Ecosystem processes such as respiration, denitrification, and methanogenesis may be quantified by measuring the flux of gases such as oxygen, carbon dioxide, nitrous oxide, and methane. For instance, we have used changes in O2 or CO2 concentrations in the water column to yield estimates of whole system metabolism in the Parker River Estuary. These fluxes must be corrected for the non-biological or diffusive flux of the gas of interest between the water and the atmosphere. Corrections are typically made by using published relationships between wind speed and a factor known as the gas transfer velocity, k. The mass flux (F) of a gas across the air-water interface is equal to the gas transfer velocity multiplied by the difference between the surface (Csfc) and equilibrium concentrations (Ceq) of the gas [F= k (Csfc-Ceq ) ] For large rivers and estuaries, k has been empirically derived using either the "dome" technique, which measures gas fluxes into a floating dome, or by injecting inert gases as tracers into the water column and directly measuring loss to the atmosphere. Ideally, these measurements are made over a variety of wind conditions. Regression relationships between the gas transfer velocity and wind velocity are then developed and used for predicting k during gas flux measurements.
Published relationships between gas exchange rates and wind speed derived from studies done in deep, wide water bodies with long fetches may not be applicable to tidal systems such as the upper Parker River which are narrow, shallow, and serpentine. The wind acting on such a system does not have a uniform effect: a) the serpentine nature of the Parker results in short reaches of river oriented in all directions, a morphometry that produces a generally short fetch; b) a three meter tide range in this narrow channel results in two vastly different situations at high and low tides, such that at high tide the river surface is somewhere near the same level as the marsh surface and exposed to the wind, and at low tide, the river surface is shielded from the wind by the high marsh banks. Another consideration in shallow systems is that currents and bottom topography may contribute significantly to surface turbulence through bottom shear, an effect not accounted for by wind speed.
To avoid the surface turbulence effects associated with the dome technique, we chose to directly measure gas exchange across the air-water interface using sulfur hexafluoride gas (SF6) as a tracer. SF6 occurs in nature at extremely low concentrations (<10-15 M), is chemically and biologically nonreactive, and is easily detectable at these very low levels. SF6 has been used successfully in open ocean, lake, and river applications.
In July, 1996, at a slack high tide, we injected ~0.004 moles of SF6 into the upper Parker River estuary. We added rhodamine dye at the same time in order to locate the tracer plume with an in situ fluorometer. The tracer was allowed to mix for one tidal cycle and we then sampled SF6 concentration in the water at each of seven successive high tides, and once again four days later. Surface water samples were drawn into 100 ml glass syringes, transported submerged in river water to a field laboratory, and analyzed within 6 hours using gas chromatography with electron capture detection. Wind speed was meaured using a continuously recording anomometer installed on the high marsh adjacent to the tracer addition site. Precipitation data were provided by the National Weather Service and were measured at a nearby monitoring station.
After the initial tidal cycle the tracer plume measured 5.2 km in length, and the distribution was approximately gaussian. The distribution of the tracer changed over time in relation to tidal and current mixing (Fig. 1). The total mass of SF6 in the estuary was calculated by integrating concentration and water cross-sectional area (as determined by several earlier surveys) by estuarine distance. The difference in mass from one high tide to the next was attributed to loss to the atmosphere. The gas transfer velocity, k, and mass are related by: k=ln(Mt/Mt-1)h/t, where Mt is the current measured mass of SF6 in the estuary, Mt-1 is the mass of SF6 measured at the previous high tide, h is the average water depth, and t is the time between samplings.
Fig. 1. Concentration and distribution of SF6 as measured from the Parker River Dam at seven successive high tides.
Calculated values for kSF6 ranged from 1.1-6.2 cm·h-1. Wind speed during the days of our study were low, ranging from about 0.3 to 2.1 m s-1. We found a linear relationship between k and wind speed for four out of seven of our data points (Fig 2). This agrees with previous studies which have found good correlations with wind speed. Three data points, however, showed higher ks than would be predicted by wind speed (Fig 2). All of these higher values were associated with rain events during or just prior to the sampling period.
Figure 2. Linear regression of k vs wind velocity, showing points off the line that were related to precipitation.
In order to compare our results to other studies, we converted our ksf6 to a k for oxygen using kO2 =kSF6 (ScSF6 / ScO2 )2/3 where Sc is the Schmidt number for the respective gas. [The Schmidt number is defined as the kinematic viscosity of water divided by the molecular diffusion coefficient of the gas.] For a given wind velocity, our ks were lower than those from established dome studies (Fig. 3). A slight tendency for dome measurements to overestimate exchange rates when compared to other techniques has been noted in the literature. Whether this is a bias introduced by the method or whether the difference is due to the application of the dome-derived relationships to shallow systems is unclear. However, in small, shallow estuarine systems, and especially those with complicated morphometry, direct measurement of gas exchange using a tracer such as SF6 may be the more appropriate technique.
Figure 3. k(O2) measured by the dome technique in the Hudson River and by SF6 evasion from the Parker River.
As an additional note, we would like to suggest that overlooking the effects of precipitation on surface turbulence may lead to errors in determining gas exchange relationships.