The authors investigate the influence of weatherability on long-term CO2 trends during the Cretaceous-Eocene period (145-34 Ma). To this end, they use climate model output (temperature and runoff) from 19 GCM simulations from HadCM3L as input for global climate and biogeochemistry model GEOCLIM. They find that palaeogeography strongly controls weatherability (primarily via changes in runoff) and that localised changes in weatherability can have an impact on global CO2 values, making a case for spatially resolved simulations of weathering-driven CO2 changes in Earth's history.

This is an interesting and solid paper which deserves publication after major review. My main concerns relate to a somewhat selective discussion in some places, a lack of depth in the analysis, and a lack of discussion of potential sources of uncertainty in the models.

Comments:

Lines 59-61: It should at least be mentioned that there is almost certainly also a contribution from CO2 drawdown by coal formation contributing to Carboniferous cooling.

Line 85: The statement that "paleogeographies are relatively well constrained" might be true for the time period investigated in this paper but certainly not in general, so please clarify.

Line 131-133: I understand that you focus on terrestrial processes here but saying that a "range of biogeochemical processes are modelled within the ocean layers" is really a bit vague for a scientific paper. Also, are there any feedbacks of these marine processes relevant for the results you are presenting in the manuscript?

Lines 171 & 181-183: Is there any impact of uncertainties in the vegetation model on the results, e.g. by using present-day PFTs?

Line 172: Since runoff is crucial to your results and your arguments, it might be important to give some more details here - and to discuss the implications of uncertainties in modelled runoff on your results!

Lines 191-192: It might be unclear to the reader how you fix the CO2 concentration in the model given a fixed degassing rate but a variable weathering rate? Are you referring to the climate model part relating to weathering?

Lines 198 & 202: This might be confusing because you ran two sets of 19 simulations each. It becomes clear from the context but should be explained in a better way to avoid unnecessary confusion.

Figure 1: Although the symbols are different, I think it is not a good idea to have both the CO2 proxy record and one of the sets of model simulations in red – there are many more colours out there...! ;)

Caption of Figure 1, lines 2-5: In my opinion, figure captions should be for descriptions rather than discussion of results/comparison with literature. This should be moved to the text.

Lines 209-210: This might be slightly confusing because you are discussing the set of simulations with fixed CO2 here. I think this can be explained better in the sense that you could say that you are comparing the two sets of simulation here...

Caption of Figure 2, line 2: Again, this sentence might be better placed in the main text. Also, I am not sure I understand the basis for this assessment from the plotted lines.

Figure 3: It might be useful to have some sort of a grid, at least in latitude, to be able to judge, e.g., where the ITCZ or the subtropical high-pressure belt would be expected etc.

Discussion section: There is no discussion on the influence of model uncertainties on the results, see the examples mentioned above and below. This needs to be added in the revised version.

Lines 241-242: This sentence lacks precision and is therefore confusing. Which "mean"? Which "runoff rates"? Also, the total global runoff is shown in Figure 2, so why are you referring to Figure 4?

Line 255, "climate patterns": This is a bit vague...

Lines 258 and 264-265, "may be": I find this choice of words a bit strange given that you should have the model diagnostics to figure out what is going on in the models...

Lines 279-299 & 375: So you find agreement for (roughly) 145–125 Ma, 90–70 Ma, and 50–45 Ma or 45 Myr or less than half out of the total 111 Myr spanned by your simulations. I find it scientifically somewhat questionable to discuss the agreement during these time periods in detail without discussing the other time periods where the comparison fails. I think that the framing in the Conclusions (lines 411-413) is alright but this is not quite reflected in the Discussion.

Figure 5: I think the vertical axes need better labels explaining that A shows a CO2 difference and B a difference in silicate weathering flux.

Section 4.4: I am really not happy to call a fictitious CO2 trend based on model simulations with strong assumptions and disagreeing with proxies for a significant portion of the time period covered a "CO2 record".

Appendix A: I was rather surprised to discover in the appendix that climate variables were extrapolated towards lower CO2 levels and frankly shocked that this was not mentioned in the Methods section of the manuscript. What are the implications for your results? Have you at least run a test simulation making sure that the extrapolation method produces reasonable results?

Minor comments:

There are a few minor typesetting issues in the manuscript, e.g. missing spaces (e.g. lines 10, 16, 142), unclosed brackets (e.g. in the caption of Figure 3), messed up citation styles (e.g. lines 147, 153) which should be corrected.

Line 141: "Arrhenius"

Line 175: "palaeogeographries"

Line 235: Add "times"?

Line 271: Is "pure" the best word here...?

Somehow the heading "References" appears twice, including a blank page.

The authors here investigate the role of paleogeography in controlling atmospheric CO₂ concentrations over the Cretaceous and early Cenozoic. The authors are at the leading edge of this kind of work, which previously coupled the GEOCLIM biogeochemical model with a low-resolution climate model FOAM. The major advance here is the use of a higher resolution climate model HadCM3L. Beyond the results specific to this study, the method is compelling and should be adopted by others.

I’ll start with two comments:

• Why was van der Meer (2014) used for degassing instead of a newer, more sophisticated record like Muller (2022; https://doi.org/10.1038/s41586-022-04420-x) or Torsvik (2021; https://doi.org/10.1002/9781119528609.ch16)? Similarly, the Cenozoic CO₂ record of Foster (2017) has been usurped by Hönisch (2023; https://doi.org/10.1126/science.adi5177). This revised record is quite different, especially for the early Cenozoic. The problem, though, is how to suture this record to the pre-Cenozoic record of Foster (2017). I think the solution is to present both, with a disconnect at 66 Ma. I think it is important for readers to know that the pre-Cenozoic portion is less certain.

• Paragraph from lines 241-250: This analysis would be more compelling if the associated figure was a cross-plot of runoff vs. silicate weathering (but connect time-adjacent points with lines so that the time evolution can also be evaluated). Statements like “strengthens”, “weakens”, and “low confidence” need statistics to back them up.

Finer-level comments:

Line 12: “switch” shouldn’t be in quotes; the phrase “global weatherability appears to switch regimes. In the Cenozoic,” can be removed without affecting the intent of the sentence.

Line 24: Weathering and burial of organic matter is important too.

Section 1.1: Somewhere in here mention that two additional factors that are held constant are the organic subcycle (burial / weathering of organic carbon) and plant evolution (especially the evolution of angiosperms during the Cretaceous). Plant type is included in HadCM3L, but the direct effect of plant type on chemical weathering in GEOCLIM is not (based on my understanding).

Line 62: Why do you say “Another” here? The previous paragraph (especially line 58) already introduces this topic.

Figure 1: “Pg” (= Paleogene) is not correct; this should be Paleocene.

Figure 3: It’s odd (to me) to see the panels organized from young to old (instead of old to young). We have been culturally conditioned to read from left to right and to tell stories in chronological order. (The same is true for the time axis in Figures 1, 2, etc…).

Lines 234-236: I’m confused by this statement: shouldn’t you be correlating runoff to CO₂ directly?