An Analysis of: Spatio-Temporal Patterns of Stream Methane and Carbon Dioxide Emissions in a Hemiboreal Catchment in Southwest Sweden


Even though they don’t cover a large area of the world, streams are an important contributor to total aquatic CO2 emissions; the estimated emissions from streams and rivers around the world exceed those of lakes and reservoirs.  The large emissions from streams is partially due to high concentrations of CO2 in streams as compared to lakes – and the gas transfer velocities are higher due to the turbulence of the water.  But though these emissions have been shown to be important, they are not typically included in carbon budgets because of the lack of available data on the subject; any previous estimates are based on data with poor spatio-temporal considerations. Spatial variability of these emissions is linked to the variability of k (due to differences in areas of stream morphology); temporal variability has been linked, on the other hand, to the variability of stream discharge (which leads to changes in the ground water level and the related input of gases – plus, large discharge events like melting snow or rainstorms can rapidly affect gas concentrations).  To make comparisons easy between values, all measurements are made to fit in relative to k600, which is the number for CO2 at 20 degrees Celsius.

Current data, though, doesn’t allow for a study of how spatio-temporal variations affect annual gas emissions; the science needs more long-term studies, since most large-scale studies of this type are of relatively short duration. There is also a need for more direct measurements, since most measurements of CO2 in particular are taken indirectly and have a tendency to overestimate concentrations in particular environments because of the methods used.



A map of the Skogaryd Research Catchment (SRC) showing the studied streams, the catchment boundaries and sampling locations.

This study was conducted in the Skogaryd Research Catchment in Southwest Sweden; this network is a mixture of flat, slow-moving areas and steep terrain (creating small waterfalls and turbulent waters).

The researchers worked in four steps: they first measured the spatio-temporal variability in k in “key” reaches of the streams network, and then used these measurements to create a model for k values for the entire network; using this and direct concentration measurements, they were able to derive CH4 and CO2 emissions that took into account spatio-temporal variability; finally they paid specific attention to the importance of turbulent sections and high discharge time periods.

To measure gas transfer velocities, gas transfer coefficients were determined by injecting volatile gas tracers into the water (propane was used) in streams ranging from 20 to 54m; the velocity of the moving gas was then measured.

CO2 concentrations were determined using plastic chambers that were equipped with CO2 sensors; 8 chambers were used in 2013, and 20 were used in 2014. In both years a total of 28463 measurements were taken.

CH4 measurements were determined using a headspace equilibration method, and measured approximately every two weeks in both years at four different monitoring stations; a total of 292 measurements were taken.

The study lasted 623 days, but excluded periods where the streams were frozen. Finally, they used the collected measurements from the tracer injections and used these to estimate velocities for all 84 sections of the stream network. Daily emissions were collected for each of the 84 sections using these specific k estimates; finally, the emissions from all the sections were added together to get the total CH4 and CO2 emissions.

To analyze the data, the concentrations, k600 and emissions for CO2 and CH4 were divided into seven groups based on location (L1 to L7) and into five groups based on slope (S1-S5).


The researchers divided the stream into 84 separate areas, based on a change in elevation of 0.5 m; this means that in areas of higher slopes the sections were shorter and in relatively flat areas they could be much longer. What the researchers found was that areas with a higher slope (including waterfall sections) lead to a greater k600 than those with flatter topography; the relationships between k600 and discharge or water velocity were strong for the areas with high slopes, but weak or nonexistent in the flatter areas.

The mean of the modelled k600 was different for different locations, slopes and times. The modelled mean of the highest discharge period was over three times higher than the overall mean k600 from the two years, but flooding events in high slopes could in themselves cause up to a seven times increase in k600.

The importance of spatio-temporal variability for total emissions and concentrations:

Velocities and turbulence seem to be the most important predictors of emissions and concentrations of CH4 and CO2 in this system.

To use specific examples, L7 had the highest mean k600; it also had CH4 and CO2 emissions which were three times higher than the overall mean. L5 had the lowest k600 and lowest mean CO2 emissions. Even though they had differences, CH4 and CO2 emissions from each area were higher in the case of higher discharge periods.

The slope areas with the highest velocities had the lowest CH4 and CO2 concentrations; concentrations were steadily lower downstream because of the added turbulence.

There were higher emissions in high slope areas and in periods of high discharge; this may seem obvious, but when we look at the facts it can be much more startling. Even though topography with a slope <1% occupied 90% of the stream’s total areal coverage, the mean CH4 and CO2 emissions (per m2) of these areas were 5 times lower than the overall mean. This shows us that the emissions are much higher in the high slope areas, even though they are quite small in surface area. Specifically, for example, S5 (which only represented 0.9% of the areal coverage) had mean emissions 3 times higher than the overall mean for CH4, and 4 times higher for CO2. S4 and S5 together, still only covering a meager 2% of the area, were the source of 18% of CH4 emissions and 30% of CO2 emissions. And with all of this being said, the emissions estimates of this study were on the conservative side – and estimates from turbulent areas were, in fact, underestimates.


Daily emissiosn of CH4 and CO2 from the studied streams in the two years

Discharge and mean emissions had a positive correlation; as the discharge went up, so did the emissions. On the few days where discharge was over 4 times the mean (5% of the overall period), the emissions of CH4 and CO2 were more than 3 times greater than the overall mean. In general, high discharge periods only occurred approximately 10% of the time – but they were responsible for 37% of CH4 emissions and 43% of CO2 emissions. This shows the real importance of accounting for temporal changes in emissions – though these periods do not last long, they are serious contributors to emissions. This also illustrates the importance of representative sampling – taking samples from one high turbulence area may overestimate emissions, but avoiding these areas altogether means seriously underestimating gas emissions.

Though steep sections may only have small gas concentrations, their turbulence means that a large percentage of these gases are being emitted – making them important areas to consider in an emission budget. In the same way, though high discharge periods may only occur seldom on an annual basis, they create a large amount of emissions. This means that changes in the hydrological patterns of an area could mean large changes in emissions. And yet, with all of this being said, studies on high discharge areas are still few and far between.


This study shows the importance of observing both changes in discharge and slope, over time and in different areas of a stream network, if we are to get accurate emissions estimates. Future work is necessary on exploring all of the different facets that comprise these emissions levels, but these researchers have started an important inquiry into the levels of greenhouse gases operating in our world.


Research performed and analysed by: Sivakiruthika Natchimuthu, Marcus B. Wallin, Leif Klemedtsson & David Bastviken