Aubry et al. present an updated dataset for stratospheric and upper tropospheric volcanic eruptions for CMIP7 simulations. Using a combination of satellite observations, ice core sulfate from Greenland and Antarctica, and geological evidence from the Global Volcanism Program, they have created a dataset of SO2 eruption mass, injection height, date, lat/lon, and vent altitude from past large volcanic eruptions that seems as close to comprehensive as possible with the available data. Despite their best efforts, small- to moderate-magnitude eruptions are likely underrepresented before the satellite era, which will result in biases in RF estimates in CMIP7 models. I’m not sure that this could be improved upon without implementing assumed volcanic eruptions based on the satellite record.

Overall, I was impressed by the authors’ meticulous attention to detail in creating this dataset. I have only minor comments, suggestions, and questions for the authors below.

Minor comments:

• Line 381: The eruption latitude vs. polar asymmetry relationship derived from ice core data is slightly different than the modeled eruption latitude vs. polar asymmetry equation. The authors explain this as caused by the overly strong stratospheric transport toward the NH in UM-UKCA. This (or in the conclusions) could be a good opportunity to call for future research, e.g. a modeled latitude vs. polar asymmetry equation using a different model. An uncertainty of ~40 ˚ latitude seems important. Could the authors comment on how this contributes to forcing uncertainty from this dataset?
• Line 423: Assuming an eruption date of January 1 may not be the best choice of arbitrary date. For example, I believe Gao et al. (2008) assumed April 1 to choose the most “neutral” seasonality possible.
• Line 450: say what the “Expected values” of 0.25-0.33 are based on (theoretical modeling? I am not familiar with these papers)
• Line 482 “We instead choose a SO2 mass enabling to match…” this phrasing doesn’t make sense
• Figure 7 is nice. I wonder what is the “climate impact” of the missing small-to-moderate eruptions from before the satellite era (figure 7c)? For example, if you estimate anthropogenic aerosol ERF by comparing a piClim-control and piClim-aer (Collins et al., 2017) with and without small-to-moderate eruptions, how different is the ERF? If this has not yet been estimated, should this be suggested as future research?
• Line 600: requesting that modelers document these parameters is an important point. Could you also state this point in the abstract, summary, and/or maybe note it in a dataset read me file? Consider also adding the suggestion from line 802 into the summary
• Line 686: should say “variability” (currently “variabilti”)
• Summary: this was a helpful section, but should be split into multiple paragraphs to make it more readable.

The paper by Aubrey et al., provides a welcome description of the volcanic sulfur injections used to underpin the CMIP7 Historical volcanic forcing datasets. It is comprehensive and clear for the most part, although there are some places that could use some further explanation which I have highlighted below by line or figure. Overall I recommend acceptance once the below issues are addressed.

Figure 2: It would be useful to specify which of the triangles are identified in which cores. Could you split up by NH, Tr, SH – so three lines of triangles that can be linked to the different hemisphere records?

Line 138 : Where does this equation come from?

Line 140: What is the error on height estimates roughly?

Line 149: Are volcanic emissions currently included in the tropospheric forcing? It should be possible to find this out so it would be good to specify if there are instances of double counting.

Line 176: Unclear what this means – what is “estimated at -50%” refering to?

Line 193-194: Is there a bias then in the period after 1900 that is based purely on Sigl 2015 and doesn’t have GISP2? Please clarify if this is an issue or not.

Line 197: is there any consideration for delay between eruption and deposition?

Line 207: Unclear here – what is the instead refering to? What was done in other cases or before?

Line 217: How robust is the assumption of constant accumulation rate at this site between seasons? How much does seasonal precipitation vary seasonally here?

Line 224: what about unattributed eruptions? What is the difference? Is the scaling different?

Lines 227-231: This is unclear what is meant. It needs to be clearer what was published in Feng paper, and what is different here. And what entails a re-assessment. And when the eruption data was a historical date v an ice core date, and the implications of that on the timing (e.g. see question about deposition timing above). These few sentences were very difficult to follow.

Line 250: I think it would be clearer if you instead went backwards in time and found the first point outside the fit.

Line 257: Maybe include a table of return rates for various VEI and how it compares to previous estimates (and include Brown et al. 2014; Rougier et al., 2016 as well as the Papale papers?).

Figure 3: It seems like the method to determine when underrecording starts is a bit flawed for the higher VEI values. Consider the VEI5 plot – there is a huge change in slope around 1600. The key thing to spot is the slope, not whether a single point falls outside the regression uncertainty. Similar with VEI6 the change in slope looks to occur around 0 BC. There have been multiple papers that address this (some cited above and in the paper) so could you use their techniques instead?

Line 280: Here the difference between eruption and deposition age is explicitly stated. What is used though? And is it assumed to be the same for all eruptions? What is the uncertainty on that offset and how is that taken into consideration?

Line 293: “no other VEI ≥ 5 eruption is a low confidence match” – not sure what is meant by this?

Figure 4: What are the two high confidence eruptions that fall outside the large uncertainty window? Can these be labelled and is there a reason why that could help us to understand when this prediction might not work well?

Line 421-422 – This is confusing how it works if there is a long lag between eruption date and deposition as above. Is that considered at all?

Lines 426- 429: This is confusing because the month is specified here, so why was it assumed a month if month was known?

Line 478: Is mean justified? Normally distributed? What is the range, standard deviation?

Line 527: A discussion of why the petrological estimates are so far off of the ice core estimates is warranted.

Line 537: the numbers seem the wrong way around as tephra volume is high then low whereas sulfur load is low then high

Line 545- 550 : It would be useful to specify how they got their S estimates given the large discrepancy between the totals.