This well-written manuscript presents a large, compiled dataset of carbon concentrations, fluxes, and isotopes for rivers in Australia. The dataset comprises DOC, DIC, POC, CO2, and CH4, while the manuscript describes DOC, DIC, CO2, and CH4.  This dataset has the strong potential to support an analysis of riverine carbon fluxes for the Australian continent and includes new, unpublished data.  Overall, I find the manuscript and dataset comprehensive and well described, and the dataset will likely support an important analysis of continental-scale carbon fluxes from a relatively understudied region. However, I have a few comments that should be addressed.

First, it might be worth re-considering not describing the POC data in the manuscript. I understand that the authors choose not to describe the POC data because of the relatively small number of samples compared to the other parameters (text says n=51, I’m seeing n=77 in the dataset).  However, POC is a part of the dataset, exhibits a large range of values (4 orders of magnitude), and is a relatively unconstrained riverine carbon flux, so the data is valuable.  If the authors don’t want to make direct comparisons between POC and DOC/DIC/CO2/CH4 given the very different number of samples, they could include a description of the POC data in a separate sub-section.  That being said, it might also be interesting to see POC in Table 2 and Figure 5.  

Second, the prevalence of river types could be considered in the discussion of spatial coverage and representativeness of the carbon dataset.  Figure 1, Table 1, and the associated text very clearly point out that arid rivers aren’t well represented in the dataset.  This point is important in understanding the dataset. However, I also wonder what proportion of streams and rivers in Australia are found in the arid interior?  How different is the proportion of samples from arid rivers (~5%) versus from non-arid rivers compared to the proportion of arid to non-arid rivers in Australia?

Other comments:

Ln 134-135: This sentence should be revised. As far as I understand, the authors did not calculate CO2 from literature values of pH and alkalinity, but rather only used literature values of calculated CO2.

Ln 182-183: Headspace equilibration samples for CO2 should be corrected for carbonate equilibria (Koschorreck et al., 2021).  At the pH range of the dataset (1^st-3^rd Quartile: 7-7.8), bicarbonate would be the dominant form of inorganic carbon, and not CO2, so this correction might be important. The authors have DIC and pH so they could apply this correction to the pCO2 values or show that it doesn’t have much of an effect.

Ln 246: Needs to be reworded to make clear that this comparison refers to observations, not concentrations. E.g., “DIC and DOC measurements were the most prevalent measurements in the dataset, across all climatic regions”

Ln 268-9: combine this stand-alone sentence into the next paragraph.

Figure 2 – It’s difficult to see the colors in the legend – consider making the lines in the legend thicker.

Figure 4a – It’s difficult to read the words in the pie chart.

Ln 306 – “The large difference for CO2 estimates *may arise* because…”

Ln 848-861 – How were the study-reach lengths choosen? What were the range of reach lengths?  Were there any groundwater inputs or lateral flows along the study reaches? If so, were dilution corrections applied?

Ln 866-876 – Generally direct measurements of CO2 are much better than calculations, given all the errors associated with pH measurements and non-carbonate alkalinity. Why were the headspace replicates so frequently (36% of samples) off by > 20%?   

Ln 877-881 – Aren’t estimates of velocity derived from a pair of breakthrough curves/EC sensors? How can only one break through curve be used?

Figure C1 – difficult to read the text on this figure.

Ln 890 – extra “40” before oxygen.

Dataset: There are two extremely high [O2] values in the dataset (37.4 and 98.6 mg/l)

Dataset: More than 6,000 entries don’t have a catchment area. It should be possible to delineate a catchment using a DEM and estimate catchment area.

References:

Koschorreck, M., Prairie, Y. T., Kim, J., & Marcé, R. (2021). Technical note: CO2 is not like CH4 – limits of and corrections to the headspace method to analyse pCO2 in fresh water. Biogeosciences, 18(5), 1619–1627. https://doi.org/10.5194/bg-18-1619-2021

I commend the authors on compiling and presenting this highly valuable and timely dataset of aquatic carbon parameters across Australia. The study addresses a significant gap in the global carbon cycle research, particularly for the under-sampled Australian continent. The collection of a harmonized dataset of Australian rivers, is a substantial achievement and represents a commendable effort. The manuscript is well-structured, the methodology for data collection and quality control appears robust, and the potential utility of this dataset is immense. It will undoubtedly serve as a critical benchmark and foundational resource for the modeling community, ecologists, and climate scientists. The authors have done an excellent job in making this dataset accessible, interoperable, and reusable) and the provision of detailed metadata and a data descriptor paper is highly appreciated. I therefore recommend a minor revision before it can be accepted for publication in ESSD.

Line-to-line comments:

Figure 1 and 4: since this manuscript mainly talk about riverine carbon, can you put major rivers in the maps.

Also can you pub climate zone here or in the supplementary information.

You also need to plot figures for POC observations as well. Can you put the information about POC in the SI?

For the dataset, the ID is very strange. You have something like “AC, AF”, you also have something as river names, numbers (“113006A”). Can you set a rule for your ID.

Additionally, it can be very easy for you to get the catchment information for each site. Can you link your data to global hydrograph dataset or national hydrograph datasets, such as Hydrosheds?