Codilean et al. present an update on the Octopus database, which now includes more data points and several new fields. They present comparisons with other calculators from Cronus and Riversand as well as comparisons among different production rate parameters calculations.

Personally, I am a big fan of the Octopus database and think it is an extremely useful resource. The new fields of glacier area, and quartz-bearing rocks are great additions, holding important information for the interpretation of the data. Moreover, the approach with an open-source code and data base to facilitate interoperability between calculators and reproducibility is important for the field of cosmogenic nuclides, where calculation parameters constantly get updated. Also the figures are of high quality. Therefore, this manuscript is suitable for Geochronology after revision.

My only major comment is that the authors should consider recalculating erosion rates using a time-variant scaling and removing topographic shielding, since they already acknowledge the problems with these approaches in the manuscript. The authors argue that this would cost too much computational time. However, all the basin-average effective atmospheric pressures have already been calculated already, therefore, the time-consuming pixel-based production rate calculation should be necessary anymore.  Aren’t all the necessary parameters for, e.g., Riversand now pre-calculated and the actual denudation rate calculation should be reasonably fast? Below I discuss these points in more detail:

Time-invariant scaling: As has been argued, e.g., by Greg Balco in a blog post

https://cosmognosis.wordpress.com/2020/10/10/version-3-erosion-rate-calculator-benchmarked-finally/

time-invariance can really become a problem for slow erosion rates. The bias arises because the current magnetic field strength is high and was lower in the past, and most calibration data are from the past 20kyr, where field strength was high. I quote from the Balco blog: “Samples with lower erosion rates reflect production during longer-ago periods of weaker magnetic field strength and higher production rates, so an erosion rate computed with time-dependent scaling will be higher than one computed with non-time-dependent scaling. “ Balco shows that this bias can be up to 40% and is therefore quite significant.

As the authors argue, many people download Octopus data for global studies and therefore use a large range of low and high erosion rates in their studies. In such a case, time-invariant production rates is a problem because it introduces a systematic bias. For instance, many studies investigate the non-linear relationship between erosion rate and river steepness (ksn) (Adams et al., 2020). Using time-invariant scaling and having a large range in erosion rates, the Ksn-E relationship would become more non-linear just due to the bias introduced by not accounting for magnetic field variation.

From my perspective, an option would be to switch from CAIRN to RIVERSAND (Stübner et al., 2023), as has been done for calculations in figure 3. I understand that requesting the recalculation of all rates using a time-dependent scaling scheme is a big ask, but I invite the authors to assess whether this is feasible.

If the authors choose to stay with the CAIRN calculation, it would be valuable to show a comparison like in Fig. 3C/D, however, using the Octopus CAIRN-St rates versus the Riversand Lm or LSDn rates. The authors selected high-relief basins for their current approach. However, for a figure comparing the time-invariant and time-variant scaling schemes, the author should select studies that contain a large gradient in erosion rates.

Lithology: More details are warranted for the estimation of quartz percentage in the basins. This is a really useful addition to Octopus. However, the authors do not describe, which lithology classes in GliM are assumed to be quartz bearing. GLiM contains layers such as mixed sediment that can be full of quartz or devoid of it. Please, provide more detail on how this crucial number was estimated.

Topographic shielding: The authors state that topographic shielding likely creates a bias towards too low erosion rates. Given that this bias can be up to 10%, it seems like a good idea to remove shielding from the erosion rate calculations once and for all. The authors argue that this is not feasible given the high computational cost. Is the re-calculation of erosion rates really so computationally expensive? As far as I understand, the computationally-expensive part is the pixel-based averaging performed on DEMs. But that part is already done. Therefore, shouldn’t you be able to recalculate erosion rates fairly quickly without shielding and with time-variant scaling scheme, with the output parameters from CAIRN in a different calculator?

L 54: What is UOW? I don’t see this defined.

L54-48: I’m confused. What is the purpose of CRN Large Basins and DRN Denudation UOW? The article mentions that these include the published denudation rates. If that is the only reason for the existence of these two collections, why aren’t the published rates added as fields to CRN International&Australia? Please, clarify.

L82: Please, state the AMS standard for normalization.

L83: It would be nice to have approximately 2-3 sentences telling the reader about the main characteristics of CAIRN: pixel-based production rate, exponential approximation of production rates, topographic shielding, etc.

Thank you for the opportunity to comment on this awesome data compilation and standardization effort.

Richard Ott

References

Adams, B. A., Whipple, K. X., Forte, A. M., Heimsath, A. M., & Hodges, K. V. (2020). Climate controls on erosion in tectonically active landscapes. Science Advances, 6(42), eaaz3166. https://doi.org/10.1126/sciadv.aaz3166

Stübner, K., Balco, G., & Schmeisser, N. (2023). RIVERSAND: A NEW TOOL FOR EFFICIENT COMPUTATION OF CATCHMENTWIDE EROSION RATES. Radiocarbon, 1–14. https://doi.org/10.1017/RDC.2023.74

First, some general comments on reviewing papers about software and data sets. Basically, this should focus on whether the paper correctly describes what the software/data does/is. It's not really fair for reviewers to demand additional software features or tell the author to process the data in some totally different way. Unfortunately, this means that the most insanely maddening aspect of the OCTOPUS website -- the fact that I can't just click on a sample name displayed on the webpage and get a simple listing of the data associated with that sample -- is not in bounds for this review. Really, I think this is completely nuts -- it is bonkers to have to download a large data set, or hack the Geoserver XML responses, just to look at the data for an individual sample. And it is such a simple piece of functionality to add live links to data pages to the sample name popups that I am utterly mystified as to why this isn't done. However, none of this is allowable in a paper review, and I am obligated to stick to the basic function of evaluating whether the paper correctly describes the data set. 

From this perspective, the paper is basically fine and should be published in approximately its present form. There are a few issues that need some attention.  

1. The relationship of 'OCTOPUS' and the various OSL, palynology, and C-14-related data sets that are briefly touched on in the introduction is unclear. It is clear that the interests of the author lie primarily in the area of erosion-rate applications of cosmogenic-nuclide data, which means that in the context of this paper the OSL and other data sets appear basically an afterthought. In addition, the title of the paper is 'updated cosmogenic-nuclide data' and not 'updated OSL data', but the discussion around line 30 indicates that there are some updates to the OSL data as well. As cosmogenic-nuclides-in-detrital-sediment data, OSL, and paleoecology data are really not very similar, and in many ways the challenges of storing OSL data in an organized way are much greater, my suggestion is for the author to just write papers about these things separately -- cover only the cosmogenic-nuclide data in this paper and then write a different paper in which the other data sets can be discussed in useful detail. 

2. What are 'partner' datasets (around line 23), and why were these two specific data sets selected from the much larger universe of geochronology databases? Why not just include anything with a geodata type feed (ICE-D, Earthchem, USGS Geochron)? Is this, like, an endorsement deal? Did money change hands? 

3. The relationship of 'collections' to data sets is very ambiguous. In one context (line 30-ish), a data type (e.g., OSL or paleoecology) is a 'collection,' whereas later discussion (line 45, 55 areas) then reveals that a single data type is composed of multiple 'collections.' Why is this? This confusing nomenclature is inherited from the earlier versions of OCTOPUS, where it was also confusing, and it would be helpful if it was explained more clearly here. 

4. The discussion around line 68 of why there are some studies that are not included in the database is mystifying, and somewhat concerning. The whole point of developing a centralized online database is that the 'return on investment' (line 69) is very high no matter how few data points there are in a study, because you only have to read the paper and ingest the data once, and then you are done. Also, of course, the computational effort scales with the number of samples, so studies with fewer data are actually less of an investment. Thus, this argument is not at all compelling; in fact, at face value it seems kind of bizarre. Furthermore, this discussion gives the idea that data sets with fewer data are deemed to be of lesser quality, which of course is a terrible approach from the science perspective. From the perspective of using the data for actual Earth science (rather than, e.g., bibliometrics) applications, how the data are grouped into 'studies' is totally arbitrary and irrelevant, so data selection should not be based on this grouping. In my view this section of the paper reveals a significant weakness of the database implementation. As this is written, it's also kind of a weird threat: include more than ten data points in your paper or else it will not be included in the database. As inclusion of data in databases of this sort is a significant element in subsequent data discoverability and reuse, this is a terrible message to send not only from the scientific perspective, but also from the perspective of outreach and student/early career development. Honestly, if I were the funding agency I would squelch this approach immediately. 

5. The discussion about how data sets from cratonic areas are sparse should probably make note of the fact that river systems in these areas are largely depositional rather than erosional systems, in which case cosmogenic-nuclide erosion rate measurements actually don't work. Really what is wanted here is an assessment of how representative these measurements are of the area in which the measurements could be made, not the area in which the measurements could not be made. 

6. The discussion in line 105-ish should be more clear about the fact that the mean atmospheric pressure is not derived from an atmosphere model (as one might reasonably expect) but by inverting the scaling model. That is, one should not assume that the mean atmospheric pressure in this field has any meteorological significance. It sort of says this, but this point should be made clear. 

7. Near line 118, later on this page, and in Figure 3, I don't understand why calculations are being made at the location of the basin outlet. If you are able to calculate the mean elevation, then you obviously know where the basin is, and can also calculate the centroid latitude. In what circumstance would you ever care about the location of the basin outlet, or want to use that in a calculation? 

8. Line 120-123 is slightly misleading. These differences are not because the calculation methods (Balco 2008 vs. CAIRN) are different, they're because the mean basin elevation is different from the effective elevation. This should be clarified. 

9. The section at the bottom of p. 6 (lines 135-140 area) is a little bit incoherent, because it is not clear what each comparison is supposed to test. It would be helpful to rewrite this to make clear what assumption is being tested with each comparison, e.g., something like, 'To quantify the effect of using the mean elevation vs. the effective elevation, we did X and here are the results. To quantify the differences between CAIRN and Balco 2008 with the same input parameters, we did Y and here are the results.' You get the idea. 

10. I agree with the other review that the paper should make clear that the glacier cover fraction and quartz-occurrence-inferred-from-lithology fields are suitable for general guidance, but probably not very quantitative. 

11. The y-axis in Fig 5 is labeled such that it looks like GLIMS is being subtracted from CAIRN, which doesn't make any sense. Also, if we assume that it is really two erosion rates that are being subtracted, then the units should be m/Ma, not percent. This needs to be labeled in a way that conforms with the units (something like 100 * (E_glaciers - E_noglaciers)/E_noglaciers ?).

12. Again following discussion of quartz-fraction estimate in other review, the discussion in line 180-ish is not helpful without some idea of which lithologies are and are not quartz-bearing. Perhaps the easiest way to handle this would be just to have a supplemental table indicating which GLiM classifications are and are not considered to have quartz.

13. In line 190-ish, the discussion here basically says that the topographic shielding calculations are wrong, but we did them anyway, which is rather odd. Again, the computational-resources argument is not super compelling, but on the other hand this paragraph does appear to correctly describe what has been done, so even if what was done is questionable, this section passes the test of whether it is accurately described. 

14. Around line 205, there needs to be more explanation of whether basins with internal drainage issues are (i) not recorded at all in the database (which in my view is bad practice) or (ii) recorded in the database with Be-10 concentrations, etc., but with no associated calculated erosion rate (preferable). Also, frankly, I don't really understand what the problem is here, because Figure 6 makes it clear that you do know what the actual correct drainage basin boundary is in lat/long coordinates - why is it possible to project the black lines into UTM but somehow impossible to project the blue lines? Also, of course, for pixel-based production rate calculations, you don't actually have to project out of lat/long - you can just weight by the actual area of each cell in real units. But that would probably require a redesign of the whole thing. 

This short communication by Codilean and Munack describes an update to the catchment CRN aspects of the OCTOPUS erosion rate database. Overall, this project is a valuable resource for the community in that it organizes an increasingly large number of catchment CRN studies and facilitates the tedious process of recalculating denudation rates that is necessary in compilation papers due to evolving understanding of cosmogenic radionuclide production rates (and evolving topographic data). I myself still find the usability a bit frustrating, but admittedly I haven’t spent much time trying to learn the ins and outs.

The present manuscript is a relatively straightforward description of some recent updates, and should be published subject to some minor clarifications listed below.

Line 17-43: The introduction describes a number of features of the OCTOPUS database that are not relevant to the current paper – (luminescence, charcoal, 10Be and 26Al exposure ages, pollen, etc.). This material should be shortened and the introduction centered on catchment CRN which is the focus of this paper.

Line 68: “limited return on investment”? I would not dismiss these studies from a database just based on the number of samples. Some may be in places with limited data coverage and thus quite valuable. I understand the resource limitations, but since these studies have already been identified I would recommend including on the OCTOPUS website a list of all identified studies, with an indication of their status (included, in process, not in process). This might streamline the possibility of a user to request addition of an important, but small study to the database?

Line 110: A quick note about how atmospheric pressure is actually derived would be helpful (I think CAIRN is starting with elevation, mapping to atmospheric pressure using NCEP2 reanalysis, and then back-calculating and effective pressure from Stone 2000, right?)

Line 119: It’s perhaps confusing to compare between “mean elevation” and the “effective atmospheric pressure”, since the issue is not one of using elevation or pressure, but going through the process to calculate the “effective” pressure (or elevation).

Line 129 and Figure 4: It seems unnecessary to discuss using outlet latitude, since the centroid latitude is already available and clearly the better approximation for production rates. And Figure 4 seems unnecessary as well.

Figure 3: The axis labels are hard to interpret without the caption – It would help readability to add some plain language to the labels (“Calculated erosion rate difference”, “Erosion rate”, etc.)

Line 150-167: There are so many issues with interpreting CRN data from currently glaciated basins (non-uniform erosion rates, time-varying erosion, storage/reworking in moraines, etc.) I would personally avoid doing this calculation at all and just flagging catchments that contain glaciers.

Line 180: what is the definition of “quartz-bearing”?

Line 187-196: I agree that the topographic shielding corrections are going to be minimal for most catchments, but it is nonetheless quite awkward to be baking in an erroneous correction into every denudation rate. It can’t be *that* hard to fix this? Or at least acknowledge that it needs to be fixed in the future?