I don’t see what is new here compared to another paper just published by this group, Bilbao et al. (2024).  Is it just by adding one forcing data set, EVA_H?  And since Bilbao et al. (2024) used multiple climate models, why does this paper use only one, and how was it chosen?  How model-dependent are the results?

Abstracts should not include references.

The abstract has acronyms that are not defined or explained.  What is EVA?  What is EVA_H?  How do they differ?  What is BSC?  What is CMIP6?

There are English errors in the abstract.  Did the native English speaker authors not edit the paper?

“Despite these differences, comparing the predicted anomalies in those variables with observations, we show that either of the forcings considered allows to make skillful predictions after the major volcanic eruptions.”  should be “Despite these differences, comparing the predicted anomalies in those variables with observations, we show that either the EVA or EVA_H forcing would allow skillful predictions to be made after major volcanic eruptions.”

I think the last two sentences (after English correction) of the abstract are fundamentally wrong: “Despite these differences, comparing the predicted anomalies in those variables with observations, we show that either the EVA or EVA_H forcing would allow skillful predictions to be made after major volcanic eruptions.  Our study thus supports both EVA and EVA_H generated forcings as reasonable choices for predicting the post-volcanic radiative responses.”

How do you define “skillful?”  How do you define “reasonable?”  The problems, as discussed below, are that these forcing data sets cannot be created just after an eruption.  The location of the eruption and the satellite-observed SO₂ emissions are not enough.  Also, the EVA algorithm is too diffusive, and the paper shows that they do not produce the correct latitudinal distribution for the three eruptions studied.  And there is no consideration of ENSO and interactions with volcanic eruptions.

And why would you want to use either of these forcings after a large volcanic eruption, when models now exist to quickly simulate the conversion of the observed SO₂ injections into sulfate aerosols, and the transport and radiative forcing from those aerosols?  And these can be updated every month incorporating new observations of how the climate and volcanic aerosol cloud are evolving.

The use of EVA is very problematic, because how is it possible to know the latitudinal extent of the stratospheric cloud a priori?  The 1982 El Chichón and 1991 Pinatubo eruptions were only 2.2° different in latitude, but their clouds ended up centered 15° apart, due to the winds on the day of the eruption.  How can the hemispheric asymmetry be known ahead of time?

In fact, Fig. 2 shows that because of the low latitudinal resolution of the EVAs, the observed hemispheric asymmetry of both 1963 Agung and 1982 El Chichón is not reproduced correctly.  This is an additional problem.

Some minor issues:

SO₂ needs to be spelled with the 2 as a subscript.

Why is the mass of S used in Table S1 and SO₂ in Table S2?  This is confusing.

The table headings in Tables S1 and S2 should not have words broken into two lines.  This is easy to fix.

El Chichón is spelled with an accent mark on the o, but this is not consistent in the text or in the figures.

Why does Fig. 4 not include observations?  Why are the observations relegated to a different section?

There were El Niños after each of the eruptions studied.  This would have a huge impact on the short term seasonal and annual forecasts, especially after El Chichón in 1982, when most of the cooling from the volcanic eruption was offset by the warming from the El Niño.  How was this taken into account?

What do the “syear” labels on each panel in Figs. 4 and 8 mean?

“Notably” and “most notably” and “Note that” are used throughout the paper.  What do these mean?  They should be deleted.  Every sentence should be noted or it should not be in the paper.

Fig. 6:  The parts of the maps that are NOT significant should be hatched, rather than the significant parts.  Don’t cover the results you want the reader to see.

Fig. 6:  The temperature changes in the panel labels need units.  And why do panels f and i only have one digit after the decimal point, while all the other have two?

There should be a figure added with the Northern Hemisphere winter responses after each eruption for the first winter, too, to see if the model can reproduce the observed winter warming over Eurasia.

The Deep-C version 5 dataset is mentioned several times but never explained.  What is it?

Bilbao et al have written a description of experiments run with different volcanic forcing data sets. It is important to be able to quickly generate new volcanic forcings for decadal predictions should a major eruption occur. Therefore, it is great that this work has been done and it is written up in a clear and exhaustive manner. I only have one major comment.

EVA_H has some problems with Agung. For people who might be using EVA_H in the future, can any lessons be learned from how the Agung forcing was produced? See minor comments below. It feels a little like this issue is ignored. If not for Agung and a weak ElChichon peak, EVA_H might be a clear choice over EVA. Can its output be improved? It is clearly capable of hemispherically asymmetrical forcings.

Technical comments:

L28 aerosols are transported sediment to the troposphere - I don't understand this

L108-110 Why use a 95% range for maps, but a 90% range for time series? Not that it matters, as long as you say what you are doing, but it might be better to be consistent.

L112 If the hindcast drift is state dependent then the drift will be different for those forecasts that contain volcanoes compared to those that do not. Is this worth mentioning?

L120 Is that not the 90% range? (95-5=90)

L136 It would also be interesting to discuss why EVA peaks earlier and has a higher peak than EVA_H consistently over the three eruptions.

L157 Repetition of line 143.

L164 Specify that you are talking specifically about Agung here

L175 Why did EVA and EVA_H produce such different latitudinal structures for Agung? Could something be done to EVA_H input to improve the outcome?

L181 Again what happened with EVA_H? Why does it not do a descending peak altitude for Agung as it does for the other two eruptions?

L313 This is wrong, DCPP-C is a "better prediction", please rephrase as it is not clear what is meant here.

Line 334 "absence of […] preferential transport to one hemisphere" - this needs to be expanded on. How does this fit with the asymmetrical forcing shown in Figure 2h? EVA_H can clearly do something. This should perhaps be described here or in the introduction.

L343 "weaker as it was calibrated accounting for the Cerro Hudson" I thought it didn't have Cerro Hudson and then this was added in an extra experiment? Please clarify.

Captions: Many of them have "volc - no volc", while this is probably widely understandable, it is perhaps worth writing it out fully for clarity in case the reader is not familiar with the language used in the community.