Jungandreas et al. investigate the influence of resolved deep convection on African monsoon rainfall and associated land-atmosphere feedbacks during the mid-Holocene in the ICON-NWP model. They compare simulations with parameterized and with resolved deep convection for a prescribed present day vegetation cover and for a more realistic prescribed mid-Holocene vegetation cover. They find a substantial northward shift of the West African monsoon regardless of the representation of convection for the mid-Holocene vegetation cover. However, the changes are larger in the simulation with parameterized convection, which is consistent with the results of an earlier paper by the authors. They argue that these differences are related to soil moisture feedbacks, which are strongly controlled by runoff. Additional simulations with a constant soil moisture field reduce the differences in the precipitation response to the prescribed mid-Holocene vegetation cover. This is a very interesting finding, pointing out the importance of the adequate representation of land-atmosphere interactions.

The manuscript is well written and clearly structured. It provides important insights for the paleo climate and the climate modeling community. I found no major issues and think it can be published after clarifying the minor comments outlined below.

1.) Throughout the text the authors refer to the period as “mid-Holocene” but in the title it says “Holocene” which of course is not wrong, but maybe a bit inconsistent.

2.) The manuscript starts with the term “storm-resolving” simulations in the second line of the abstract, which means resolved deep convection on the kilometer scale. However, the authors refer to it simply as “explicitly resolved convection” throughout the text, before calling it “deep-convection” in the last sentence of the conclusion. The terminology could be clarified in the beginning.

3.) The authors refer to the simulations with present day vegetation cover as “Desert Sahara”. The word “Sahara” is derived from the Arabic word for desert, making the naming “Desert Deserts”. A more applicable nomenclature without changing the abbreviation “DS” could be “Dry Sahara”.

4.) In line 102 the authors note: “We select two years after this 15-year soil moisture- spinup phase and start our nesting experiments for the boreal summer monsoon season.” I would like to know why these two years were chosen and which of them was/were used in the analysis?

5.) I would be interested to know whether the precipitation in the GS and GS-cSM simulations is sufficient to sustain the prescribed vegetation cover.

6.) In Figure 10 a) there seems to be a “separation” of 40 km-P data-points at about 10 % runoff and between 2 and 4 mm/day precipitation. This “separation” is also visible in Fig. 10 b) at about 10 % runoff with a wide range of corresponding soil moisture changes. I would be interested in an explanation for these results. Are these data- points related to a specific time during the JAS season (and thereby maybe also a region)? This is particularly interesting to me because, if these data-points are neglected, the precipitation (mm/day) to runoff (%) relationship appears to be similar for the 40 km-P and 5 km-E simulations in the overlapping range of 2.5 to 4.5 mm/day, i.e. all data-points roughly follow a linear trend.

7.) In some Figures the dpi seems to be too low (e.g. Fig. 3 or 5).

--------------------------------------------------------------------------------------------------------------------------------

Formalities and Typos:

• Figure 1 and 8: Change “sahel” and “sahara” to “Sahel” and “Sahara”.

• Figure 11 and 12: The meaning of the black dashed lines is not described in the caption.

• Line 34: Change “feed back” to “feedback”.

• Line 246: Change “becomes” to “becoming”.

• Line 282: Change “daily-change” to “daily change”.

Review of the paper “How does the explicit treatment of convection alter the precipitation-soil hydrology interaction in the Holocene African humid period?” by Jungandreas et al., submitted for publication in Climate of the Past.

General comments

The paper discusses the simulation of precipitation during the Green Sahara period performed by using a storm-resolving set up, highlighting the effect of the interaction between precipitation and soil moisture. The paper is well written, in a clear and concise manner. Relevant literature is cited, the methodology is suitable for the purpose of the study and clearly presented, conclusions follow from evidence. Results are original and relevant for the field, and I recommend the paper for publication. However, a few minor issues should be addresses first.

My main comment is about the presentation of the implications of the results, which are very relevant for the modelling of the GS. The main finding of the paper highlights the importance of considering the precipitation-soil moisture interaction when high resolution simulations are used. I have the impression that this is not sufficiently highlighted in the abstract and in the introduction, when the objective of the paper is stated.

For instance, in the abstract the authors could add a few lines to briefly explain how run-off influences soil moisture instead of referring to a previously published paper. And make a more logical link with the following sentence.

Similarly, in the Introduction, after the description of previous results by Jungandreas et al. (2021), a clear statement is needed of what is missing in the previous paper and what the present paper aims at.

Specific comments

L14: I don’t understand how to connect the conclusions on the role of precipitation type in this sentence with the conclusions on the role of soil moisture in previous sentences. Please make a clearer logical link.

L82-87: I believe that you should not summarise here your findings. You could outline here the content of the next sections.

L110: why do you use 5 km for simulating explicit convection? Usually in CPM/storm resolving setups, 2-3 km resolution is used. Running the simulations at higher resolution would change your results?

L130: this sentence is unclear to me. Do you simulate land cover by using IFS or do you use the same land cover used in IFS (as I guess)? Please clarify.

L132-135: I don’t understand how you use MPI-ESM Holocene simulations, desert fraction and the present-day vegetation to build your GS land cover. Please rephrase and clarify.

L144: which version of ERA? ERA5?

L182: in the discussion of the changes in the heat fluxes at the surface, you mention the changes in cloud cover, which are not presented anywhere in the paper. A figure is needed to illustrate cloud cover changes.

L187: moisture is not limited with regard to what? Do you mean “abundant”? Please clarify.

Section 3.1.2: a map showing the differences in the lon-lat projection would be helpful in illustrating the changes in horizontal transports.

L243: please add brief descriptions of CAPE and CIN.

L246: LFC is not defined.

Section 3.1.4: this section is not very clear in my opinion. At L252, you state that the increase of vegetation in the coastal region leads to a decrease of precipitation, which I find rather counterintuitive. I’d say that the decrease in precipitation is mostly lead by the change in the thermodynamical gradients due to the dramatic increase in vegetation in the Sahara-Sahel, in turn leading the changes in the dynamics, which you also highlight at the end of the section. Please rephrase to clarify these aspects.

Technical corrections

L75-78: the sentence does not read very well, please revise the location of commas.

L150: We.

L335: please check this sentence, something is wrong/missing.

Summary

Jungandreas et al. report convection-permitting simulations of the mid-Holocene for North Africa. They show that without parameterised convection (-E) there is a weaker feedback between the land and the atmosphere compared to the parameterised runs because more of the rainfall falls during intense events and ends up as runoff so that less is thus available to re-seed further convection.

The paper is well written and thorough, especially because the authors include a comparison at a common resolution of 10km in addition to the main results presented at 5 and 25km grid-spacing. The simulations are analysed in some detail and the results should be of broad interest. I have a few minor comments given below, but otherwise would recommend publication of this very interesting study.

Main comments:

* At 5 km there is explicit convection but it may be more accurately named convection-permitting rather than explicit convection or storm-resolving.

* One of the major conclusions here is that the intensity of precipitation events is very different between the E and P models. This difference is shown to modify the soil moisture and hence the land-atmosphere feedbacks. One thing that is missing is any analysis of how this intensity distribution of precipitation events actually differs between the E and P models. I realise that it has already been shown in the previous work but it might help to include this here also.

* Related to this, many convection-permitting models overestimate the intensity of rainfall events (e.g. Kendon et al (2021, doi: /10.1098/rsta.2019.0547) because they do not resolve all convection at this resolution. The land-atmosphere feedbacks are dependent on this intensity, so can you comment on this potential caveat? Does ICON overestimate these downpours? More speculatively, would more-fully resolving convection (e.g. down to sub-kilometre scale resolution) address this, or could it sbring to light other effects not considered?

Minor points:

Page 7, line 147-156 and point 2: It's not clear what is meant here by the 75th and 85th percentile values? What are these used for? This explanation does not make sense to me.

The fact that the overall precipitation response is not wildly different between the explicit and parametrised convection agrees with studies of future precipitation change in Africa, e.g. Kendon et al (2019, doi: 10.1038/s41467-019-09776-9). It might be worth citing that study here.

Technical corrections:

Appendix: Please can you add the full descriptions of what each figure shows to the captions in the Appendix figures as well as writing "as in figure x".  Otherwise the reader has to switch between the main text and the Appendix to understand what each figure shows.

Page 7, line 150: "Ee" should be "We".

Table 1: This values would be better at 1 decimal place. Adding an anomaly column GS - DS would be helpful.

Line 204 "Africa thus influences" -> "Africa and thus influences".

Line 366: "Also how the land surface (soil moisture and runoff) reacts to specific precipitation characteristics (drizzle or shower) needs" 

-> 

"How the land surface (soil moisture and runoff) reacts to specific precipitation characteristics (drizzle or shower) also needs"

Line 377: "This study highlights the importance to consider..." 

-> 

"This study highlights the importance of considering...", or similar.