The model-data discrepancy with respect to the climate response to volcanic eruptions is a long-standing problem that has been documented in IPCC AR5. Uncertainties can arise from both the model and data sides, and numerical experiments are a great tool to investigate the impact of volcanic forcing on the climate from aspects such as magnitude and hemispheric symmetry of the forcing, eruption location, eruption season, etc.

This study by Timmreck et al. takes the case of the unidentified 1809 eruption, a large eruption nearly missing in historic documents but its existence is apparent according to ice core sulfur records. The authors conducted numerical experiments with the MPI-ESM1.2-LR climate model, investigating the climate response to different volcanic forcing patterns regarding orders of magnitude and spatial structure (hemispherically symmetric/asymmetric forcing). In addition, they surveyed several instrumental observations and reconstructions based on different proxy types, and performed detailed model-data intercomparisons over the tropics and the Northern Hemisphere extratropics, based on which, they gave an estimate of the VSSI level for the 1809 eruption, and confirmed the importance of the spatial structure of the forcing. Therefore, this study has a topic that is both timely and of importance, and is so far a unique modeling study focusing on the 1809 eruption, which contributes to the understanding of the impact of volcanic forcing on the climate response. The model-data intercomparisons are also valuable for the paleoclimate reconstruction community.

Overall, I think the manuscript is in high quality: the text is well organized with a clear structure; the analyses are to the point and presented with instructive figures. That said, some cleaning up and clarification needs to be done, which I listed below. Once those have been addressed, I recommend the work be accepted for publication.

Details

• Section 2.1.1: It might be worth introducing the parameterization scheme for the aerosol microphysical processes in MPI-ESM1.2-LR. As suggested in recent studies (e.g. LeGrande et al., 2016, Nat. Geosci.), some CMIP5-era climate models can produce overly strong volcanic cooling due to unrealistic aerosol microphysics. How is the scheme in MPI-ESM1.2-LR different?

• Eruption timing: In L107 it is mentioned that the 1809 eruption is set to occur on Jan 1st of 1809. Recent studies (e.g., Predybaylo et al., 2020, Commun. Earth & Environ) have suggested that the eruption timing may also affect the climate response, especially the ENSO response, due to different circulation conditions and ENSO phases. Since the SOI response is also assessed in Fig. 6, it might make this study more complete to add one extra experiment testing the sensitivity to the eruption timing.

• Fig. 2: It might be better to use green triangles or other readable colors and symbols to denote the location of the tree-ring proxies. Red can be misleading given that red also represents a high temperature anomaly. The colorbar may also be adjusted to drop the white color to differentiate the missing values.

• Fig. 5 & 6d: It might be worth showing the relative sea surface temperature (RSST) (Khodri et al., 2017, Nat. Commun.) to highlight the impact of volcanic forcing on ENSO relative to tropical mean cooling in the supplementary information. It may or may not affect the results significantly, but either way it is a valuable assessment.

• Fig. 6: If I am not misunderstanding, the volcanic forcing magnitude in Best, Low, and Hight experiments can be ranked as High > Best > Low according to Fig. 1, and there's no difference in their meridional structure. I am curious about the reason that the SOI response in the Low experiment seems to lie between that of the Best and Hight experiments during both winter and summer. Particularly, does the fact that the High and Best experiments show opposite signs during the summer indicate that the SOI response is actually more internally driven than externally forced? Similar doubts exist for other indices that the impact seems not following the same monotonic order of the magnitude of the forcing. It might be good to reorder the bars in the figures as pre, Low, Best, High to highlight the potential impact of magnitude, no matter exists or not.

• Fig. 7a and L200-204: In L200-204, it is mentioned that the authors accounted for the sparsity and irregularity in spatial and temporal sampling of the EEIC data, but it is unclear how good the performance of the processing is, and the authors still see overall dampened tropical SST anomalies in EEIC compared to model simulations in Fig. 7a. I was wondering what the comparison would look like if compare the mean of the model simulated SST anomalies over grid cells nearest to the locales of the EEIC logs to the mean of the original EEIC data. Similar strategy might be worth taking for other comparisons if the observations/reconstructions are available over multiple sites instead of a processed regional mean.

• Fig. 9: It seems that the strategy mentioned above is taken here in Fig. 9 as the model simulations are "similarly sampled". Perhaps can add an extra column for the visualization similar to Fig. 7a, comparing EEIC to ensemble means, but for two separated regions. Is it overall a better agreement in Fig. 9 than in Fig. 7a? If so, does it mean the sparsity of the EEIC logs is not well accounted for as mentioned in L200-204?

• Fig. 13 & 14: What is the rationale that the anomalies are calculated with respect to the years 1806-1820 here instead of 1800-1808 as in previous figures? The decision will largely affect the model-data comparison on the response to volcanic forcing.

• L527: a typesetting issue (delta-18-O)

Dear editor, dear authors, 

the manuscript preprint cp-2021-4 by Timmreck and co-researchers is a valuable contribution to our understanding about the climatic effects of an unidentified volcanic eruption sometime in 1808 or 1809 CE. 

The manuscript reports (i) results from a set of dedicated climate simulations sampling the forcing uncertainty and (ii) the comparison of these simulations with available observed and reconstructed local, regional, and hemispheric data sets. 

I do not have any major concerns with the manuscript, but I have a small number of minor comments. One of these minor notes is tending slightly towards major. 

That is: I do not believe that the manuscript does what the authors claim in the title. While I appreciate a catchy title, I think overselling is a net-minus. My understanding of the title is that the authors claim to show why 1809 remains a cold case. The title plays with the cold-case terminology from criminal investigations and procedural crime series on TV and streaming. However, from my point of view the manuscript does not show enough in that respect to claim this title. A positive reading is, the title already announces the failure to provide major new insights into location, seasonality, or strength of the eruption, a negative reading is that the title suggests large new insights, why we won’t have much success in becoming more certain about this eruption. I don’t think these large new insights are provided. The manuscript is in a sense incremental while also presenting some very valuable - and I think new - simulations and in addition supporting previous understanding about the eruption in 1809.

Minor

Line 19: The claim that the early 19th century is the coldest period of the last 500 years is based on a reference to Jungclaus et al. (2017). While I am not entirely sure whether this is meant to be a global statement or a hemispheric one, there have been a number of recent reconstruction efforts spanning the period of the last 500 years and beyond, which may or may not require to reassess or qualify the statement. That is, such an absolute statement requires assessing the newest evidence, which in this case may include, for example, Büntgen et al. (2020), PAGES2k Consortium (2019), and more.

A note on the paragraph starting in line 78: the authors detail changes in their atmospheric component but they only mention in passing that there were also changes in their land component. That is fine, but the authors probably also can be briefer in describing the atmospheric model.

Figures: some of the Figures have a strangely low resolution (on my screen). While this will be caught by the technical editing at Copernicus anyway, I wanted to mention it. 

Data: EEIC. I have to admit I am not really up to date and may express my ignorance but how does the EEIC data differ or improve on the most recent ICOADS data? 

Data: Stations. I was thinking there should exist tropical stations for the period of interest. Do they have too little temporal coverage, or were they not of interest?

At line 284 I was wondering whether a very short comparison between reconstructed and observed results would be of value there. 

Index comparisons: Line 347ff. I fail to see the relevance of the circulation results for the manuscript. They feel unrelated to the rest of the manuscript. They are again referred to in the discussions but to me it also remains unclear there why they are relevant for the argument of the manuscript. The point in the discussion does not depend on the analysis, does it? This is not necessarily a problem but skipping the relevant parts may make the manuscript more concise. 

Discussions of central Asia: The authors give good reasons for potentially weak reconstructions in Asia and particularly central Asia. Has there been a general evaluation of how well the MPI-ESM performs in these regions? 

TR: I invite the authors to skip the abbreviation TR. “Tree ring” is not too long to be spelled out everywhere. 

NA: similarly to TR, is it really necessary to abbreviate North America? 

There are a small number of grammar/spelling/typos etc that I assume are artefacts from tracking changes in a document. I only mention Line 655: proxi -> proxy 

Is a “comparative assessment” a comparison? 

I wonder whether it makes sense to spell out S in line 638.

Note

Finally, I have to mention that I cooperated with some of the authors in the past.