers ranging from about 10⁻² to 10⁻⁵. Two cases were examined: (1) the so-called cool-skin case where the water surface was significantly cooler than the bulk water temperature and (2) the warm-skin case where the water surface was warmer than the bulk. In the cool-skin case, the low wind speed results reveal a cellular structure reminiscent of earlier results in which the lateral length scale of the cells (or fish-scales) varies as the inverse of the friction velocity. The imagery clearly reveals the progression from non-breaking gravity waves, to a system of omnidirectional breaking which seems to create a nearly isotropic surface temperature field. Though no wind waves were present at low wind speeds, the thermal imagery reveals the existence of persistent, highly coherent, Langmuir-like cell structures which were marked by surface convergent zones in which ambient surfactant may have accumulated. Imagery obtained for the case in which the water-side thermal boundary layer is stable constitutes a novel aspect of this work. In this warm-skin case, the cellular (fish-scale) structure appears as it does in the unstable case, strongly suggesting that these small scale features are due to shear instabilities in the surface layer. In addition, they are more clearly revealed as the natural convective instability of the thermal boundary layer is suppressed. This appears to reduce the appearance of the smallest scales of surface turbulence.

A novel technique is presented that makes it possible to measure the viscous shear stress τ_μ from active thermography. With a CO₂ laser, patterns are written to the sea surface. This temperature structure is distorted by the linear velocity profile in the viscous boundary layer. Due to the non-zero penetration depth of both the laser and the infrared camera, this vertical velocity profile can be resolved. By resolving the velocity profile, the viscous shear stress can be extracted from the recorded image sequences. At the same time, the flow field at the water surface can be measured accurately. Estimating both quantities is only possible by modeling the imaging process as well as the velocity profile in the boundary layer. The model parameters can then be computed in a standard parameter estimation framework. This novel technique was tested both on simulated data and on measurements conducted in a small annular wind-wave flume. The friction velocity computed in this fashion compared favorably to independent measurements. Although not tested yet, this technique should be equally applicable to field measurements.

Water surface infrared images were obtained during the GASEX2001 experiment in the South Equatorial Pacific waters and during the laboratory experiment in the AEOLOTRON wind wave tank at University of Heidelberg in October 2004. The infrared imagery during these experiments reveals coexistence of roller type turbulence and intermittent breaking events. Previous interpretations of the infrared images relied on the surface renewal model, in which the water surface is assumed to be occasionally renewed by bursts of turbulent eddies reaching the water surface. A new complementary model (eddy renewal model) based on stationary and spatially periodic turbulent eddies is developed to reinterpret the infrared images of near surface turbulence. The model predicts warm elongated patches bounded by cold streaks aligned with mean wind, being consistent with field and laboratory infrared images. The model yields bulk temperature estimates and mean heat flux estimates that are very close to those based on the surface renewal model.

This paper theoretically investigates the influence of intermittency on determining average transfer velocities using different measuring techniques. It is shown that all measuring techniques can significantly be biased by intermittency. Mass balance and eddy correlation measurements are only biased when the concentration difference between the air and the water is spatially or temporally inhomogeneous over the measurement interval. Mean transfer velocities calculated either from mean boundary layer thicknesses or from thermographic techniques, which compute the mean transfer velocity either from concentration differences of from time constants, are biased toward lower values. The effects can be large and a simple stochastic bimodal model is used to estimate the effect.

The processes at the ocean surface play an important role in the exchange of carbon dioxide (CO₂) to and from the atmosphere. Despite this, some fundamental mechanisms that control the rate of interfacial transfer have not been well resolved. Globally, the ocean absorbs a significant amount of atmospheric carbon, however, the ocean has strong local sinks and sources. The carbon that is exchanged between the ocean and atmosphere is primarily as a dissolved gas that is not highly soluble in water. This means that molecular diffusion and turbulent mixing at the aqueous boundary layer are the significant transport processes controlling CO₂ exchange rates across the surface. The aqueous boundary layer and the corresponding CO₂ exchange rate are affected by the local climate and environmental conditions in both the atmosphere and ocean. For the first time, the turbulent transport through the marine atmospheric boundary layer has been measured in the North Atlantic and the Equatorial Pacific using micrometeorological techniques to quantify the CO₂ flux. Measurements also provided an understanding of the environmental conditions controlling the exchange rate. The North Atlantic was a large sink of atmospheric CO₂ with a high variability in flux ranging from 1.2–4.2 mol m² yr⁻¹ that was forced primarily by the wind. The Equatorial Pacific was a strong source of atmospheric CO₂. Here, the CO₂ flux ranged from 3.0–4.2 mol m² yr⁻¹ and was forced primarily by diurnal cycles.

The advection of horizontal inhomogeneous CO₂ concentrations by a wind field with vertical gradient causes a height dependent vertical CO₂ flux. It is shown that the corresponding bias between the flux at few meters above the surface and the flux through the surface is in the order of the natural CO₂ flux variability through the air-sea interface, if horizontal CO₂ gradients are assumed that are typical for land conditions or epicontinental seas. Although this bias is a zero mean effect on the long term, it is suspected that it may contribute significantly to the estimation uncertainty of air-sea CO₂ transfer velocities based on eddy correlation flux measurements. A simple model is suggested to retrieve the surface flux from extrapolation of flux measurements at two height levels.

Fluxes of primary marine aerosol in the sub-micron fraction were measured using a flux package consisting of a sonic anemometer, a Condensation Particle Counter (CPC) and an optical particle counter (OPC) equipped with a heated inlet. Whereas the CPC provides the total particle number flux of particles larger than 10 nm, the OPC measures size segregated fluxes for particles between 0.5 and 5 μm radius. By varying the temperature of the OPC inlet, particles of certain composition can be selected. Results are presented with the inlet temperature at 100°C (dry particles retaining most of the aerosol material, representative of sea spray aerosol) or at 300°C which volatilizes all material except sea salt and elementary carbon. Temperature scans confirm the choice of these temperatures to discriminate between sea spray and sea salt. This flux system was deployed at the end of the 560 m long pier of the US Army Corps of Engineers Field Research Facility in Duck (NC, USA). Initial results show the increase of the sea spray fluxes with wind speed u, roughly varying as u ³ for u up to 16 ms⁻¹.