Summary of Shi et al.

Shi et al. has performed a mid-Holocene (MH) time slice experiment with the new isotope enabled Earth System Model AWI-ESM-wiso. The authors validate the control run and go on to compare the model with isotopic archives. Although the model performs well for the control run and the spatial pattern of the MH anomalies are correct the overall amplitude of the isotope anomalies appear underestimated, which could mean that the MH climate anomalies are underestimated. The authors investigate the spatial and temporal relations of isotopes and climate. One conclusion is that the modern spatial slope can be used as surrogate for the temporal slope in Greenland and Antarctica. Finally, the authors offer an alternative method employing a combination of isotope anomalies and precipitation amount to determine the onset of the West African summer monsoon.

General comments.

I find the manuscript overall well-written, well-structured and with appropriate figures to illustrate the results. As the detailed comments below show I have a number of suggestions and concerns that I think should be taken into account before considering publication. One major concern is the somewhat superficial treatment of the debate on temporal versus spatial slope, which is a topic that has been extensively researched, and I find some key references missing in this context. Temporal and spatial changes in climate are different in fundamental ways, and this is reflected in the difference in the temporal and spatial slope. I think we are past the point where we are asking if the temporal and spatial slopes are interchangeable, which is summarized by the publications listed below (see comment to L304-307 and L319-324).

In summary I find that minor revisions should be made to accommodate the major and minor comments in this review.

Detailed comments.

Abstract

L2 “Straightforward” do you mean “direct”?

L6 Move “well” to right before the comma.

L9-10 “The ratio of the MH-PI difference … is reasonable …” clarify if you are comparing to measured d18O.

L39-43 The isotope-temperature relationship can also be affected by the proportion of continental sources and recycling of vapor (Werner et al. (2001), Sjolte et al., (2014)).

L59 “Straightforward” do you mean “direct”?

L139 Is spring equinox fixed to March 21? This information can be added to Table 1.

L172 “Straightforward” do you mean “direct”?

Section 2.3.5 “Assimilation product”. I suggest a more informative title such as “Marine d18O reanalysis data”.

L214-215 High bias in d18O over Antarctica has also been linked to a low bias in water vapor for cold regions leading to a precipitation weighting towards higher d18O (Masson-Delmotte et al., 2008).

Figure 3/4/5/6. Are the simulated MH-PI  anomalies significant?

Figure 6b Why not plot the RMSE between the different species which you write in the text? I think the reader expects this since it is included in the other plots. It doesn’t take much to explain the details in the figure caption.

L304-307 I think it would be good to move this to the introduction and include some relevant paper discussing this (Cuffey et al., 1992; Sime et al., 2009; Sjolte et al., 2011; Kindler et al., 2014; Sjolte et al., 2014; Guan et al., 2016).

Figure 8: It is risky to calculate a slope between only two points, but one could use the STD of the interannual variability to estimate the uncertainty or use the annual/seasonal data to make bootstrap estimates of the uncertainty in slope.

L323 The estimate is affected by the selection of the data. For example, if coastal data is included or only higher altitude sites on the ice sheet (Sjolte et al., 2011).

L319-324 Vinther et al. (2009) estimated a slope of 0.5 permil/degree C using borehole temperature and ice core data. I think it is well documented from the papers I mention above that the spatial slope and temporal slope are not interchangeable and that there are large regional differences in both. To quote Guan et al. (2016): “Finally, the relation between temporal and spatial slopes is understood using a semiempirical equation that is derived based on both the Rayleigh distillation and a fixed spatial slope. The slope equation quantifies the Boyle's mechanism and suggests that the temporal slope is usually smaller than the spatial slope in the extratropics mainly because of the polar amplification feature in global climate change, such that the response in local temperature at middle and high latitudes is usually greater than that in the total equivalent source temperature.”

L341 What is the choice of 0.5 degree C anomalies based on? Can’t you base it on whether the anomalies are significant?

L363 Judging by Figure 8d there is still quite some regional differences in slope. What is the slope of key ice core locations in Greenland and Antarctica, and what is the reconstructed temperature anomalies from ice core d18O based in this slope? This can then be compared to the simulated temperature anomaly.

L373 Again I wonder how the results would look if one based the maps on significant anomalies instead of a somewhat arbitrary threshold? Did you look at the slope for the NH/SH monsoon season (JJAS/DJFM) instead of annual mean?

Section 6: In relation to this section the definition of the calendar. Maybe it’s because the calendar effect is neglectable given the variance of the onset of the monsoon, but the calendar effect is a constant bias, so I suggest taking it into account. In any case I think it would be useful to discuss the role of the calendar effect in relation to the onset and duration of the monsoon.

L483 From the model setup I understood that the model is run with active vegetation? L141-142: “In our simulations, the dynamic vegetation is interactively calculated via the land surface model JSBACH.” If not, lacking vegetation-albedo feedbacks could explain some of the too weak modelled climate response to MH conditions, if the vegetation is indeed interactive the model might still be underestimating the changes in vegetation.

References

Cuffey, K., R. Alley, P. Grootes, and S. Anandakrishnan (1992), Toward using borehole temperatures to calibrate an isotopic paleothermometer in central Greenland, Global Planet. Change,  98(2-4),  265– 268.

Guan, J.,  Liu, Z.,  Wen, X.,  Brady, E.,  Noone, D.,  Zhu, J., and  Han, J. (2016),  Understanding the temporal slope of the temperature-water isotope relation during the deglaciation using isoCAM3: The slope equation, J. Geophys. Res. Atmos.,  121,  10,342– 10,354, doi:10.1002/2016JD024955.

Kindler, P., Guillevic, M., Baumgartner, M., Schwander, J., Landais, A., and Leuenberger, M.: Temperature reconstruction from 10 to 120 kyr b2k from the NGRIP ice core, Clim. Past, 10, 887–902, https://doi.org/10.5194/cp-10-887-2014, 2014.

Masson-Delmotte, V., et al. (2008), A review of Antarctic surface snow isotopic composition: Observations, atmospheric circulation, and isotopic modeling, J. Clim.,  21(13),  3359– 3387, doi:10.1175/2007JCLI2139.1.

Sime, L., Wolff, E., Oliver, K. et al. Evidence for warmer interglacials in East Antarctic ice cores. Nature462, 342–345 (2009). https://doi.org/10.1038/nature08564

Sjolte, J.,  Hoffmann, G.,  Johnsen, S. J.,  Vinther, B. M.,  Masson-Delmotte, V., and  Sturm, C. (2011), Modeling the water isotopes in Greenland precipitation 1959–2001 with the meso-scale model REMO-iso, J. Geophys. Res.,  116, D18105, doi:10.1029/2010JD015287.

Jesper Sjolte, Georg Hoffmann & Sigfús Jóhann Johnsen (2014) Modelling the response of stable water isotopes in Greenland precipitation to orbital configurations of the previous interglacial, Tellus B: Chemical and Physical Meteorology, 66:1, DOI: 10.3402/tellusb.v66.22872

Vinther, B., Buchardt, S., Clausen, H. et al. Holocene thinning of the Greenland ice sheet. Nature 461, 385–388 (2009). https://doi.org/10.1038/nature08355

Martin Werner, Martin Heimann & Georg Hoffmann (2001) Isotopic composition and origin of polar precipitation in present and glacial climate simulations, Tellus B: Chemical and Physical Meteorology, 53:1,53-71, DOI: 10.3402/tellusb.v53i1.16539

1. Line 80-84, should motivate why studying the monsoon onset is important. Line 495-507 in the discussion part can be moved here.
2. Line 125, "The model has been widely used with its standard configuration", may cause confusion if it refers to model with or without isotope, better present "The AWI-ESM model has been widely used with its standard configuration".
3. This is the first document for implementing the isotope in ocean model, is it possible to evaluate if everything is done correctly with ocean-only simulation without coupling? Otherwise how to identify any biases in coupled version is due to the ocean model implementation or due to coupling?
4. A bit confusing for the PI period, Line 130 says 1850 CE, and Line 177 says "and PI period (1950-1990CE)".
5. Line 226-227, "For instance, our model underestimates the isotopic composition of the South Atlantic and overestimates the δ18O values in the Southern Ocean", these under/over estimation is due to ocean model or due to implemented isotope? Any comments on these biases would be helpful to understand the origin of the biases.
6. Colour bar in Fig 1,2,5,6,8 needs to be improved. For example, the red colour in Fig.1 for -6 to -4 looks the same to me.
7. To be consistent, mark RMSE in Fig6 as that in Fig5.
8. Line 374, for gradient, "-4 ‰mm−1d" should be "-4 ‰mm−1/d".
9. Line 387, dexp is never mentioned earlier, for those who are not familiar with water isotopes, the meaning of d-excess and definition of dexp should be introduced.
10. Fig10, better to indicate "day of a year" for x-axis in the figure, the same for the other similar figs.
11. Fig10 caption, "The two black lines respond to 4.7°N and 10.3°N." There is one black line and one red line. Here the 10.3°N is marked in the figure, but in the text description often mentioned 12.3°N, typo?
12. For Fig S2 to S4, is it proper to use 4.7°N and 10.3°N for the specific year? If you apply EOF analysis for specific year, I think the maxima precipitation will have different latitude location.