The manuscript entitled ” GRiMeDB: The global river database of methane concentrations and fluxes” proposes an a comprehensive database for riverine CH4 and the associate drivers. The proposed databased is based on the earlier work by Stanley et al. (2016). The authors present the flow chart for generating the database, and also the data analysis.



The topic of the manuscript is interesting and relevant to the earth system science community, as methane emission is a potent source of greenhouse gas. Overall, this is a well written manuscript without any apparent flaws. I can recommend the publication of this manuscript with minor revisions.



I also have a minor remark about the ‘first database’ stated in the abstract. I must confess, though, that I did not quite understand the difference between this database and the previous one(MethDB). I think MethDB is the first comprehensive database for river CH4. This work is an extension with significant efforts.

Figs. 5 and 6: Can you differentiate the sites for ebullitive and diffusive flux, respectively. It is very important for modelers.

The figures are very nice and useful. Not all of them need to be in color, though. I also struggled a little bit with the legends: I think Figures 11, 12 and 13 should have legends to show the meaning of the colors.

Dear editor and authors,

My review is attached in the form of a supplement,

Best regards

This data paper substantially updates and expands upon a previous fluvial methane database (MethDB) from a similar group of authors that was published in 2016 (Stanley et al. 2016).  It will be an important resource for the aquatic biogeochemistry community in general.  The paper identifies key gaps in the existing spatial representativeness of stream and river methane emission data (missing data from arid, high altitude, and arctic regions) and also highlights the lack of long time series methane flux data in fluvial ecosystems. The dataset is structured in a unique way that allows users to explore spatio-temporal variation.  Instead of reporting mean emissions for a given system, the database provides within-system site and date specific emissions (and summary statistics across the day when relevant).  The dataset will be a key resource for those interested in upscaling aquatic greenhouse gas emissions.

The paper contains helpful visualizations and tables that orient the reader to key aspects of the dataset.  One useful addition the authors could consider adding is a database schema diagram that shows how unique IDs can be used to link the four.csv files (concentrations, fluxes, sites, and source).  This type of diagram could provide a visual guide for a data user describing which connections are “one to one” versus “one to many”.  For example, on first look at the data files, I found myself a bit confused as to how I would link the flux data to the concentration data.  Is there always a “one to one” link where each row of flux data has a matching row of data in the “concentration” data file? 

Another novel aspect of this database is the inclusion of both diffusive and ebullitive methane emissions and the comprehensive annotation of specific methodological approaches (where the previous MethDB database only reported diffusive emissions).  I think this aspect of the dataset should be mentioned in the abstract.  I also think it would be interesting to visualize and/or further describe differences in the emissions that are generated via different methods.  There is already some discussion of the potential importance of ebullition in overall fluvial methane flux (lines 522-523 citing certain papers), but the authors do not report any statistics regarding the fraction of emissions that are ebullitive in their own dataset (for sites with independent estimates of both flux pathway).  In my work with reservoir methane emissions, the potential predictors of emission became clearer when we stopped combining diffusive-only emission estimates with those that integrated across both flux pathways (Deemer et al. 2016).   I think this division may be a bit harder to make in the river literature (since it may be harder to discern which estimates were truly diffusive-only), but I think some basic summary of the flux data by method would be helpful.  In looking at the data, it looks like you only have 8 rows of data with total methane flux recorded.  You do mention that 85% of the data is diffusive-only, but I’m surprised there are only 8 rows that have estimates of both diffusive and ebullitive emission.  This could be explicitly called out.  

Thank you to the authors for this important contribution to the field.

Line by Line Comments

Line 35- If you can fit it, I suggest including the ranges and/or standard deviations here

Lines 88-95- I assume beaver ponds were not included as “marginal” fluvial systems, but you might explicitly mention this here.  Also, what about river reaches upstream of weirs?

Line 131-132- Figure 2 doesn’t really make this distinction regarding sites that were used in multiple studies.  Consider either adding this or annotating it somehow directly in figure legend.

Line 175- Consider changing the wording in this title (and/or in the text directly below it) to “Concentrations Table and Fluxes Table” to make extra clear that they are two separate tables.

Line 183- Delete duplicate use of the word “both”

Lines 221-228- So, is it true that in some cases the same concentration data might be applied to many rows of flux estimates (one to many)?

Lines 287-289- This pattern is also true for lake and reservoir methane data—65% of the lake/reservoir methane emission estimates in a recent dataset were collected since 2015 (Rosentreter et al. 2021, Deemer and Holgerson 2021)

Lines 292-294- Wow!  I can’t believe how short the longest flux record is—much shorter than the lake literature.

Line 335- Do you mean 4% of the global river surface?  I don’t think you mean land surface from looking at the map (more than 4% seems to be shaded darker tones of orange, but some of this is in surface-water poor areas like the Sahara).

Line 408- Include the definition for “IMP” like you do for the other site types.

Figure 12- The relationship between flux and total N & P looks stronger than for DOC or dissolved oxygen, but this isn’t called out where you discuss drivers (lines 571-590) .  You might consider citing some of the wetland, lake and reservoir literature that has linked methane emission to productivity/chlorophyll a.

Line 480- You could discuss insights on spatial/temporal resolution from the lake literature here.  Wik et al. 2016 Geophysical Research Letters showed that spatial and temporal under-representation generally led to underestimates of emission in lakes.

Line 509- Remove either “few” or “several”

Line 522- It isn’t clear if this 30-90% range comes from your entire database, or just from the few papers cited here.  Or maybe there are only three papers that quantify both pathways together?  In the lake and reservoir literature, the fraction of emissions that are ebullitive can range dramatically (undetectable to almost all of the emission), with ebullitive emission contributing a median of 78% of methane emissions in reservoirs and 54% in lakes- Deemer and Holgerson 2021).

Line 584- Rosentreter et al. 2021 also used latitude to upscale stream and river emissions.

Line 599- I thought Burns et al. 2018 reported rather high methane emissions from glacial systems? 

Line 628- Add the word “from” between “data” and “world”

Tables A3 and A4- I suggest explicitly clarifying that “new” units are the relevant units for the data you report.