Summary of the manuscript

The reviewed manuscript by Koschorrek et al. quantifies variability of methane and carbon dioxide fluxes between the atmosphere and a temperate low-land river at scales of hours and hundreds of meters. Based on a three-day sampling campaign, including flux chamber measurements in the river and in nearshore areas, the authors found considerable diurnal variability in carbon dioxide fluxes and variability from near-shore to off-shore areas in methane fluxes. The authors also discuss consequences of different sampling strategies for upscaled gas fluxes, concluding that accurate flux estimates require continuous measurements.

Overall assessment

Scientific significance: As well introduced by the authors, rivers play an important component of the global carbon cycle and emit carbon gasses at globally significant rates. Yet, there are large uncertainties in emission estimates due to very large spatiotemporal variabilities. The research question on spatiotemporal greenhouse gas fluxes in rivers is not particularly novel, but the focus of this study on small-scale variations (diurnal, near-shore / off-shore) fills a poorly studied niche in the literature that is well worth investigating. I also appreciate the comparison of aquatic and terrestrial gas fluxes, which is rarely done, but highly relevant given that rivers can vary largely in their aerial extend, depending on discharge fluctuations.

Scientific quality: The scope of the study including 3 days of measurements in a 1 km river reach may not appear overly impressive and representative for other conditions. However, relative to many other studies, the authors managed to collect an impressive and interesting data set at very small spatial and temporal scales. Overall, the authors address the research question by using state-of the art techniques. The study design could be acceptable, overall, but some design-related questions should be addressed first (see major concerns below). I agree with most data interpretations and conclusions, but a mismatch in the results shown should be resolved (see major concerns below). I also have a few concerns about the statistical analysis of the data, as outlined below.

Presentation quality: The manuscript is well written, logically structured and clearly and concisely presented. Overall, the figures and tables, including the supplementary material, are adequately chosen and well designed, but I have some concerns and suggestions for improvements, as listed below.

Overall, I find that the manuscript is well within the scope of Biogeosciences.

Major concerns

A main focus of the manuscript is to compare spatial and temporal variability in gas fluxes. I wonder to what extent this analysis may be biased by the fact that spatial and temporal assessments were not fully independent? I understand that for practical reasons (limited availability of gas analysers), it is impossible to perform simultaneous measurements at the different locations. However, I would expect a discussion on the consequences of the sampling design for the analysis of spatial and temporal variability in aquatic gas fluxes. For example, I would like to see at what time the different floating chamber measurements were performed. Given that each measurements takes 2-5 min, I would expect that daytime may affect measurements, in addition to location. Did the authors account for time in their assessment of spatial variability?

Related to the major comment above, it is unclear to me how potential temporal variability in the gas transfer velocity was accounted for in calculations of diel gas fluxes. I appreciate the high temporal resolution of dissolved gas concentrations, but for accurate calculations of gas fluxes, temporal variability in k should also be characterized. K may or may not vary on a diel basis (see e.g. Attermeyer et al. 2021 Comm. Earth&Env). Please clarify how time series fluxes were calculated and discuss any potential shortcomings, in case concentration and k estimates differ in temporal resolution.

I think there is a mismatch in gas fluxes and concentrations shown in Table 1 and in Figures 3/4. According to Table 1, CO2 fluxes range up to 13.9 mmol m-2 h-1, with medians up to 2.8 mmol m-2 h-1. In contrast, the Figure 4 shows maximum fluxes of near 30 mmol m-2 h-1 and medians of up to 10 mmol m-2 h-1. Also, CH4 concentrations in Figure 3 range up to 240 nmol/L, compared to 320 nmol/L in Table 1. Shouldn’t the data shown in Table 1 and Figures 3/4 be the same? Table 1 suggests no considerable difference in CO2 fluxes between aquatic and terrestrial habitats, while Figure 4 does. The mismatch may have implications for results (L. 216) and conclusions (L. 432). This issue must be addressed, through corrections or clarifications, before the manuscript can be considered for publication.

The authors mention the major effect of salty water inflow (river Saale) affecting water chemistry along the western shore (L. 203-206). The authors sampled the western shore and main part of the river, but not the eastern shore, which seems not to be affected by the salty water inflow. I understand that the focus of this study was on the Groynes located along the western shore. However, given the focus on spatial variability of this study, I think it would have been valuable to also study the eastern shore as a “reference” to better evaluate the effect of the Groynes and the salty inflow. Why did the authors did not do any attempt to also study the eastern shore? To what extent could the salty inflow have affected results? Would there be any way to disentangle the spatially overlapping effects of the salty water inflow and the groynes? I would appreciate a brief discussion on this issue.

What is the role of ebullition for gas fluxes in the studied system? Given the potentially large role for total fluxes as well as spatial and temporal variability of methane fluxes, I think this should be discussed more in the manuscript (extending the statement in L. 422). In particular, did you observe sudden jumps in the within-chamber gas measurements that would indicate ebullition? If so, how did you treat such data and how would the exclusion of ebullition affect gas flux estimates?

I would like to see more details on the statistical analyses used. For example, the choice of methods described in L. 187-192 should be justified and the used R functions / packages should be explained/cited. What explanatory variables (fixed and random effects) were investigated in the Linear mixed models (L. 191)? Were fluxes always positive so that log-transformation is justified (L. 190)? How was temporal / spatial autocorrelation tested/accounted for in the analyses? How does the correlation analysis and linear mixed effects modelling help to address the stated research question? Can you please add details of statistical analysis (Wilcox test statistics, mixed effects model parameters / AIC, degrees of freedom). This could be added in the main text or as tables, e.g. in the supplementary material.

Specific comments

L. 14: Can the authors motivate their statement that most existing studies were carried out in small streams? Perhaps by referring to published work (review, metaanalysis). Personally, I don’t have a complete / up-to date overview of the existing literature, but I don’t necessarily have the impression that smaller streams are represented more than larger rivers. For air-water gas exchange work in larger rivers, see e.g. Yao et al. (2007, Sci Total Environ), Alin et al. (2011, JGR), Hall et al. (2012, L&O), Beaulieu et al. (2012, JGR), Striegl et al. (2012, GBC), Huotari et al. (2013, GRL), Borges et al. (2016, Nat. Geosci.), Qu et al. (2017, Sci. Reports), Rosentreter et al. (2017, L&O), Paranaiba et al. (2018, ES&T).

L. 28 This may be a matter of taste, but could the title “Necessity of upscaling/quantification of GHG emissions from rivers” be shorted? Starting the manuscript with a less bulky title may approach a wider readership.

L. 37 Raymond et al. (2013) relied mainly on calculated CO2 based on pH, alkalinity and temperature, not “measured concentrations” as written here.

L. 38 Perhaps “gas transfer velocities” could be defined/introduced to make the manuscript more accessible for a wider readership?

L. 38 The term “multiplied” confuses me, because the other terms of the equation that is referred to here (concentrations, gas transfer velocity) are simply mentioned without any mathematical characterization of their relationship. I suggest to rephrase the statement to be more consistent in the language.

L. 38 I agree that most datasets seem to contain weekly or monthly data, but could the authors provide (a) reference(s) for their statement? Perhaps a metaanalysis/review? For example, Marx et al. (2017, Reviews of Geophysics) mentions “knowledge gaps with respect to high-resolution temporal (i.e., diurnal) and spatial variations of carbon fluxes”.

L. 40/ L. 124 I agree with Lorke et al. (2015, Biogeosciences) that floating chamber measurements can be problematic in flowing water. This has also been evaluated by Vingiani et al. (2021, Biogeosciences) under a range of hydraulic conditions. I would appreciate if the authors could give more details in the methods section on their floating chamber design. How did the authors minimize potential experimental artifacts (e.g. by using “flying” chambers such as described by Lorke et al. and Vingiani et al.)?

L. 53-54 Please provide (a) reference(s) to support the statement “While a single water sample might be representative of a certain specific reach in a small stream this is undoubtedly not the case in larger rivers.” Why would spatial variability be higher in larger systems? Greenhouse gas fluxes can be highly heterogeneous in headwater systems (see e.g. Marx et al. 2017, Reviews of Geophysics; Lupon et al. 2019 L&O; Horgby et al. 2019, JGR). I am not aware of any systematic analysis of variability relative to system size, but I would be happy if the authors can substantiate their statement.

L. 80 Elsewhere in the manuscript it says the campaign was 3 days long, but here it says 4 days. Can you clarify this difference, please?

L. 91 Why was the outer boundary of the groyne fields set to 15 m into the river? Is this based on previous research?

L. 116 Measurements of turbidity and chlorophyll are mentioned here but there is no data shown. This should be consistent. I would be happy to see data on chlorophyll as it could indicate the level of primary production and hence provide important context to diel CO2 concentrations.

L. 118 I had to look up the term “moon pool of the albis”. I can imagine that there are more potential readers that are unfamiliar with this term. Consider clarification.

L. 123-174 The authors used three different portable gas analyzers and a gas chromatograph. Have you performed cross-characterizations of the analyzers to make sure that concentration measurements and flux estimates are comparable between the study systems?

L. 139 Which “instruments” are referred to here?

L. 153 I appreciate that CO2 was measured in the air continuously. However, I cannot find any data on this in the manuscript or any statement on how the data was used. Did you use this data in flux calculations?

L. 158 “Sampling points” are mentioned here, but I would appreciate a clarification of the exact sampling setup, perhaps already in the section with the study site description. How many chambers were deployed in total / per vegetation zone? This is implicit in Figure 1, but it is not clear to me until this point, whether chambers were deployed in all Groyne fields. Also, based on what criteria was the location of the soil flux chamber chosen? Fig. 1c) suggests that the vegetated site C3 was located very close to muddy area, which makes me wonder how representative this site was for the vegetated area?

L. 179-182 Gas transfer coefficients were calculated from CH4 fluxes and then converted to CO2. This conversion could potentially be erroneous in the presence of bubbles (Klaus et al. 2022, JGR), so I would appreciate a brief note on the role of bubbles in gas exchange in the study system.

L. 185-186 It is unclear to me why “Probe measurements of CO2 and CH4 concentrations were converted to fluxes using the measured gas transfer velocity of k600 = 5.5 m d-1 (Table 1).” Why was this constant value chosen here, given that it varies substantially, as the data in Table 1 suggests.

L. 186 Please clarify “converted to kCO2 and kCH4 as explained in Striegl et al. (2012)”. Do you mean the Schmidt number conversion as explained in L. 182? What is the difference between the conversions you mention in L. 182 and 186?

L. 187-188 Please clarify the definition of “day” and “night”. Some details are given in Table 2, but they should also appear or be moved to the methods section. Figure 5 suggests some offset between the timing of day-night shifts and changes in PAR. What is the reasoning behind this offset?

Figure 3 I appreciate this map, but I wonder why spatial patterns are only shown for CH4 and not for CO2? I would like to see a map of CO2 measurements.

L. 226 What criteria did you use to extract the areas manually from a google earth image? Was there a clear division between the different areas or is the manual extraction prone to uncertainties/errors?

Figure 4 It is not totally clear to me what the p-values of pair-wise comparisons refer to. Three values are given, but they are all aligned with the same arrow. Please modify the arrows so it becomes clear which p-value belongs to which comparison.

L. 235-236 Data on pH and O2 is mentioned here, but I cannot find this data in Figure S5. Please add the data to the figure.

Figure 5 Please add units of light and temperature to panel c).

L. 259 / Figure S6 Why did you perform the correlation analysis only for CO2 fluxes, but not for CH4 fluxes?

L. 260 Please clarify how you treated the “high” autocorrelation of light and temperature in the statistical analysis.

L. 267-268 The statement on fixed and random factors should be moved to the methods section.

L. 269 Do you mean the most parsimonious model?

L. 279 Why did you not measure temporal changes in CH4 flux at the dry sites?

L. 302 and L. 383 Can you provide a reference for the statement that CH4 is primarily produced in the sediment? Methane can potentially also be produced in the water column, even under oxic conditions (e.g. Guenthel et al. 2019, Nat. Comm.). Is anything known about the sources of CH4 in the Elbe river or other lowland rivers?

L. 307 I cannot find any statistical support for the statement that there was “a significant difference between aquatic and terrestrial CH4 emissions”. Please provide this support (e.g. in Fig. 4).

L. 319 I appreciate the pioneering effort on small-scale spatial variability in gas transfer velocities. However, the spatial variability is not explicitly shown in the manuscript. Data on k600 is given in Table 1, but only median values and ranges are shown and it remains unclear whether they represent spatial or temporal variability. While showing variability in k600 is not critical to the focus on fluxes in this paper, it might still be interesting to provide more detailed information on spatial vs temporal variability (e.g. CV) in k600 and gas concentrations in the river. I leave it to the authors to decide whether they want to add this information to substantiate the statement in L. 319, or whether they want to leave this part of the story out.

L. 319-321 The floating chamber is not the only method applicable to rivers. See Huotari et al. 2013 (GRL) for deployment of the eddy-covariance technique in a river. Also, what is a “large stream” relative to a “river”? I suggest to use consistent terms throughout the manuscript.

L. 326-329 The statement “While higher k600 values at the side of the river were expected,” leaves me to wonder why you expected this. The explanation comes indirectly in the following sentence, but perhaps you can rephrase / reorder the sentence to improve logic?

L. 331-333 Why do the authors refer to stream metabolism here? I agree that k is critical to metabolism calculations based on the free-water oxygen method, but there could be many other examples on exchange of other gases (e.g. Hg, Rn) where k values are relevant. Perhaps rephrase the sentence to reflect that metabolism calculations is just one example where k is relevant?

L. 342 Why did you exclude plants from the chambers?

L. 357 I would replace “obviously” by “most likely”, to reflect uncertainties.

L. 362 Abbreviation “DIC” should be explained.

L. 385 I don’t quite understand why the CH4 pool in the water would buffer fluctuations in CH4 emissions caused by CH4 oxidation at the sediment interface. CH4 could also potentially be oxidized in the water column. Some of the co-authors show this for the Elbe river estuary (Matousu et al. 2017, Aqua Sci). Can you please substantiate your argumentation, e.g. by referring to references?

L. 405 / 414 Please use consistent terms to describe the “optimal” or “perfect” approach.

Supporting Information: Please clarify the symbols and units of the data included in the excel sheet that contains Time series data. Why are is the spatially resolved data not provided?

Figure S2 I think it would be useful to show a length metric as x-axis.

Figure S4 Please add letters to the panels. Also, I think that either the panel headings or the the panel descriptions for b and c in the figure caption are mixed up.

Figure S6 Please give a clarification of the symbols / abbreviations used in the figure.

Technical notes

L. 21 imoprove -> improve

L. 54 or those or -> delete ”or”

L. 98 The lines in Fig. 1 b are orange, not red as indicated here

L. 187 Here, both “emissions” and “fluxes” are used. I would suggest to use consistent terms throughout the manuscript. Personally, I prefer the more neutral term “fluxes”.

L. 259 Remove “tried”. It is apparent that you did the analysis.

L. 395 Remove “very”

L. 399 logyrhytmic -> logarithmic

L. 406 under estimation -> underestimation

The work is within the scope of the journal as it addresses CO2 and CH4 pattern in some typical European river. The approach is sound, careful and the arguments are generally convincing

My first major issue – on the real usefulness of diel monitoring of CO2 emissions in rivers.

The sentence in L 25 of the Abstract is questionable. It is unclear why continuous measurements of fCO2 are really necessary given that day/night median values in water habitat (Table 2) which mostly contribute to C emissions (Table 1) are same, within +/- 20% (Table 2). This uncertainty is within the internal measurement uncertainty and hence can be neglected.

The second issue is about potential importance of diurnal variations on the annual scale. Peak summer season and sunny weather may not have the same CO2 pattern as cloudy and cooler weather during other period of the year in this part of Germany. At present, extrapolation to year-round time scale is not warranted

Some specific issues

L77-79 Please provide a justification for this hypothesis, based on available literature

L201 The rainfall during continuous monitoring is rather unfortunate and highly undesirable event, adding a new dimension (variable). Please explain how it was taken into account.

Fig 3 is useful; however, the same plot for pCO2 is needed. Please also consider presenting a plot for CO2 fluxes of this kind to demonstrate spatial variability.

L310 Is this ‘small’ in GHG potential equivalent, or ‘small’ for the total C balance?

L315-316 What is the reason of lower Kt at high pCO2 – lower turbulence?

L353-355 This statement is too general. See works on large tropical and temperate rivers (Mississippi, Congo) or subarctic rivers. On the latter (for instance, Taz, Ket, Lena, see https://doi.org/10.5194/bg-19-1-2022; https://doi.org/10.5194/bg-18-4919-2021; doi: 10.3389/fenvs.2022.98759), the diurnal dynamics of CO2 emissions is not strongly pronounced.

L368 In water, there is no constant CO2 flux during the night (Fig 5a)

L372-375 Justification of analogy with marine sediments is necessary

Comments to the manuscript by Koschorreck et al, “Diurnal versus spatial variability of greenhouse gas emissions from an anthropogenic modified German lowland river”

Overview:

In this study the authors investigate the temporal and spatial variability in greenhouse gas emissions along a German lowland river that has been heavily impacted by anthropogenic activities. Specific focus is on how different greenhouse gases (CO2 and CH4) differ in their respective variability between different locations along a 1 km river reach and over a diurnal temporal scale. The study is based on a three to four days (dependent on variable) long measurement campaign where different locations (middle and side of river channel, different habitats at the terrestrial-aquatic interface) within the river are monitored. The authors conclude that the variability in aquatic emission is gas-specific and that CH4 are more variable on a spatial scale than CO2, whereas CO2 is more variable on the diurnal scale than CH4. They further show that the non-aquatic parts of the river (i.e. parts that are temporally flooded but that were not flooded during the measurement campaign) contributed 10% of the total GHG flux during the campaign.

The manuscript focus on an important topic that is very suitable for publication in Biogeosciences. Although I do not fully agree with the authors that large rivers are much more heterogeneous in their spatial variability of GHG emissions than small streams, I agree that large rivers are somewhat less studied. I also agree that, due to the relative lack of data, GHG emissions from large rivers are often estimated without any validation. Although small streams (< stream order 4) dominate the total global stream and river length, the water surface area that larger rivers (> stream order 4) are representing is a large share of the total surface area of running water. Hence, understanding the temporal and spatial variability and what controls these variabilities are essential for representative GHG emission estimates.

General comments:

With this background the manuscript is an important contribution to the research field. I appreciate the detailed and small scale focus of the study which highlight fundamental differences in how CO2 and CH4 are sourced from a central Europe (and likely also elsewhere?) common type of river environment. The manuscript is in general well written and presented but I have some points that needs to be clarified/added prior to a publication of the study. 1) I have some problems to follow the method section and the structure of it. There are also some unclear parts of the methods which at least to me is confusing and which makes it a bit hard to interpret the results. 2) The study is based on a variety of different measurement approaches and setups for capturing GHG fluxes at different habitats/scales. It is currently hard to assess the uncertainty of each approach which makes it hard to understand their absolute or relative difference when measurements are compared. 3) The conclusion section of the study could be stronger given the high detail of the measurements. Currently the text is rather vaguely written and not fully representing, or capitalizing on, the outcome of the study.

Detailed comments:

Ln 15-16, Although not clear but I interpret this sentence starting with “Here quantification…”  that small streams (which are mentioned in the sentence before) are not displaying spatial and temporal variability in their GHG emissions. If this is what the authors mean I strongly disagree. I would claim that the spatial and temporal variability in GHG emissions could be even more pronounced in small streams. I however certainly agree with the authors that small scale assessments of temporal and spatial variabilities are rarely made in larger rivers and even less simultaneously. I suggest that the authors rephrase this sentence.

Ln 17, It later on says in the text that the campaign lasted for four days. I suggest that the authors keep it consistent although I think I understand that the difference stems from that different variables/habitats were measured during different number of days.

Ln 24-25, yes the data confirms the hypothesis, but I think it needs to be transparently stated that this was just true for this river section and during the three or four days long campaign. Whether this is a more universal pattern requires measurements covering more extensive temporal and spatial scales.

Ln 29-31, I think the authors can update the CH4 referencing, here and elsewhere suitable, with the very relevant and recent global stream and river studies (Rocher-Ros et al. 2023; Stanley et al. 2023).

Ln 53-54, I think this motivation statement (comparison with small streams) to why it is necessary to study small scale variability in large rivers is not really true and also not really needed. I have measured very high small scale variability along stream channels in both concentrations and emissions of different GHG´s (especially for CH4). In comparison, I have also measured low spatial variability in GHG´s across larger rivers. To conclude, whether concentrations and emissions are spatially variable are highly site specific and not necessary related to the size of the water body. I agree though that little is known about spatial (and temporal) variability in larger rivers, a good motivation of the study as such. I suggest a rephrasing of this statement.

Ln 80, in abstract “three days’ campaign”. I suggest to be consistent.

Figure 2, what is meant with “mean low discharge”? Unclear to me.

Ln 107, section header. This section contains more than just flow velocity and depth measurements, I suggest to give a more suitable header.

Ln 111-113, These velocity measurements are more suitable for the result section to me, or why are they place in the methods?

Ln 117-119, “The water supply for both sensors was the moon pool of the Albis”. This comes without any introduction to the reader, what is moon pool and what is Albis?? Please clarify!

Ln 122-147, This section is a bit unclear to me, and at the same time the core of many of the measurements included. I suggest that the authors go through it and make a more logical structure of the text. For example:

• Are the same spatial measurements described starting in the lines 123 and 139? If so, why are they separated in the text? If not, what is the difference between them?
• The dissolved gas mapping described in ln 130, why was it just done for CH4 and not also for CO2? I though the LGR instrument handled both gases?
• The conversion of ppm values of CH4 to concentrations determined from the mapping was made by a regression equation I assume? I suggest to show the data for this conversion in the SI
• Ln 144, again, what is the moon pool of Albis? Maybe obvious for a “ship-based” researcher but not for me.
• Ln 145-147, how was the CH4 ppm values from the Contros converted to absolute concentrations, similar to above I suggest to show this conversion in some way. As the use of CH4 sensors are in the forefront for this kind of research it would be highly useful for other researcher to show how this was done.

Ln 148, this section contains more than “terrestrial measurements”. The first part of the section describes for example atmospheric measurements. I suggest the authors give a more suitable section header.

Ln 185, what probe measurements? The ones conducted at the ship? Please clarify!

Ln 186, why was a fixed value of k600 (5.5 m d-1) used? I don’t see 5.5 in Table 1 as referred to. Also, what habitat do these fluxes represent?

Ln 213, I believe it should be “at the sides” instead of “at the sites”? or?

Figure 3, why was not CO2 measured at the same time?

Ln 243, “were” instead of “are”.

Figure 5, I assume the time point 12.00 on the x-axis refer to mid-day, I suggest to clarify this.

Table 2 and related text. Here comes one of my larger concerns. It is currently hard to assess how the different method approaches correspond to each other. Although a range is given for all emissions it is hard to understand with what certainty each individual emission determination (done with a different measurement approach) is made with. Some kind of uncertainty estimate would certainly help the reader with this.

Figure 6, similar to above, the CV values are good and illustrate well the variability associated with each gas, variability and habitat. However, to what extent these differences in CV values are dependent on variable certainties in the different methods involved in measuring these emissions is currently unknown for the reader.

Ln 320, The sentence that starts with “The floating chamber….” Used for what? Do you mean GHG emission measurements? What about the eddy covariance method? There are a few examples of the application of the EC method on rivers e.g. Huotari et al. 2013, Guseva et al. 2021. I suggest to rephrase this sentence.

Ln 390, here refered to as “stream”, in other places “river”, I suggest to be consistent and use river. Here and elsewhere in this section specifically, but also throughout the ms.

Ln 399, “logyrhymtic scale”??

Ln 405 and onwards, I appreciate this exercise of simulating different monitoring approaches and what you might miss/capture with a certain approach. This is good information for the reader and it also put some more perspectives on the detailed data basis that this study provides.

Ln 426, I suggest the authors rewrite the conclusions to make them stronger and better reflecting the outcome of the study. I appreciate the starting sentence that the study “just” represent a snapshot of one river reach, still I think the conclusions could be more concise, focused and with a stronger message. The study deserves that.

References

Guseva, S., Aurela, M., Cortés, A., Kivi, R., Lotsari, E., MacIntyre, S., et al. (2021). Variable Physical Drivers of Near-Surface Turbulence in a Regulated River. Water Resources Research, 57(11), e2020WR027939.

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), 2013GL057705.

Rocher-Ros, G., Stanley, E.H., Loken, L.C. et al. Global methane emissions from rivers and streams. Nature 621, 530–535 (2023). https://doi.org/10.1038/s41586-023-06344-6

Stanley, E. H., Loken, L. C., Casson, N. J., Oliver, S. K., Sponseller, R. A., et al: (2023) GRiMeDB: the Global River Methane Database of concentrations and fluxes, Earth Syst. Sci. Data, 15, 2879–2926, https://doi.org/10.5194/essd-15-2879-2023.