Investigations of Air-Sea gas Exchange in the CoOP Coastal Air-Sea Chemical Exchange Project

Author Posting. © Oceanography Society, 2008. This article is posted here by permission of Oceanography Society for personal use, not for redistribution. The definitive version was published in Oceanography 21, 4 (2008): 34-45.

AbStr ACt.The exchange of CO 2 and other gases across the ocean-air interface is an extremely important component in global climate dynamics, photosynthesis and respiration, and the absorption of anthropogenically produced CO 2 .The many different mechanisms and properties that control the air-sea flux of CO 2 can have large spatial and temporal variability, particularly in the coastal environment.The need for making short-time-scale and small-spatial-scale estimates of gas transfer velocity, along with the physical and chemical parameters that affect it, provided a framework for the field experiments of the Coastal Ocean Processes Program (CoOP) Coastal Air-Sea Chemical Exchange (CASCEX) program.As such, the CASCEX project provided an opportunity to develop some of the first in situ techniques to estimate gas fluxes using micrometeorological and thermal imagery techniques.The results reported from the CASCEX experiments represent the first step toward reconciling the indirect but widely accepted estimates of gas exchange with these more direct, higher-resolution estimates over the coastal ocean.These results and the advances in sensor technology initiated during the CASCEX project have opened up even larger regions of the global ocean to investigation of gas exchange and its role in climate change.

gAS ExChANgE IN COAStAL ENVIrONmENtS
The exchange of CO 2 and other gases across the ocean-air interface is an extremely important component in global climate dynamics, photosynthesis and respiration, and the absorption of anthropogenically produced CO 2 .The many different mechanisms and properties that control the air-sea flux of CO 2 can have large spatial and temporal variability, particularly in the coastal environment.
For example, the surface concentration of CO 2 can vary with temperature, salinity, bubble entrainment, physical mixing, and biological productivity.Also, gas transfer rates can vary with sea state and surface chemical enhancement.
Laboratory experiments (e.g., Liss and Merlivat, 1986) show that gas exchange can be reasonably modeled using where F G is the gas flux, K G is the gas transfer velocity, G w and G a are the gas concentrations in bulk seawater and in the air, and H is the appropriate Henry's law solubility constant (McGillis et al., 2000;Hintsa et al., 2004).This approach is closely related to the bulk flux method (e.g., Fairall et al., 1996Fairall et al., , 2003;;Hare et al., 2004) in which, where C G is a dimensionless transfer coefficient for gas exchange and U r is the wind speed relative to water.Therefore, we would expect K G to vary at least linearly with wind speed.Studies also show that K G varies with oceanic and atmospheric stability, sea state, Schmidt number (a gas-dependent function of temperature and salinity), and a host of surface processes (e.g., Liss, 1983;Liss and Merlivat, 1986;Liss et al., 1988;Brutsaert and Jirka,1984;Wilhelms and Gulliver, 1991;Jähne and Monahan, 1995;Donelan et al., 2002;McGillis et al., 2004a;Borges and Wanninkhof, 2007;Garbe et al., 2007).
In from 12-24 cm hr -1 (Peng et al., 1979;Broecker and Peng, 1982;Jähne, 1985;Broecker et al., 1986).Although a few estimates over the Great Lakes by Donelan and Drennan (1995) were more in line with these estimates, most previous estimates of transfer velocity using direct covariance (eddy correlation) flux estimates were orders of magnitude larger (e.g., Jones and Smith, 1977;Wesely et al., 1982;Smith and Jones, 1985).Although physical processes may have been responsible for some of this discrepancy, Broecker et al. (1986) provided compelling evidence that CO 2 sensors available at the time were simply not sensitive enough to measure the CO 2 signal.

A fIrSt StEP tOwArd COmmUNIty CONSENSUS
The  The ship-based systems included a high-resolution infrared imaging system capable of estimating the air-sea gas exchange, and a direct covariance flux system (DCFS) (Edson et al., 1998).
Vertical profiles of DMS were measured from both the ship and LADAS to provide additional estimates of the gas flux using the profile method and a newly developed micro-gas chromatograph (micro-GC) (Dacey et al., 1999;McGillis et al., 2001a).The catamaran carried systems to quantify the surface wave characteristics using a scanning laser slope gauge and a wave-wire array (Hara et al., 1998), and an additional meteorological package to quantify the near-surface atmospheric turbulence.A fluorometry package measured microlayer enrichments of colored dissolved organic matter (CDOM) (Frew et al., 2004).(1996) and Jähne and Haußecker (1998); those from the CoOP '97 cruise were published by Schimpf et al. (2004) and Frew et al. (2004).These investigations relied heavily on the use of thermal imagery to compute the transfer velocity using surface renewal theory (Schimpf et al., 2004).These estimates were in good agreement with previously published wind-speed-dependent parameterizations of transfer velocity.(Edson, 2001).The direct covariance method correlates the gas concentration with the vertical velocity typically measured using sonic anemometers to provide a direct estimate of the flux: where the overbar represents a time average, and w' and G' are the fluctuating components of the vertical velocity and gas of interest, respectively.The profile method assumes that the gradient is directly proportional to the flux: where z is the height above the surface and K M is the variable of proportionality known as the eddy diffusivity for mass.
The eddy diffusivity is typically determined using parameterizations that rely on Monin-Obukhov similarity theory (e.g., Edson et al., 2004).
Another important consideration in accurately measuring these fluxes concerns the nature of the gas measurement itself.For example, if the gas sensor measures the mixing ratio, r G (mass of gas per mass of dry air), then the fluctuating where w is commonly estimated using the approach of Webb et al. (1980) and includes terms to account for the sensible and latent heat fluxes.The community now refers to this approach as the Webb correction.A recent review of the Webb correction and similar approaches is given in Fuehrer and Friehe (2002).
A correction of this sort is required for all forms of micrometeorological methods, for example, the direct covariance, profile, and bulk methods (see Webb et al., 1980;Fairall et al., 1996).Although this term often represents a negligible correction for water vapor fluxes, it is often a significant correction for CO 2 fluxes.For example, Figure 2  Our approach in the CoOP project and subsequent field programs has been to measure the moisture and temperature fluctuations and "dry" the sample numerically (McGillis et al., 2001b).

OCEANIC APPLICAtIONS
Application of these techniques to measure CO 2 flux over the ocean is a challenge due to the low signal levels in the marine atmosphere (i.e., very small r C fluctuations).The small signal is a direct consequence of CO 2 being in near equilibrium because of its long lifetime in the atmosphere and because it is a waterside-controlled gas (Liss and Slater, 1974;McGillis et al., 2000).The result is very small gradients in the atmospheric surface layer, even in regions of large The left panel shows that the correction is two to three times larger than the true flux under low relative humidity conditions and large latent heat fluxes.The right panel shows that the correction is considerably smaller at near neutral conditions, but is not negligible even at a relative humidity of 95%.
was paired with the DCFS (Edson et al., 1998)   Later experiments simply collected the gas in Tedlar bags, which allowed the investigators to store the DMS without loss for extended periods until it could be concentrated and analyzed in a more controlled laboratory setting (Zemmelink et al., 2002).The success of these experiments led to the development of more sophisticated profiling systems deployed during the Fluxes, Air-sea Interaction, and Remote Sensing (FAIRS) (Hintsa et al., 2004) and GasEx-2001experiments (Zemmelink et al., 2004).In these subsequent experiments, the use of Tedlar bags to collect DMS samples allowed investigators to implement the relaxed eddy accumulation method.In this method, air is conditionally sampled into two separate bags, one containing the air samples measured in updrafts (w' > 0) and the other in downdrafts (w' < 0).Therefore, this method is essentially using the vertical velocity measurements shown in Figure 3 and sampling the air into one bag when the fluctuations are above the w' = 0 line and into another bag when they are below the line.After a given time (usually 10-30 minutes), the gas concentrations in the two samples are measured, and their difference is proportional to the flux: where is the standard deviation of the vertical velocity, the concentrations in the updrafts and downdrafts, respectively, and is a calibration coefficient.This approach was successfully used to estimate fluxes during the FAIRS and GasEx-2001 experiments and showed reasonably good agreement with fluxes determined using the profile method (Zemmelink et al., 2004;Hintsa et al., 2004).
The most recent advance in the measurement of DMS flux over the ocean has involved the implementation of atmospheric pressure ionization mass spectrometry (APIMS).APIMS detectors have developed to the point where they can continuously measure DMS concentration at high flow rates to estimate the flux using the direct covariance method (Bandy et al., 2002;Huebert et al., 2004;Marandino et al., 2007).The successful implementation of the direct covariance technique for DMS opens up a much larger portion of the global ocean to investigation of gas exchange due to its larger signal.

OCEANOgr APhIC SENSOrS ANd tEChNIQUES
Waterside measurements provide the concentration difference required to compute the gas transfer velocity when combined with any of the above figure 4. The LAdAS platform just before it is lowered into the ocean, showing airside and waterside instrumentation, including, again, the four dmS samplers and two sonic anemometers.The microlayer sampler is visible as two small pontoons cascaded between the larger pontoons.The white canister above and yellow torpedo-shaped device below make up the scanning laser slope gauge.Also visible but less distinct are components of the wave-wire and subsurface-AdV arrays.There is considerable interest in quantifying the air-sea exchange of CO 2 in coastal ecosystems.As repositories of terrestrial organic carbon that cannot be accounted for in sediment deposits, early researchers presumed that coastal waters were regions of significant CO 2 efflux to the atmosphere.however, recent global syntheses of direct estimates of CO 2 air-sea fluxes on continental shelves generally suggest that shelf waters are net sinks for atmospheric CO 2 globally (Cai et al., 2006;borges et al., 2005), removing roughly 0.3 Pg C a -1 , while nearshore and estuarine waters are generally a net source.

Photo by Sean McKenna
The exclusive focus on air-water exchange, however, overlooks the potential contribution from intertidal plants, such as salt marsh grasses and mangrove trees, which remove CO 2 directly from the atmosphere and release a significant portion directly to coastal waters through losses from roots and by the degradation of leaf litter and other detritus (figure 1).A significant portion of this carbon will not immediately escape back to the atmosphere.rather, it may be transported later-

SUmmAry
The CoOP-sponsored CASCEX project produced a number of important scientific findings on many of the processes that control gas exchange, for example, the role of gravity-capillary waves (Hara et al., 1998), surface renewal (Garbe et al., 2004), surfactants (Frew et al., 2004), and atmospheric forcing (Schimpf et al., 2004) 1994, the Coastal Ocean Processes (CoOP) Program supported a study to examine the exchange of CO 2 and other climate-and weather-relevant compounds (CWRC) in the coastal ocean.The study was in recognition of the growing concern for the impact of greenhouse and other gases on climate and of the key role the coastal ocean may play in processing these gases.Therefore, CoOP initiated the Coastal Air-Sea Chemical Exchange (CASCEX) project, which involved field experiments off the US west coast at Monterey, California, and off the east coast in the Mid-Atlantic Bight.The project concentrated on investigations to improve our ability to estimate CWRC fluxes in coastal regions.In particular, the project provided an opportunity to develop some of the first in situ sensors to estimate the fluxes on shorter temporal and spatial scales using micrometeorological and thermal imagery techniques.Improvement of these techniques was critical to investigating the processes governing gas exchange in the coastal environment where biogeochemical and physical processes act on shorter time scales than in the open ocean.Specifically, although the coastal ocean's role in the uptake of anthropogenic CO 2 is widely debated (Borges, 2005), its importance in processing the carbon from terrestrial sources (e.g., through rivers, runoff, and groundwater) and exporting carbon to the open ocean has been well established (Ducklow and McCallister, 2004).Additionally, there was a need in the gas-exchange community to either reconcile or explain the significant differences found between the micrometeorological and indirect, but better established, estimates of gas transfer velocities.Prior to the CoOP experiments, determining the rate at which CO 2 was exchanged between the ocean and atmosphere was a very controversial subject.Specifically, indirect methods such as the radiocarbon, radon, and radioisotopic methods give long, timeaveraged transfer velocities ranging of in situ instruments to measure gas transfer velocity; momentum, heat, water vapor, dimethylsulfide (DMS), and CO 2 fluxes; directional wave spectra; surface microlayer chemical enrichment; and near-subsurface turbulence on the short time scales commensurate with natural forcing variability.Measurements were made with shipbased instrumentation and with a newly developed, remotely operated research catamaran known as LADAS (Laser Slope Gauge/Acoustic Doppler Current Profiler/Anemometry Package/Surface Sampler).Figure 1 shows LADAS being deployed from R/V Oceanus.
Subsurface instrumentation included an array of acoustic current meters to measure the near-surface current profile and various turbulence statistics.These field experiments represent the first coincident measurements of atmospheric forcing, surface chemical enrichment, small-scale surface roughness, and gas transfer rates, allowing several aspects of gas exchange mechanisms to be studied.Results from the CoOP '95 cruise were published by Haußecker

Frew
figure 1. deployment of the LAdAS platform from r/V Oceanus during the 1997 CASCEx cruise.The instrumented mast on the starboard side of LAdAS supports two sonic anemometers and four samplers with tenex traps for dmS profiles.LAdAS was released from its tether and remotely controlled to operate beyond the wake of the ship.Photo by Sean McKenna where a is the mean density of dry air.However, many gas analyzers measure the density of the gas, G , in mass or moles of gas per volume of moist air.When the density is used to estimate the flux,Webb et al. (1980) showed that the flux must be corrected for the density variations caused by water vapor and temperature.The need for this correction can be illustrated by considering an upward buoyancy flux of warm, moist air from the ocean surface.This air is typically replaced by cooler, dryer (denser) air from aloft.Therefore, a small mean vertical velocity must be present to preserve the assumption of zero mean vertical mass flux of dry air(Fuehrer and Friehe, 2002).To see how this vertical velocity affects the gas flux, we consider a perfectly level and accurate anemometer over a horizontally homogeneous surface.In this setting, such an ideal device would be able to measure the appropriate vertical mean and the fluctuations about it.This vertical velocity could then be correlated with the gas density to estimate the flux, i.e., F G = w G .Even when the ocean is considered horizontally homogeneous, this approach is not practical in the marine surface layer where platform motion and flow distortion make it nearly impossible to measure the true mean vertical velocity from real sensors.Therefore, the direct covariance fluxes are computed by correlating the fluctuations about their respective means and applying a correction for the missing vertical velocity contribution: shows the magnitude of the correction over an oceanic CO 2 sink region, where the Webb correction reduces the measured flux in the presence of a positive latent heat flux.The correction can be substantially larger than the flux itself in regions of strong heat fluxes.In fact, the correction can change the sign of the measured versus actual flux over CO 2 source regions with large positive heat fluxes.This observation has led investigators to reduce this correction by physically drying air samples before they reach the detector to remove the moisture.This step allows the sensors to directly measure a concentration that can be more readily converted to the mixing ratio, but at the expense of sensor response.

∆pCO 2 .
Figure 6 in McGillis et al., 2004b) in the upper left-hand panel next to a simulation of the resulting CO 2 fluctuations normalized by the mean atmospheric concentration.To resolve the profile requires a precision of 0.1 ppm, which is less than 0.1% of the mean atmospheric concentration.As such, vertical gradients and fluctuations of CO 2 in the marine surface layer are extremely difficult to resolve with sufficient accuracy to estimate the flux using the profile method over the ocean.The small flux also results in a very low signal-to-noise ratio for CO 2 , which makes it difficult to resolve the turbulent fluctuations required for the direct covariance method.For example, the standard deviation of the fluctuations shown in figure 3. typical profiles of CO 2 and dmS at right and simulations of the turbulent fluctuations of CO 2 , dmS, and vertical velocity based on these profiles at a wind speed of 10 m s -1 .The CO 2 profile is representative of a sink region with a bulk air-sea concentration difference of 100 ppm (i.e., the waterside bulk concentration is approximately 278 ppm).The dmS profile is representative of the typical oceanic source of dmS found over the ocean.The CO 2 signal is negatively correlated with the vertical velocity signal while dmS is positively correlated, indicating the expected downward and upward fluxes, respectively.The dmS flux is uniformly positive over the world's ocean.The downward CO 2 flux shown here is most commonly found at high latitudes, while upward fluxes are expected over the source regions commonly found at lower latitudes.The fluctuations are plotted after normalization by their respective mean concentrations at a height of 10 m.The figure clearly shows how small the CO 2 fluctuations are relative to the mean concentrations, while the dmS fluctuations are orders of magnitude larger compared to their means.On the other hand, the mean concentrations for dmS are orders of magnitude smaller than CO 2 as indicated by the profiles.
results and a range of wind speed-gas exchange parameterizations.
figure 5. gas-transfer velocity versus wind-speed measurements from CoOP and gasEx-1998.A cubic wind speed relationship from mcgillis et al. (2001b), with a finite zero-wind-speed offset, is fit to this data and shown for comparison.transfer velocity versus wind speed relationships based on 14 C budgets (solid line) and on lake and wind-tunnel experiments (dotted line) are shown for comparison.
many gases of interest.Fortunately, the sources of velocity contamination due to platform motion and waves can be significantly reduced by averaging the waterside gas concentration at a given depth over many wave periods.Because both CO 2 and DMS are predominantly under waterside transport control, they often exhibit large gradients just beneath the ocean surface.This condition makes it possible for wave-following moorings and drifters to resolve the near-subsurface gradient, which can then be used with the micro-oceanographic profile method to estimate the flux (McGillis et al., 2004b).Considerably slower sensors can also be used, such that the analytical demands are greatly relaxed for these measurements.These micro-oceanographic approaches have become an important component of the GasEx and other field programs.Their flux estimates provide an excellent complement to micrometeorological measurements for investigating the processes that drive air-sea exchange.
on their transfer.The CASCEX project is also notable for the technological advances in gas exchange sensors and techniques that were developed.For example, CASCEX investigators pioneered the use of micro-GCs to measure DMS profiles and estimate DMS flux.This technique was refined and successfully deployed during the 2000 FAIRS experiment to estimate DMS fluxes alongside a relaxed eddy flux system.The CASCEX cruises also marked the initial deployment of a system using the controlled flux technique at sea.This system helped spur the development of other active and passive IR techniques for use during GasEx and FAIRS.These CASCEX studies also spurred innovations in dissolved gas measurement technology on the ocean side of the air-sea interface.These innovations include the development of in situ pCO 2 sensors, low-power and small-sized gas equilibrators, total gas tension sensors, and rapid, high-sample-throughput systems for shipboard analysis (e.g., for DMS).Not only are these measurements required for calculating gas transfer rates from airside direct flux measurements, but they also have advanced our understanding of the processes that control dissolved gases.Further advances in rapid-response, in situ chemical sensors (e.g., for O 2 ) are making waterside direct flux measurements possible.CASCEX provided an important step toward reconciling the indirect but widely accepted estimates of gas exchange with more direct estimates based on micrometeorological approaches and thermal imagery.Micrometeorological techniques have been used subsequently to estimate CO 2 fluxes and transfer velocities over regions of the global ocean where ∆pCO 2 is large.For example, the fluxes from the GasEx campaigns are being used to improve our understanding of how high winds, sea state, and oceanic stratification impact these parameterizations.Significant progress has also been made in our ability to estimate DMS fluxes using micrometeorological techniques, including the direct covariance approach.These and other advances in sensor technology are opening up even larger regions of the global ocean to investigation of gas exchange and its role in climate change.