Abstract

The amount of CO2 exiting headwater streams through degassing plays an important role in the global carbon cycle, yet quantification of CO2 degassing remains challenging because of the morphology of headwater streams and because of uncertainty about whether floating or suspended chambers provide valid measurements in moving water. We show that experiments using large and small floating chambers in flowing water over a moderate range of water velocities (0.13–0.23 m s−1) in a laboratory flume resulted in similar k600s to published field measurements with similar water velocities. We confirmed the flume experiments with paired stirred-still beaker experiments, where resulting k600s fell within the extrapolated trend of the flume experiments. We propose that the floating chambers can provide good estimates of CO2 degassing, particularly in shallow, low-velocity, morphologically complex headwater streams, permitting quantification of this important contributor to the global carbon cycle.

HIGHLIGHTS

  • Headwater streams contribute significantly to the C efflux from inland waters, yet the size of their contribution has only recently begun to be quantified.

  • Floating chambers, to date, are only generally accepted for use in relatively still waters (lakes, oceans), yet are increasingly used in characterizing headwater streams. Thus, we designed laboratory experiments to compare CO2 degassing using a floating chamber in a flume, with other methods of quantifying degassing rates.

  • In our experiments, k600-CO2 values varied linearly with water velocities from 0.13 to 0.23 m s−1, velocities that are typical of headwater streams. This confirms the importance of water velocity-induced turbulence to k600-CO2. It was not clear whether floating chamber design strongly affected the k600-CO2 results.

  • We compared CO2 degassing in floating chambers to calculated CO2 degassing based on the pH increase in stirred beakers of flume water, and found that the beaker experiments, although at lower water velocities, fell on the extended trend of the flume experiments.

  • These results support the use of floating chambers to characterize CO2 degassing in headwater stream type environments.

INTRODUCTION

Headwater streams and rivers have recently been recognized as significant sources of carbon to the atmosphere (e.g., Cole et al. 2007; Battin et al. 2009; Butman & Raymond 2011; Hotchkiss et al. 2015; Crawford et al. 2017; Marx et al. 2017; Ulseth et al. 2019). The magnitude of degassing from streams and rivers is quantified globally in only a few studies (Liu et al. 2008; Battin et al. 2009; Aufdenkampe et al. 2011; Marx et al. 2017), and more direct and site-specific measurements of gas efflux are needed to improve these estimates. Numerous methods exist for obtaining estimates of the flux of gases in environments with flowing water (Hall & Ulseth 2020). These techniques include wind-based estimates (Raymond & Cole 2001), the use of gas tracers as a proxy for the gas of interest (e.g., Genereux & Hemond 1992; Schelker et al. 2016), estimates based on hydrogeomorphic equations (Streeter & Phelps 1925; Owens et al. 1964; Raymond et al. 2012), chamber-based techniques (Crawford et al. 2013), and more (Neal et al. 1998; Morse et al. 2007). Newer methods include using a headspace equilibration method to determine atmospheric and aqueous CO2 (Schade et al. 2016), floating chamber with CO2 sensors and acoustic Doppler velocimeters to correct for turbulence-induced degassing (Ribas-Ribas et al. 2018), floating chambers with automatic chamber purging that allows continuous measurements (Martinsen et al. 2018), injecting CO2 and using upstream and downstream CO2 sensors to determine gas transfer velocities (McDowell & Johnson 2018), and using dissolved inorganic carbon and stable carbon isotopes to quantify degassing without using a gas transfer coefficient (Marx et al. 2018). Many calculate the gas transfer coefficient rather than directly measuring it (e.g., Aho & Raymond 2019; Horgby et al. 2019).

Chamber-based methodologies have proven a logistically simple and economic method of measuring gas efflux from lentic and ocean systems (Kremer et al. 2003; Cole et al. 2010; Mazot & Bernard 2015). However, controversy remains about the accuracy of chamber measurements under certain conditions (Matthews et al. 2003; Vachon et al. 2010), and limited testing has been conducted on chambers’ effectiveness in shallow, turbulent, flowing water environments (Cole et al. 2010; Lorke et al. 2015). Despite a lack of developed evidence supporting the use of chamber-based techniques, there have been numerous studies using them to measure gas fluxes in environments with turbulent flowing water (e.g., Bastien & Demarty 2013; Denfeld et al. 2013; Crawford et al. 2015).

The floating chamber works by encapsulating a portion of the water surface with an impermeable chamber and measuring the change in concentration of a gas of interest inside the chamber over time. The chamber is supported by flotation devices in the water (i.e., floating chamber, floating dome, floating helmet, or static chamber method) or suspended directly over the water. Different chamber designs have been shown to yield different results (Lambert & Fréchette 2005), making chamber design an important variable.

There are two principal limitations of chamber use in flowing water caused by chamber design and field conditions: (1) the chamber may over-predict the flux from streams because the chamber disturbs the water surface and induces water turbulence and (2) the chamber is difficult to deploy in areas where there is significant turbulence, the stream is narrow, or the stream is shallow. Failing to be able to use the chamber in the most turbulent areas and in narrow and/or shallow streams causes an underestimation of gas fluxes from streams.

Users of the chamber in oceanographic and lake environments were the first to question and test the floating chamber for accuracy of flux measurements (Matthews et al. 2003; Vachon et al. 2010; Martinsen et al. 2018). It has been visually observed (Campeau et al. 2014) that chambers likely add bias to flux estimates in streams by disturbing the water surface and increasing turbulence, but to date, few studies have attempted to quantify this effect or incorporate it into chamber design (Lorke et al. 2015; Ribas-Ribas et al. 2018). The main objectives of this study were to test the validity of chamber-based flux measurements in moving water in a laboratory setting, thereby limiting the tested variable to water velocity and its induced turbulence. This removes the multitude of other factors affecting degassing (e.g., Hall & Ulseth 2020) by maintaining identical ‘stream channel’ characteristics and similar atmospheric conditions, as well as removing the possibility of spatial and temporal variability in CO2 additions to stream water, as is common in natural settings (e.g., Deirmendjian & Abril 2018; Horgby et al. 2019; Luo et al. 2019; Riml et al. 2019). We also evaluated the effects of chamber design by testing two different chambers and independently verified degassing rates in a stirred-beaker experiment by monitoring pH, which is affected by dissolved CO2 concentrations.

METHODS

Flume experiments

Laboratory experiments were conducted in the University of Kansas Water Resources Laboratory. The flume system there simulates a simplified, headwater stream with relatively no variations in roughness or topography compared to a natural stream channel because of its smooth sides and bottom. The flume is a straight, shallow channel, 0.76 m (width) × 19.2 m (length) × 0.89 m (depth) in size (Figure 1). In our experiments, the height of the water in the flume channel was held constant using a 15.25 cm weir on the downstream end of the flume channel (Figure 1(b)); at the CO2-measurement location, the height was 17 cm. Water velocities between 0.13 and 0.23 m s−1 mimicked those observed in a low-gradient, low-ordered headwater stream at the Konza Prairie Biological Station in northeastern Kansas; at these velocities, water discharge was 17–30 L s−1.

Figure 1

The experimental flume. (a) Schematic of the flume plumbing. The system was sealed off to the atmosphere except for the flume channel. CO2 was dissolved in both the holding tank and the constant-head tank. Water recirculates through the system as indicated by the arrows; valves were operated manually. (b) Map view of the flume channel, showing measuring points. (c) Image of the flume, looking upstream at the weir, from the approximate location where chamber measurements were made.

Figure 1

The experimental flume. (a) Schematic of the flume plumbing. The system was sealed off to the atmosphere except for the flume channel. CO2 was dissolved in both the holding tank and the constant-head tank. Water recirculates through the system as indicated by the arrows; valves were operated manually. (b) Map view of the flume channel, showing measuring points. (c) Image of the flume, looking upstream at the weir, from the approximate location where chamber measurements were made.

The reservoir system circulated water between a constant-head tank and a holding tank (Figure 1(a)); CO2 was added to a covered, constant-head tank using diffusion stones. CO2 dissolved into the water for approximately 1.5 h until concentrations reached levels similar to those measured in naturally occurring shallow groundwater (log10pCO2 ≈ −1.5; e.g., Macpherson 2009).

Water velocity and the height of water in the flume were measured for each experimental trial. The velocity of the water was measured by observing the movement of a dye tracer in the flume channel (Figure 1(b)). The dye was introduced through two tubes, one about 5 cm and one about 12 cm from the side of the flume, at a water depth of about 5 cm. These locations were intended to better reflect the near-surface water velocity that would most affect CO2 degassing.

The atmospheric concentrations of CO2 in the room were maintained using two ventilation fans in the ceiling of the laboratory after it was observed that the room concentrations increased to very high levels on one of the days when multiple experiments were run. For the calculations, the y-intercept of the CO2 concentration versus time plots was used as the CO2 concentration in the atmosphere. An AZ-77535 CO2/temperature/relative humidity meter also measured the indoor, atmospheric conditions of the room periodically throughout the duration of the experiments.

Floating chamber measurements

Two floating chamber designs were tested: a large version, modified from Liu (2014) (large floating chamber (LFC); 5.73 L volume) and a smaller, streamlined version designed for this work (streamlined floating chamber (SFC) 1.03 L volume; Table 1). The SFC was designed to be more hydrodynamic than the LFC to improve performance in narrow, shallow streams. It was manufactured using 3D-printed acrylonitrile butadiene styrene (ABS) plastic and was coated with paraffin wax to reduce gas permeability. The float-support structure of the SFC differs from that of the LFC, in that it is separated from the chamber (Table 1). The gas-integrity of the SFC was verified before deployment (Rawitch 2016). The LFC was used in 26 experimental trials, and the SFC was used in 19 experimental trials, over a range of water velocities from 0.13 to 0.23 m s−1. Both chamber designs were tethered in place at the mid-section of the flume.

Table 1

Floating chamber characteristics

LFCSFC
  
Contact surface area: 0.035 m2 Contact surface area: 0.0094 m2 
Volume: 5.73 L Volume: 1.03 L 
Material: Inverted polypropylene bucket sealed to an inverted funnel. Tubing: Tygon® tubing, 1/8 inch inside diameter, ¼ inch outside diameter, 1/16 inch wall Material: 3D-printed ABS P-240 plastic coated in paraffin wax to reduce gas permeability. Tubing: Tygon® tubing, 1/8 inch inside diameter ¼ inch outside diameter, 1/16 inch wall 
Other characteristics: Styrofoam float immediately adjacent to where the device contacted the water surface. Other characteristics: Styrofoam floats were extended away from the device where it contacted the water surface. 
LFCSFC
  
Contact surface area: 0.035 m2 Contact surface area: 0.0094 m2 
Volume: 5.73 L Volume: 1.03 L 
Material: Inverted polypropylene bucket sealed to an inverted funnel. Tubing: Tygon® tubing, 1/8 inch inside diameter, ¼ inch outside diameter, 1/16 inch wall Material: 3D-printed ABS P-240 plastic coated in paraffin wax to reduce gas permeability. Tubing: Tygon® tubing, 1/8 inch inside diameter ¼ inch outside diameter, 1/16 inch wall 
Other characteristics: Styrofoam float immediately adjacent to where the device contacted the water surface. Other characteristics: Styrofoam floats were extended away from the device where it contacted the water surface. 
Using Equation (1), a chamber-based CO2 flux (FChamber) was calculated for each experimental trial. The chamber was sealed at the water surface, and continuous measurements of CO2 were made for approximately 5 min immediately after deployment. Chamber volume, V (L), and the area of the water surface sealed by the chamber, Achamber (m2), differed between the two-chamber designs (Table 1). The change in mole fraction (μmol per mol) of CO2 inside each chamber with time was measured at 1 Hz by circulating chamber air through an LI-820 CO2 infrared gas analyzer (IRGA). The sampled air was drawn from the top of both chamber designs, filtered through a Drierite® granule filter to remove moisture, analyzed for CO2 content, and then recirculated back to the chamber to avoid causing pressure changes inside the chamber. Between sampling events, the chamber CO2 concentration was allowed to equilibrate with the room atmosphere to avoid CO2 accumulation inside the chamber. The air temperature in degrees Celsius (converted to K for Equation (1)), measured using a hand-held meter as described in the Methods, Flume Experiments section, and R, the ideal gas constant (L atm mol−1 K−1) are also required to solve for FChamber (Equation (1)). Atmospheric pressure (p) during the trials was 1 atm ±1%. To determine , a linear regression was fit to the linear portion of the measured change in CO2 mole fraction per second, accepting those with R2 > 0.90.
formula
(1)
The FChamber obtained in Equation (1) can be applied to the flux equation (Equation (2)) to solve for k, the gas transfer coefficient (Cole & Caraco 1998). The gas transfer coefficient is considered representative of the depth of water that equalizes with atmospheric gas concentrations per unit of time.
formula
(2)
The gas transfer coefficient is temperature dependent and is often corrected to a temperature of 20 °C where the temperature-corrected value of k is expressed as k600 (Equation (3)). In Equation (3), is the Schmidt number of CO2 at the temperature of interest, and 600 is the Schmidt number of CO2 at 20 °C. The ratio is raised to the power of −½, which is appropriate for use in rough surface waters (Jähne et al. 1987).
formula
(3)

Stirring plate simulations

The residence time of water in the flume was too short for measurable pH change (>0.01 standard pH units), which change theoretically can be used as an independent measure of CO2 degassing rates. For this reason, a second set of laboratory experiments was conducted to verify the results of the flume experiments. During this set of experiments, two 800 mL-aliquots of flume water, placed in glass beakers with a depth of 15.25 cm to match the approximate depth of water within the flume channel during the flume experimental trials, were charged with CO2 by bubbling ∼100% CO2 through diffusion stones. The first glass beaker was placed on a magnetic stirring plate (stirred sample), while the second sample remained adjacent to it in the same room (still sample). The stirred sample was mixed using a magnetic stir bar rotating at a rate of approximately 90 revolutions per minute (rpm) to simulate the water velocity observed in flume experiments; 90 rpm provided a radial velocity of about 0.1 m s−1, similar to the lowest linear water velocity of the flume system (0.13 m s−1). Water velocities were confirmed by measuring the travel time of a dye around the circumference of the beaker. Two experiments of both the stirred and still samples were run, one at higher CO2 (log10pCO2-water ≈ −0.1) and one at lower CO2 (log10pCO2-water ≈ −2.2). In both cases, the partial pressures were much higher than in the laboratory atmosphere to ensure that degassing occurred. For both the stirred and still samples, pH and temperature were recorded periodically throughout each experiment using an Orion Star A329 meter with an Orion 81007UWMMD Ross Ultra pH/ATC triode-type electrode (precision 0.01 standard units).

The flux from each of the beakers (FBeaker) was calculated using Equation (4) where ΔCO2 is the change in dissolved CO2 over the duration of the experiment, calculated from the change in pH; Abeaker is the surface area of the beaker (m2); and t is the duration of the experiment(s). A value for k600 was then calculated using Equation (3).
formula
(4)

RESULTS

Flume experiments

Water chemistry

Alkalinity varied only slightly throughout the period during which experiments were run, with a mean of 136.5 ± 3.9 ppm as (1 standard deviation (SD); n = 32), and coefficient of variation equal to 3% (Supplementary Table S1). The difference between the cations and other anions in the two water samples was very small (Supplementary Table S2).

The water in the flume was supersaturated with CO2 compared to the atmosphere of the laboratory on all experiment days but one: log10pCO2-water ranged from −1.46 to −2.37. Laboratory atmospheric CO2 concentrations, controlled by the ventilation system in the laboratory, fluctuated between 400 and 5,000 parts per million by volume (ppmv) CO2 (median and mean concentration of 1,610 and 2,160 ± 1,660 ppmv CO2) over the study period. Laboratory atmospheric CO2 was high because of leakage from the basement reservoir headspace to the laboratory room containing the flume when the room-air ventilation was inactive, which occurred only during 1 day of trials.

The pH of flume water ranged from 6.58 to 7.50 with a mean pH of 6.81. The PHREEQC geochemical speciation model (Parkhurst & Appelo 1999) results showed that waters were undersaturated with respect to all minerals in the PHREEQC.dat database, including the carbonates (log saturation indices of around −1.1, −1.3, and −2.8 for calcite, aragonite, and dolomite, respectively). Because of this, we attribute changes in pH solely to degassing and not to carbonate-mineral precipitation. With negligible mineral precipitation, any change in the H+ concentration of the water was due to changes in the CO2 or concentrations (Equations (5) and (6)). Because alkalinity, dominated by , remained consistent throughout the experiment, changes in CO2 were the primary driver of changes in pH (Equations (6) and (7)).
formula
(5)
formula
(6)
formula
(7)

Results suggest that the chemistry over the length of the flume would likely only be influenced by the movement of gas species between the water and the atmosphere. The constant oversaturation of aqueous CO2, with respect to the laboratory atmosphere, suggests that the flume water was most likely to degas CO2 rather than absorb CO2, except for one set of experiments, described later.

CO2 fluxes from floating chambers

Physical and environmental conditions during the flume experiments were held relatively constant (Supplementary Table S3). Of the 45 flume experiments, equipment malfunction or breaking of the seal between the chamber and water occurred in 5 experiments, so 39 experiments comprise the data set for this study (Supplementary Table S3).

More than 90% of all chamber CO2 measurements varied linearly with respect to time (R2 ≥ 0.90; Supplementary Table S3); we used this criterion to select data for further investigation. SFC measurements converted to k600 ranged between 2.48 × 10−8 and 5.67 × 10−8m s−1 with an average value of 4.86 × 10−8 ± 7.61 × 10−9 m s−1 (1 SD; Figure 2). The LFC measurements converted to k600 ranged between 5.69 × 10−10 and 9.17 × 10−8 with a mean value of 2.70 × 10−8 ± 2.70 × 10−8 m s−1 (1 SD; Figure 2). The values of k600 measured during all flume experimental trials with both chamber designs ranged between −5.69 × 10−10 and 9.17 × 10−8m s−1 with a mean value of 3.50 × 10−8 ± 2.45 × 10−8 m s−1 (1 SD; Figure 2). The trend of the k600 data with flume water velocity (Figure 2), using only FC data, is significant with R2 of 0.37 and p-value (at 0.05 confidence level) of 6.53 × 10−5. Trend, confidence intervals, prediction intervals, and statistics are based on the entire data set of FC measurements (n = 37) and are shown on both Figure 2(a) and 2(b).

Figure 2

Floating chamber and beaker experiment gas transfer coefficients versus water velocity. (a) Individual experiments. (b) Data averaged by four levels of water velocity. The points plotted are averages over velocity ranges with coefficient of variation used to represent uncertainties. Numbers indicate the number of experiments at each water-velocity range. The highest water velocity incorporates all points greater than or equal to 0.22 m s−1, the second highest water velocity averages all points between 0.19 and 0.21 m s−1, second lowest velocity averages points between 0.015 and 0.018 m s−1, and the lowest water velocity averages all points with velocity of 0.13 m s−1. LFC is large floating chamber; SFC is small, streamlined floating chamber.

Figure 2

Floating chamber and beaker experiment gas transfer coefficients versus water velocity. (a) Individual experiments. (b) Data averaged by four levels of water velocity. The points plotted are averages over velocity ranges with coefficient of variation used to represent uncertainties. Numbers indicate the number of experiments at each water-velocity range. The highest water velocity incorporates all points greater than or equal to 0.22 m s−1, the second highest water velocity averages all points between 0.19 and 0.21 m s−1, second lowest velocity averages points between 0.015 and 0.018 m s−1, and the lowest water velocity averages all points with velocity of 0.13 m s−1. LFC is large floating chamber; SFC is small, streamlined floating chamber.

Beaker experiments

The residence time of water in the flume, from upstream to downstream sampling points, was so small that pH changes were mostly smaller than the precision of the pH electrode, making it impossible to estimate CO2 degassing rates in the flume as an independent check on the floating chamber degassing rates. For this reason, beaker experiments were designed to simulate conditions in the floating chamber during the flume experiments while allowing pH measurement over sufficient time periods. During the stirred-beaker experiment, pH increased from 6.41 to 6.93 over a period of 2.77 h during the relatively low-CO2 experiment. In the companion still-water experiment, pH changed from 6.42 to 6.53 over a period of 2.74 h. Both experiments yielded linear changes in pH, with R2 of 0.999 for the stirred experiment and 0.980 for the still experiment (Rawitch 2016), and the slope of the stirred best-linear fit was approximately five times larger than that of the still experiment. In the relatively high CO2 experiment, pH increased from 5.17 to 5.36 over 31 min in the stirred beaker (R2 = 0.980) and 5.16 to 5.22 over 24 min (R2 = 0.75) in the still beaker, and the slope of the stirred best-linear fit was ∼15 times larger than that of the still experiment.

Fluxes of CO2 from the water in the beakers to the laboratory atmosphere were modeled using PHREEQC (Parkhurst & Appelo 1999) to calculate pCO2 in equilibrium with the water. For the relatively low-CO2 experiment, the log10pCO2 decreased from −1.32 to −1.83 in the stirred-beaker experiment and from −1.33 to −1.44 in the still-beaker experiment. The CO2 fluxes were used with the measured CO2 concentration gradient between the air and the water to calculate a value for k (Equation (6)) and then temperature corrected to a k600. The values of k600 measured during the beaker experiments were 2.41 × 10−9m s−1 in the stirred-beaker experiment and 3.03 × 10−9 m s−1 in the still-beaker experiment. For the higher-CO2 experiment, the log10pCO2 decreased from −0.07 to −0.16 in the stirred-beaker experiment and from −0.06 to −0.12 in the still-beaker experiment. The k600 values were 1.26 × 10−8m s−1 in the stirred-beaker experiment and 1.08 × 10−8 m s−1 in the still-beaker experiment. The beaker experiments plot on the trend established by the flume experiments (Figure 2(a)).

DISCUSSION

Inability to quantify CO2 efflux from streams severely limits robust quantification of the global carbon cycle. Floating chambers provide a relatively fast and effective method of providing this estimate, and with confirmation of their validity, they can be used to improve our understanding of carbon cycling and the importance of stream efflux. Establishing the validity of floating chambers requires a controlled environment, and in this work, the flume experiments were designed to simulate a simplified headwater stream environment, physically and chemically. The range of water velocities during the experiments (0.13–0.23 m s−1) falls within the range observed in many low-gradient, low-ordered headwater streams, and the water temperature remained within a range acceptable for many stream environments. All flume experiments in this study were done with the chamber fixed in one place (static, i.e., anchored rather than moving). In general, the expectation is a linear dependency of k on the water velocity for chamber measurements, similar to those reported as unpublished data by Teodoru et al. (2015) for degassing in the Congo River and to data from other locations (Table 2).

Table 2

Selected CO2 gas efflux measurements using floating chambers in streams, rivers, and wetlands

ReferenceWater velocity (V, m s−1) or discharge (D, L s−1)aEnvironmentGases measuredFluxes measuredChamber design details
Dinsmore et al. (2010)  D: 0.5–632 Peatland streams CO2, CH4, and N21,200 μg C m−2s−1
2 μg C m−2s−1
0.025 μg N m−2s−1 
Opaque injection-molded polypropylene box with dimensions of 61 cm × 30 cm × 15 cm with a volume of 0.0191 m3 
Alin et al. (2011)  V: 0.054–1; D: 100–38,000 Low-gradient rivers CO2 0–25 μmol m−2 s−1 Equipped with an internal fan to simulate wind 
Belger et al. (2011)  Not provided by author (n/a) Amazonian interfluvial wetlands CO2 and CH4 2,193 and 60 mg C m−2 d−1 Vented, with a fan to circulate air inside, 25 cm diameter with a volume of 10 L, sample time was 15 min, and gas samples were taken at 5-min intervals with a syringe 
Neu et al. (2011)  D: 13–52 Amazonian first-order headwater stream CO2 and CH4 5,994 ± 677 and 987 ± 221 g C m−2 y−1 Dynamic Plexiglas floating chamber (0.125 m2 footprint with 13.5 L volume) connected to an IRGA 
Sha et al. (2011)  n/a Riverine wetlands CH4 0–160 mg CH4 m−2 h−1 Rectangular 0.21 m2 footprint and 56 L plastic storage containers equipped with foam for buoyancy and wrapped with duct tape for waterproofing 
Beaulieu et al. (2012)  V: 0.07–1.19; D: 380,000–8,076,000 Large rivers CO2, CH4, and N2Reported gas transfer coefficients (k) and not fluxes 20 L acrylic chambers with a 0.164 m2 footprint and a height of 17.8 cm, equipped with a 1 cm diameter vent hole, supported within an air-filled flotation collar, chamber walls extended 1–2 cm into the water 
Sand-Jensen & Staehr (2012)  V: 0.01–0.82 Danish lowland streams CO2 170–1,200 mmol m−2 d−1 Constructed as a half-cylinder 94 cm long and 18 cm in internal diameter made of PVC painted white to prevent heating 
Sawakuchi et al. (2012)  n/a Large tropical rivers in the amazon basin CH4 59–2,974 mmol m−2 y−1 7.5 L volume chamber with a 30 cm diameter and covered with reflective aluminum tape to reduce heating 
Striegl et al. (2012)  n/a Yukon rivers and streams CO2 and CH4 6–548 mmol C m2 d−1
0.07–4.80 mol C m2 d−1 
Allowed to float alongside a boat in rivers and anchored in place in streams 
Bastien & Demarty (2013)  n/a Reservoirs, rivers CO2 and CH4 0–5,000 mg CO2 m−2 d−1 and 0–15 mg CH4 m−2 d−1 17.6 L volume and 0.16 m2 footprint, sampled for 7 min continuously with air recirculating to an analyzer 
Crawford et al. (2013)  D: 10–1,320 Oxygenated wetland streams CO2 and CH4 1–20 and 0–15 μmol m−2 s−1 Suspended clear chamber 
Denfeld et al. (2013)  D: ∼2–5 million Arctic river basin including streams and rivers CO2 0–26 g C m−2d−1 Circular plastic chamber with a fixed headspace, and allowed to float freely alongside a boat. Recirculated air sample through an IRGA 
Han et al. (2013)  n/a Rivers CO2, CH4, and N21,023 mg m−2 h−1, 89 mg m−2 h−1, and 151 μg m−2 h−1 The static chamber with gas samples analyzed using chromatography 
Huotari et al. (2013)  V: 0.15–1.17; D: 234,000 Boreal river CO2 83 ± 37 μmol m−2 d−1 Not described in detail 
Juutinen et al. (2013)  D: 1.9–16.1 Northern peatland-stream-lake continuum CO2 and CH4 479 and 0.4–17.4 g C m2 a−1 Chamber walls were submerged 5 cm below the water surface. 6.6 L volume and 451 cm2 area footprint. 6.6 L, 452 cm2 area, with sidewalls penetrating 5 cm into the water column. Allowed to drift next to a boat, and sampled for 20–60 min 
Krenz (2013)  n/a Small boreal lake and its connecting streams CO2 29–1,616 g C m−2 y−1 7.5 L volume chamber with a 30 cm diameter and covered with reflective aluminum tape to reduce heating 
Rasera et al. (2013)  n/a Amazonian rivers CO2 0.8–15.3 μmol m−2 s−1 0.125 m2 footprint, 10.6 L volume, made of Plexiglas with a 1 mm diameter hole to provide for pressure equalization, recirculated air sample through an IRGA, and sample passed through a water trap, chamber walls extended 2–3 cm into the water 
Xia et al. (2013)  V: 0.01–0.3 Sewage-enriched river N262 μg N m−2 h−1 Constructed of round polypropylene ‘cake’ containers 0.0064 m3 and 0.07 m2 footprint were floated with Styrofoam annuli and penetrated 1–2 cm into the water column. Covered with silver paper to prevent heating 
Campeau & Del Giorgio (2014)  V: 0.001–1 Boreal rivers and streams CO2 and CH4 aReported as k600 Covered with aluminum foil to reduce solar heating and equipped with an internal thermometer to monitor temperature 
Campeau et al. (2014)  V: 0.001–1 Boreal rivers and streams CO2 and CH4 888 and 97.8 mg C m−2 d−1 Measurements of chamber concentration taken every minute for 10 min 
Khadka et al. (2014)  D: 30,000–300,000 River system CO2 400–1,200 g C m−2 y−1 28 L volume plastic storage bin with an 0.11 m2 footprint, and allowed to penetrate 6 cm into the water column, recirculated air sample through an IRGA, left on the water for 7–10 min 
Bednařík et al. (2015)  D: 312, 81 Third-order stream CH4 0.07–0.73 mmol m−2 d−1 3.1 L volume and 0.024 m2 footprint equipped with styrene floats, sampled every 30 min with a syringe for 3 h 
Billett et al. (2015)  D: 1–374 Peatland streams 14C and δC13 isotope N/A In-line molecular sieve for CO2 isotopes 
Borges et al. (2015)  D: 60,000–41 million Rivers CO2 and CH4 186–1,149 mmol m−2d−1
0.5–18 mmol m−2d−1 
IRGA for CO2 and in-line gas extraction for CH4 
Chen et al. (2015)  n/a River with hydroelectric reservoirs N210–30 μg m−2 h−1 Thermometer located at the top of the chamber to measure inside air temperature 
Crawford et al. (2015)  D: 0–600 High elevation mountain streams CO2 and CH4 −40 to 146 mmol CO2 m−2d−1 and −21.2–642 μmol CO2 m−2 d−1 Suspended clear chamber 
Deshmukh et al. (2015)  D: 2000 Downstream of a subtropical hydroelectric reservoir CH4 1.14–3.3 mmol m−2 d−1 Allowed to float alongside a boat, and connected to an IRGA 
Gómez-Gener et al. (2015)  V: 0.09–0.64; D: 120–2,720 Intermittent Mediterranean fluvial network CO2 and CH4 120 ± 33 and 13.9 ± 10.1 mmol m−2 d−1 Monitored the gas concentrations in the chamber every 30 s for a total of 10 min after passing through an in-line moisture trap at a rate of 2.9 L min−1 
Jones et al. (2015)  V: 0.01–0.3 Streams and pools in karstic cave systems H20–80 μmol m−2 s−1 The chamber was connected to a hand-held gas detector 
Luan & Wu (2015)  n/a Drainage ditches CH4 0–16 CH4 mg m−2 d−1 50 cm in height, 26.3 cm in diameter, and equipped with a capillary tube to maintain atmospheric pressure 
Müller et al. (2015)  V: 0.7 ± 0.7, 0.8 ± 1.0, 2.5 ± 1.4; D: 160,000; 490,000 Tropical peat-draining river CO2 1.8–28.5 g C m−2 d−1 Equipped with a vent tube to maintain atmospheric pressure, and edges extended 1 cm into the water 
Ran et al. (2015)  n/a The Yellow River CO2 7.9 ± 1.2 Tg C y−1 Volume of 0.042 m3 and a footprint of 0.2 m2, walls 3–5 cm into the water column for 40–70 min 
Teodoru et al. (2015)  D: 10,000–3,500,000 River and major tributaries CO2 and CH4 3,380 and 48.6 mg C m−2d−2 Temperature continuously monitored, 17 L volume, and extended 7 cm below water surface 
Tran et al. (2017)  n/a City canals CO2 35–446 mmol m−2 h−1 n/a 
This study V: 0.13–0.23; D: 17–30 Laboratory flume simulating headwater stream CO2 −0.7 to 36.7 μmol s−1 m−2 See text 
ReferenceWater velocity (V, m s−1) or discharge (D, L s−1)aEnvironmentGases measuredFluxes measuredChamber design details
Dinsmore et al. (2010)  D: 0.5–632 Peatland streams CO2, CH4, and N21,200 μg C m−2s−1
2 μg C m−2s−1
0.025 μg N m−2s−1 
Opaque injection-molded polypropylene box with dimensions of 61 cm × 30 cm × 15 cm with a volume of 0.0191 m3 
Alin et al. (2011)  V: 0.054–1; D: 100–38,000 Low-gradient rivers CO2 0–25 μmol m−2 s−1 Equipped with an internal fan to simulate wind 
Belger et al. (2011)  Not provided by author (n/a) Amazonian interfluvial wetlands CO2 and CH4 2,193 and 60 mg C m−2 d−1 Vented, with a fan to circulate air inside, 25 cm diameter with a volume of 10 L, sample time was 15 min, and gas samples were taken at 5-min intervals with a syringe 
Neu et al. (2011)  D: 13–52 Amazonian first-order headwater stream CO2 and CH4 5,994 ± 677 and 987 ± 221 g C m−2 y−1 Dynamic Plexiglas floating chamber (0.125 m2 footprint with 13.5 L volume) connected to an IRGA 
Sha et al. (2011)  n/a Riverine wetlands CH4 0–160 mg CH4 m−2 h−1 Rectangular 0.21 m2 footprint and 56 L plastic storage containers equipped with foam for buoyancy and wrapped with duct tape for waterproofing 
Beaulieu et al. (2012)  V: 0.07–1.19; D: 380,000–8,076,000 Large rivers CO2, CH4, and N2Reported gas transfer coefficients (k) and not fluxes 20 L acrylic chambers with a 0.164 m2 footprint and a height of 17.8 cm, equipped with a 1 cm diameter vent hole, supported within an air-filled flotation collar, chamber walls extended 1–2 cm into the water 
Sand-Jensen & Staehr (2012)  V: 0.01–0.82 Danish lowland streams CO2 170–1,200 mmol m−2 d−1 Constructed as a half-cylinder 94 cm long and 18 cm in internal diameter made of PVC painted white to prevent heating 
Sawakuchi et al. (2012)  n/a Large tropical rivers in the amazon basin CH4 59–2,974 mmol m−2 y−1 7.5 L volume chamber with a 30 cm diameter and covered with reflective aluminum tape to reduce heating 
Striegl et al. (2012)  n/a Yukon rivers and streams CO2 and CH4 6–548 mmol C m2 d−1
0.07–4.80 mol C m2 d−1 
Allowed to float alongside a boat in rivers and anchored in place in streams 
Bastien & Demarty (2013)  n/a Reservoirs, rivers CO2 and CH4 0–5,000 mg CO2 m−2 d−1 and 0–15 mg CH4 m−2 d−1 17.6 L volume and 0.16 m2 footprint, sampled for 7 min continuously with air recirculating to an analyzer 
Crawford et al. (2013)  D: 10–1,320 Oxygenated wetland streams CO2 and CH4 1–20 and 0–15 μmol m−2 s−1 Suspended clear chamber 
Denfeld et al. (2013)  D: ∼2–5 million Arctic river basin including streams and rivers CO2 0–26 g C m−2d−1 Circular plastic chamber with a fixed headspace, and allowed to float freely alongside a boat. Recirculated air sample through an IRGA 
Han et al. (2013)  n/a Rivers CO2, CH4, and N21,023 mg m−2 h−1, 89 mg m−2 h−1, and 151 μg m−2 h−1 The static chamber with gas samples analyzed using chromatography 
Huotari et al. (2013)  V: 0.15–1.17; D: 234,000 Boreal river CO2 83 ± 37 μmol m−2 d−1 Not described in detail 
Juutinen et al. (2013)  D: 1.9–16.1 Northern peatland-stream-lake continuum CO2 and CH4 479 and 0.4–17.4 g C m2 a−1 Chamber walls were submerged 5 cm below the water surface. 6.6 L volume and 451 cm2 area footprint. 6.6 L, 452 cm2 area, with sidewalls penetrating 5 cm into the water column. Allowed to drift next to a boat, and sampled for 20–60 min 
Krenz (2013)  n/a Small boreal lake and its connecting streams CO2 29–1,616 g C m−2 y−1 7.5 L volume chamber with a 30 cm diameter and covered with reflective aluminum tape to reduce heating 
Rasera et al. (2013)  n/a Amazonian rivers CO2 0.8–15.3 μmol m−2 s−1 0.125 m2 footprint, 10.6 L volume, made of Plexiglas with a 1 mm diameter hole to provide for pressure equalization, recirculated air sample through an IRGA, and sample passed through a water trap, chamber walls extended 2–3 cm into the water 
Xia et al. (2013)  V: 0.01–0.3 Sewage-enriched river N262 μg N m−2 h−1 Constructed of round polypropylene ‘cake’ containers 0.0064 m3 and 0.07 m2 footprint were floated with Styrofoam annuli and penetrated 1–2 cm into the water column. Covered with silver paper to prevent heating 
Campeau & Del Giorgio (2014)  V: 0.001–1 Boreal rivers and streams CO2 and CH4 aReported as k600 Covered with aluminum foil to reduce solar heating and equipped with an internal thermometer to monitor temperature 
Campeau et al. (2014)  V: 0.001–1 Boreal rivers and streams CO2 and CH4 888 and 97.8 mg C m−2 d−1 Measurements of chamber concentration taken every minute for 10 min 
Khadka et al. (2014)  D: 30,000–300,000 River system CO2 400–1,200 g C m−2 y−1 28 L volume plastic storage bin with an 0.11 m2 footprint, and allowed to penetrate 6 cm into the water column, recirculated air sample through an IRGA, left on the water for 7–10 min 
Bednařík et al. (2015)  D: 312, 81 Third-order stream CH4 0.07–0.73 mmol m−2 d−1 3.1 L volume and 0.024 m2 footprint equipped with styrene floats, sampled every 30 min with a syringe for 3 h 
Billett et al. (2015)  D: 1–374 Peatland streams 14C and δC13 isotope N/A In-line molecular sieve for CO2 isotopes 
Borges et al. (2015)  D: 60,000–41 million Rivers CO2 and CH4 186–1,149 mmol m−2d−1
0.5–18 mmol m−2d−1 
IRGA for CO2 and in-line gas extraction for CH4 
Chen et al. (2015)  n/a River with hydroelectric reservoirs N210–30 μg m−2 h−1 Thermometer located at the top of the chamber to measure inside air temperature 
Crawford et al. (2015)  D: 0–600 High elevation mountain streams CO2 and CH4 −40 to 146 mmol CO2 m−2d−1 and −21.2–642 μmol CO2 m−2 d−1 Suspended clear chamber 
Deshmukh et al. (2015)  D: 2000 Downstream of a subtropical hydroelectric reservoir CH4 1.14–3.3 mmol m−2 d−1 Allowed to float alongside a boat, and connected to an IRGA 
Gómez-Gener et al. (2015)  V: 0.09–0.64; D: 120–2,720 Intermittent Mediterranean fluvial network CO2 and CH4 120 ± 33 and 13.9 ± 10.1 mmol m−2 d−1 Monitored the gas concentrations in the chamber every 30 s for a total of 10 min after passing through an in-line moisture trap at a rate of 2.9 L min−1 
Jones et al. (2015)  V: 0.01–0.3 Streams and pools in karstic cave systems H20–80 μmol m−2 s−1 The chamber was connected to a hand-held gas detector 
Luan & Wu (2015)  n/a Drainage ditches CH4 0–16 CH4 mg m−2 d−1 50 cm in height, 26.3 cm in diameter, and equipped with a capillary tube to maintain atmospheric pressure 
Müller et al. (2015)  V: 0.7 ± 0.7, 0.8 ± 1.0, 2.5 ± 1.4; D: 160,000; 490,000 Tropical peat-draining river CO2 1.8–28.5 g C m−2 d−1 Equipped with a vent tube to maintain atmospheric pressure, and edges extended 1 cm into the water 
Ran et al. (2015)  n/a The Yellow River CO2 7.9 ± 1.2 Tg C y−1 Volume of 0.042 m3 and a footprint of 0.2 m2, walls 3–5 cm into the water column for 40–70 min 
Teodoru et al. (2015)  D: 10,000–3,500,000 River and major tributaries CO2 and CH4 3,380 and 48.6 mg C m−2d−2 Temperature continuously monitored, 17 L volume, and extended 7 cm below water surface 
Tran et al. (2017)  n/a City canals CO2 35–446 mmol m−2 h−1 n/a 
This study V: 0.13–0.23; D: 17–30 Laboratory flume simulating headwater stream CO2 −0.7 to 36.7 μmol s−1 m−2 See text 

aReported units of discharge and velocity were converted to L s−1 and m s−1, respectively, for easier comparison.

We found that the k600 values for both the large and small chamber designs were of a similar magnitude (Figure 2(a)), so we propose that chamber design did not have a significant effect on measured gas transfer in this study. Since the two-chamber designs resulted in comparable measurements, they are grouped together. We note that the validity of the small chamber design is promising for quantifying efflux from headwater streams as they are usually narrow. A linear trend of k versus water velocity was observed for all of the data (Figure 2(a)), with significant R2 of 0.37 (p-value of 6.53 × 10−5), suggesting that the turbulence created by water velocity, even in the smooth-walled and smooth-floored flume, is likely an important control on degassing. Given some uncertainty likely present in our velocity measurements, values for k600 were grouped into four water-velocity ranges (Figure 2(b)). The results show increasing variability in k600 with increasing water velocity, which could indicate the increasing difficulty of making reproducible measurements at higher water velocity. The positive relation between water velocity and k600 we found in the flume supports the natural log-linear relation between k600 and discharge (10–266 L s−1) found in a steep headwater stream (McDowell & Johnson 2018), although Ulseth et al. (2019) suggest different linear relations between k600 and the energy dissipation rates in low- and high-slope streams in a compilation of data, and less difference in the low- and high-slope streams’ relation between stream velocity (all data, range of 0.006–1.2 m s−1) and k600 (all data, 0.1–4,100).

The nine flume experiments used to calculate the average k at the lowest water velocity (0.13 m s−1; trials 5–1 to 5–9 in Supplementary Table S2) were obtained on the same day and provided values almost two orders of magnitude lower than other experiments. An experiment with a slightly higher flume water velocity (0.15 m s−1; trial 4–6 in Table S2) obtained during a different sampling event provided a value that was comparable to many of the other values for k. Whether this represents some kind of threshold behavior is unclear because the low-velocity experiments with the lowest values for k occurred when laboratory atmospheric concentrations of CO2 were abnormally high (mean of 4,900 ppmv CO2 compared to a mean of 1,200 ppmv for all other trials). The high laboratory-air CO2 concentrations resulted in the lowest values of (CO2-water – CO2-air), which were in fact negative (CO2 moving from the atmosphere into the water). This inadvertent testing of negative fluxes is beneficial for future field applications as negative fluxes have been reported from field measurements (Liu et al. 2015). It should be noted that the reported field measurements had the opposite reason for the negative flux (very low water CO2 rather than very high atmospheric CO2). Using the absolute value of the negative flux experimental results will not change the outcome substantially because of the low numeric values.

To validate the floating chamber results, we used the pH change that results from degassing CO2 as in an independent measure of the k600. The limited residence time of the water in the flume did not allow the water to degas enough CO2 to result in a significant pH change that could be measured within the precision of the pH probe. Similar unresolvable changes in pH were observed in the beaker experiments over the same time periods as the residence time of water in the flume. Over longer time periods, however, greater pH changes were measured in the beaker simulations so that the rate of CO2 degassing was quantifiable. The water velocities in the beaker experiments were similar to but lower than the lowest velocities measured in the flume. The CO2 flux out of the beaker in the stirring experiments was always higher than the flux of the non-stirring experiments (Rawitch 2016 and Supplementary Table S4). The k600 estimates of the stirring plate experiments (2.4 × 10−8 and 1.3 10−8 m s−1, lower CO2 and higher CO2, respectively) were larger than the k600 estimates of the still experiments (3.1 × 10−9 and 1.1 10−8 m s−1, lower CO2 and higher CO2, respectively). The ratio of the k600 estimates in the stirred to still experiments was 7.9 and 1.2 (lower CO2 and higher CO2, respectively) even though the stirring rate was the same in the stirred experiments. This suggests that the high CO2 exsolves out of the water disproportionately, being driven largely by the high concentration gradient. The beaker experiment k600 estimates fall within the extrapolated trend of the floating chamber experiments (Figure 2(a)).

Previous research into the use of the floating chamber in a lake (Matthews et al. 2003) suggests that the floating chamber method may produce artificially high measurements of CO2 caused by the disturbance of the surface boundary layer. In a different study, Vachon et al. (2010) used an acoustic Doppler velocimeter to demonstrate that the floating chamber technique in a lake setting disturbs the surface boundary layer and the chamber design overestimates fluxes due to enhanced turbulence. Although multiple authors have acknowledged the effect of chambers disturbing the water surface, there have been relatively few attempts to adequately quantify this disturbance (Kremer et al. 2003; Matthews et al. 2003; Vachon et al. 2010; Lorke et al. 2015) and only one known study attempting to quantify the effect of chamber-induced turbulence on fluxes in systems with flowing water (Lorke et al. 2015). Lorke et al. (2015) suggested that static chambers used in flowing water consistently measure higher flux values than chambers allowed to drift. Lorke et al. (2015) suggested a mechanism by which anchored chambers increase the turbulence under chambers and elevate the observed flux values due to this artificially induced turbulence. We noted during our flume experiments that both chambers did visibly disturb the surface of the water, and presumably the water under the chamber, although the disturbance was not apparent in our results. A newer chamber design for a buoy system deployed on seawater incorporates acoustic Doppler velocimeters to allow compensation for turbulence (Ribas-Ribas et al. 2018), although the size of this equipment renders in infeasible for headwater stream deployment.

To test whether the floating chamber results could reflect accelerated degassing, we compare the k600 estimates of the flume experiments with the k600 estimates of the beaker experiments. The stirred-beaker experiment k600 estimates were lower than the flume experiments because the water velocity in the beaker was lower than the lowest flume experiment, so we are unable to confirm an acceleration of degassing caused by the floating chamber. The beaker results do fall on the trend established by the flume experiments, however, suggesting chamber-induced degassing may not be large. Therefore, we suggest that the floating chamber provides a reasonable measurement of flux and k600, and that previous estimates using this technique to quantify efflux from streams should be accepted, at least over the range of water velocities tested here.

This study limited the variables affecting k600 to a single one, water velocity, using it as a proxy for turbulence. Other factors also influence turbulence and the k600 value, including geomorphic factors (e.g., Hall & Ulseth 2020), and future work could be designed to test these independently, as well.

CONCLUSIONS

The stirred and still beaker experiments at lower and higher dissolved CO2 levels showed a larger contrast in k600 estimates in the lower CO2 experiment than in the higher-CO2 experiment. This suggests that degassing rates may be affected, in part, by the contrast between the water and air CO2 concentrations in concert with water velocity, where in higher CO2 water the difference in degassing rates between stirred and still was small.

This study supports the idea that the floating chamber can accurately measure CO2 gas fluxes in environments with flowing water. The results of flume experiments that approximated water velocities in headwater streams show a positive linear correlation with water velocity, corroborating the idea that CO2 degassing is proportional to stream turbulence. The flume experimental results from this study were also within the range of other published field measurements, suggesting the floating chamber method may be robust in moving water. Beaker experiments simulating flume water depth and water velocity resulted in k600 estimates that fell within the extrapolated trend of the floating chamber data. This suggests that the floating chamber does not accelerate CO2 degassing, although further work is needed to verify this result.

Nevertheless, careful consideration should always be taken when applying chamber-based methods in environments where water flows against the sides of the chamber and where the structure of the chamber disturbs the interface between the surface water and the atmosphere. When considering methods for quantifying gas flux from streams and rivers, we propose the floating chamber remains a relevant and viable method due to its low cost and simple field application.

ACKNOWLEDGEMENTS

The authors thank the reviewer and editor for their comments that improved this manuscript. We also thank Pamela Sullivan and Walter Dodds for discussion about this project. This study was funded by the University of Kansas (KU) Department of Geology, the Geology Associates Board of the KU Endowment Association, the KU Department of Geography and Atmospheric Science, scholarships to the first author from the Geological Society of America (Engineering Geology Division), the American Water Works Association, Descendants of the Signers of the Declaration of Independence, and the Great Lakes National Scholarship Program. In-kind support was provided by the KU Water Resources Laboratory.

REFERENCES

Aho
K. S.
Raymond
P. A.
2019
Differential response of greenhouse gas evasion to storms in forested and wetland streams
.
Journal of Geophysical Research: Biogeosciences
124
(
3
),
649
662
.
Alin
S. R.
de Fátima
M.
M
F. L.
Salimon
C. I.
Richey
J. E.
Holtgrieve
G. W.
Krusche
A. V.
Snidvongs
A.
2011
Physical controls on carbon dioxide transfer velocity and flux in low-gradient river systems and implications for regional carbon budgets
.
Journal of Geophysical Research: Biogeosciences
116
(
G1
),
G01009
.
doi:10.1029/2010JG001398
.
Aufdenkampe
A. K.
Mayorga
E.
Raymond
P. A.
Melack
J. M.
Doney
S. C.
Alin
S. R.
Aalto
R. E.
Yoo
K.
2011
Riverine coupling of biogeochemical cycles between land, oceans, and atmosphere
.
Frontiers in Ecology and the Environment
9
(
1
),
53
60
.
Battin
T. J.
Luyssaert
S.
Kaplan
L. A.
Aufdenkampe
A. K.
Richter
A.
Tranvik
L. J.
2009
The boundless carbon cycle
.
Nature Geoscience
2
(
9
),
598
600
.
Beaulieu
J. J.
Shuster
W. D.
Rebholz
J. A.
2012
Controls on gas transfer velocities in a large river
.
Journal of Geophysical Research
117
(
G2
),
G02007
.
doi:10.1029/2011JG001794
.
Bednařík
A.
Čáp
L.
Maier
V.
Rulík
M.
2015
Contribution of methane benthic and atmospheric fluxes of an experimental area (Sitka Stream)
.
Clean – Soil, Air, Water
43
,
1
7
.
Belger
L.
Forsberg
B. R.
Melack
J. M.
2011
Carbon dioxide and methane emissions from interfluvial wetlands in the upper Negro River basin, Brazil
.
Biogeochemistry
105
(
1–3
),
171
183
.
Borges
A. V.
Darchambeau
F.
Teodoru
C. R.
Marwick
T. R.
Tamooh
F.
Geeraert
N.
Omengo
F. O.
Guérin
F.
Lambert
T.
Morana
C.
2015
Globally significant greenhouse-gas emissions from African inland waters
.
Nature Geoscience
8
(
8
),
637
.
Butman
D.
Raymond
P. A.
2011
Significant efflux of carbon dioxide from streams and rivers in the United States
.
Nature Geoscience
4
(
12
),
839
842
.
Campeau
A.
Lapierre
J. F.
Vachon
D.
Giorgio
P. A.
2014
Regional contribution of CO2 and CH4 fluxes from the fluvial network in a lowland boreal landscape of Québec
.
Global Biogeochemical Cycles
28
(
1
),
57
69
.
Chen
J.
Cao
W.
Cao
D.
Huang
Z.
Liang
Y.
2015
Nitrogen loading and nitrous oxide emissions from a river with multiple hydroelectric reservoirs
.
Bulletin of Environmental Contamination and Toxicology
94
(
5
),
633
639
.
Cole
J. J.
Caraco
N. F.
1998
Atmospheric exchange of carbon dioxide in a low-wind oligotrophic lake measured by the addition of SF6
.
Limnology and Oceanography
43
(
4
),
647
656
.
Cole
J. J.
Prairie
Y. T.
Caraco
N. F.
McDowell
W. H.
Tranvik
L. J.
Striegl
R. G.
Duarte
C. M.
Kortelainen
P.
Downing
J. A.
Middelburg
J. J.
Melack
J.
2007
Plumbing the global carbon cycle: integrating inland waters into the terrestrial carbon budget
.
Ecosystems
10
(
1
),
172
185
.
Cole
J. J.
Bade
D. L.
Bastviken
D.
Pace
M. L.
Van de Bogert
M.
2010
Multiple approaches to estimating air-water gas exchange in small lakes
.
Limnology and Oceanography: Methods
8
,
285
293
.
Crawford
J. T.
Striegl
R. G.
Wickland
K. P.
Dornblaser
M. M.
Stanley
E. H.
2013
Emissions of carbon dioxide and methane from a headwater stream network of interior Alaska
.
Journal of Geophysical Research: Biogeosciences
118
(
2
),
482
494
.
Crawford
J. T.
Dornblaser
M. M.
Stanley
E. H.
Clow
D. W.
Striegl
R. G.
2015
Source limitation of carbon gas emissions in high-elevation mountain streams and lakes
.
Journal of Geophysical Research: Biogeosciences
118
,
1
14
.
Crawford
J. T.
Stanley
E. H.
Dornblaser
M. M.
Striegl
R. G.
2017
CO2 time series patterns in contrasting headwater streams of North America
.
Aquatic Sciences
79
(
3
),
473
486
.
Denfeld
B. A.
Frey
K. E.
Sobczak
W. V.
Mann
P. J.
Holmes
R. M.
2013
Summer CO2 evasion from streams and rivers in the Kolyma River basin, north-east Siberia
.
Polar Research
32
,
19704
.
Deshmukh
C.
Guérin
F.
Pighini
S.
Vongkhamsao
A.
Guédant
P.
Rode
W.
Godon
A.
Chanudet
V.
Descloux
S.
Serça
D.
2015
Low methane (CH4) emissions downstream of a monomictic subtropical hydroelectric reservoir (Nam Theun 2, Lao PDR)
.
Biogeosciences Discussions
12
,
1113
11347
.
Dinsmore
K. J.
Billett
M. F.
Skiba
U. M.
Rees
R. M.
Drewer
J.
Helfter
C.
2010
Role of the aquatic pathway in the carbon and greenhouse gas budgets of a peatland catchment
.
Global Change Biology
16
(
10
),
2750
2762
.
Genereux
D. P.
Hemond
H. F.
1992
Determination of gas exchange rate constants for a small stream on Walker Branch Watershed, Tennessee
.
Water Resources Research
28
(
9
),
2365
2374
.
Gómez-Gener
L.
Obrador
B.
von Schiller
D.
Marcé
R.
Casas-Ruiz
J. P.
Proia
L.
Acuña
V.
Catalán
N.
Muñoz
I.
Koschorreck
M.
2015
Hot spots for carbon emissions from Mediterranean fluvial networks during summer drought
.
Biogeochemistry
33
(
2
),
349
360
.
Hall
R. O.
Jr.
Ulseth
A. J.
2020
Gas exchange in streams and rivers
.
Wiley Interdisciplinary Reviews: Water
7
(
1
),
e1391
.
Han
Y.
Zheng
Y.-f.
Wu
R.-j.
Yin
J.-f.
Xu
J.-x.
Xu
P.
2013
Greenhouse gases emission characteristics of Nanjing typical waters in Spring
.
China Environmental Science
8
,
004
.
Horgby
Å.
Boix Canadell
M.
Ulseth
A. J.
Vennemann
T. W.
Battin
T. J.
2019
High-resolution spatial sampling identifies groundwater as driver of CO2 dynamics in an Alpine stream network
.
Journal of Geophysical Research: Biogeosciences
124
(
7
),
1961
1976
.
Hotchkiss
E.
Hall
R.
Jr.
Sponseller
R.
Butman
D.
Klaminder
J.
Laudon
H.
Rosvall
M.
Karlsson
J.
2015
Sources of and processes controlling CO2 emissions change with the size of streams and rivers
.
Nature Geoscience
8
(
9
),
696
699
.
Huotari
J.
Haapanala
S.
Pumpanen
J.
Vesala
T.
Ojala
A.
2013
Efficient gas exchange between a boreal river and the atmosphere
.
Geophysical Research Letters
40
(
21
),
5683
5686
.
Jähne
B.
Münnich
K. O.
Bösinger
R.
Dutzi
A.
Huber
W.
Libner
P.
1987
On the parameters influencing air-water gas exchange
.
Journal of Geophysical Research: Oceans
92
(
C2
),
1937
1949
.
Jones
D. S.
Polerecky
L.
Galdenzi
S.
Dempsey
B. A.
Macalady
J. L.
2015
Fate of sulfide in the Frasassi cave system and implications for sulfuric acid speleogenesis
.
Chemical Geology
410
,
21
27
.
Juutinen
S.
Väliranta
M.
Kuutti
V.
Laine
A.
Virtanen
T.
Seppä
H.
Weckström
J.
Tuittila
E. S.
2013
Short-term and long-term carbon dynamics in a northern peatland-stream-lake continuum: a catchment approach
.
Journal of Geophysical Research: Biogeosciences
118
(
1
),
171
183
.
Kremer
J. N.
Nixon
S. W.
Buckley
B.
Roques
P.
2003
Technical note: conditions for using the floating chamber method to estimate air-water gas exchange
.
Estuaries
26
(
4A
),
985
990
.
Krenz
J.
2013
Measuring CO2 Emissions From a Small Boreal Lake and its Connecting Streams Using Automatic Floating Chambers
.
MS Thesis
,
Swedish University of Agricultural Sciences
.
Lambert
M.
Fréchette
J.-L.
2005
Analytical techniques for measuring fluxes of CO2 and CH4 from hydroelectric reservoirs and natural water bodies
.
Greenhouse Gas Emissions - Fluxes and Processes (A. Tremblay, C. Varfalvy, Roehm & M. Garneau, eds). Springer, Netherlands, pp. 37–60
.
Liu
H.
2014
Inorganic and Organic Carbon Variations in Surface Water, Konza Prairie LTER Site, USA and Maolan Karst Experimental Site, China
.
MS Thesis
,
University of Kansas Geology
.
Liu
Z.
Dreybrodt
W.
Wang
H.
2008
A possible important CO2 sink by the global water cycle
.
Chinese Science Bulletin
53
(
3
),
402
407
.
Liu
H.
Liu
Z.
Macpherson
G. L.
Yang
R.
Chen
B.
Sun
H.
2015
Diurnal hydrochemical variations in a karst spring and two ponds, Maolan Karst Experimental Site, China: biological pump effects
.
Journal of Hydrology
522
,
407
417
.
doi:10.1016/j.jhydrol.2015.01.011
.
Lorke
A.
Bodmer
P.
Noss
C.
Alshboul
Z.
Koschorreck
M.
Somlai
C.
Bastviken
D.
Flury
S.
McGinnis
D.
Maeck
A.
2015
Technical note: drifting vs. anchored flux chambers for measuring greenhouse gas emissions from running waters
.
Biogeosciences Discussion
12
,
1461
14645
.
Luo
J.
Li
S.
Ni
M.
Zhang
J.
2019
Large spatiotemporal shifts of CO2 partial pressure and CO2 degassing in a monsoonal headwater stream
.
Journal of Hydrology
579
,
https://doi.org/10/1016/j.jhydrol.2019.124135
.
Martinsen
K. T.
Kragh
T.
Sand-Jensen
K.
2018
Technical note: a simple and cost-efficient automated floating chamber for continuous measurements of carbon dioxide gas flux on lakes
.
Biogeosciences
15
,
5565
5573
.
https://doi.org/10.5194/bg-15-5565-2018
.
Marx
A.
Dusek
J.
Jankovec
J.
Sanda
M.
Vogel
T.
Van Geldern
R.
Hartmann
J.
Barth
J. A. C.
2017
A review of CO2 and associated carbon dynamics in headwater streams: a global perspective
.
Reviews of Geophysics
55
(
2
),
560
585
.
Marx
A.
Conrad
M.
Aizinger
V.
Prechtel
A.
van Geldern
R.
Barth
J. A.
2018
Groundwater data improve modelling of headwater stream CO2 outgassing with a stable DIC isotope approach
.
Biogeosciences
15
(
10
),
3093
3106
.
Matthews
C. J. D.
St. Louis
V. L.
Hesslein
R. H.
2003
Comparison of three techniques used to measure diffusive gas exchange from sheltered aquatic surfaces
.
Environmental Science & Technology
37
(
4
),
772
780
.
Mazot
A.
Bernard
A.
2015
CO2 Degassing From Volcanic Lakes
.
Volcanic Lakes, Springer
,
Berlin, Heidelberg
, pp.
341
354
.
McDowell
M. J.
Johnson
M. S.
2018
Gas transfer velocities evaluated using carbon dioxide as a tracer show high streamflow to be a major driver of total CO2 evasion flux for a headwater stream
.
Journal of Geophysical Research: Biogeosciences
123
(
7
),
2183
2197
.
Morse
N.
Bowden
W. B.
Hackman
A.
Pruden
C.
Steiner
E.
Berger
E.
2007
Using sound pressure to estimate reaeration in streams
.
Journal of the North American Benthological Society
26
(
1
),
28
37
.
Müller
D.
Warneke
T.
Rixen
T.
Müller
M.
Jamahari
S.
Denis
N.
Mujahid
A.
Notholt
J.
2015
Lateral carbon fluxes and CO2 outgassing from a tropical peat-draining river
.
Biogeosciences
12
,
10389
10424
.
Neu
V.
Neill
C.
Krusche
A. V.
2011
Gaseous and fluvial carbon export from an Amazon forest watershed
.
Biogeochemistry
105
(
1–3
),
133
147
.
Owens
M.
Edwards
R.
Gibbs
J.
1964
Some reaeration studies in streams
.
Air and Water Pollution
8
,
469
.
Parkhurst
D. L.
Appelo
C.
1999
User's Guide to PHREEQC (Version 2): A Computer Program for Speciation, Batch-Reaction, One-Dimensional Transport, and Inverse Geochemical Calculations
.
U.S. Geological Survey Water-Resources Investigations Report 99-4259, p. 312
.
Ran
L.
Lu
X. X.
Yang
H.
Li
L.
Yu
R.
Sun
H.
Han
J.
2015
CO2 outgassing from the Yellow River network and its implications for riverine carbon cycle
.
Journal of Geophysical Research: Biogeosciences
120
,
1334
1347
.
Rasera
M. d. F. F. L.
Krusche
A. V.
Richey
J. E.
Ballester
M. V.
Victória
R. L.
2013
Spatial and temporal variability of pCO2 and CO2 efflux in seven Amazonian Rivers
.
Biogeochemistry
116
(
1–3
),
241
259
.
Rawitch
M. J.
2016
Stream Degassing: Review of Methods and Laboratory Validation of Floating Chambers
.
Unpublished MS Thesis
,
University of Kansas
, p.
143
.
Raymond
P. A.
Cole
J. J.
2001
Gas exchange in rivers and estuaries: choosing a gas transfer velocity
.
Estuaries and Coasts
24
(
2
),
312
317
.
Raymond
P. A.
Zappa
C. J.
Butman
D.
Bott
T. L.
Potter
J.
Mulholland
P.
Laursen
A. E.
McDowell
W. H.
Newbold
D.
2012
Scaling the gas transfer velocity and hydraulic geometry in streams and small rivers
.
Limnology & Oceanography: Fluids & Environments
2
(
1
),
41
53
.
Ribas-Ribas
M.
Kilcher
L. F.
Wurl
O.
2018
Sniffle: a step forward to measure in situ CO2 fluxes with the floating chamber technique
.
Elementa
6
(
article 14
),
15
.
Riml
J.
Campeau
A.
Bishop
K.
Wallin
M. B.
2019
Spectral decomposition reveals new perspectives on CO2 concentration patterns and soil-stream linkages
.
Journal of Geophysical Research: Biogeosciences
124
(
10
),
3039
3056
.
Sawakuchi
H.
Rasera
M.
Krusche
A.
Ballester
M.
2012
Methane emission from tropical rivers
. In
Proceedings EGU General Assembly Conference Abstracts 2012
.
14
, p.
11440
.
Schade
J. D.
Bailio
J.
McDowell
W. H.
2016
Greenhouse gas flux from headwater streams in New Hampshire, USA: patterns and drivers
.
Limnology and Oceanography
61
(
S1
),
S165
S174
.
Schelker
J.
Singer
G. A.
Ulseth
A. J.
Hengsberger
S.
Battin
T. J.
2016
CO2 evasion from a steep, high gradient stream network: importance of seasonal and diurnal variation in aquatic pCO2 and gas transfer
.
Limnology and Oceanography
61
(
5
),
1826
1838
.
Sha
C.
Mitsch
W. J.
Mander
Ü.
Lu
J.
Batson
J.
Zhang
L.
He
W.
2011
Methane emissions from freshwater riverine wetlands
.
Ecological Engineering
37
(
1
),
16
24
.
Streeter
H.
Phelps
E.
1925
A Study of the Pollution and Natural Purification of the Ohio River, Illinois, Factors Concerned in the Phenomenon of Oxidation and Reaeration. Public Health Service Bulletin, No. 146
.
Striegl
R. G.
Dornblaser
M.
McDonald
C.
Rover
J.
Stets
E.
2012
Carbon dioxide and methane emissions from the Yukon River system
.
Global Biogeochemical Cycles
26
(
4
).
https://doi.org/10.1029/2012GB004306
.
Teodoru
C.
Nyoni
F.
Borges
A.
Darchambeau
F.
Nyambe
I.
Bouillon
S.
2015
Dynamics of greenhouse gases (CO2, CH4, N2O) along the Zambezi River and major tributaries, and their importance in the riverine carbon budget
.
Biogeosciences
12
(
8
),
2431
2453
.
Tran
T.
Nguyen
N.
Do
H.
Nguyen
D.
2017
Surveying emissions of greenhouse gas CO2 in the canals of Ho Chi Minh city by floating chamber method
.
Science & Technology Development Journal - Science of The Earth & Environment
1
(
M1
),
5
4
.
https://doi.org/https://doi.org/10.32508/stdjsee.v1iM1.436
.
Ulseth
A. J.
Hall
R. O.
Canadell
M. B.
Madinger
H. L.
Niayifar
A.
Battin
T. J.
2019
Distinct air–water gas exchange regimes in low-and high-energy streams
.
Nature Geoscience
12
(
4
),
259
263
.
Xia
Y.
Li
Y.
Li
X.
Guo
M.
She
D.
Yan
X.
2013
Diurnal pattern in nitrous oxide emissions from a sewage-enriched river
.
Chemosphere
92
(
4
),
421
428
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with no derivatives, provided the original work is properly cited (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Supplementary data