We thank M. Frezzotti for the interesting comment and for starting the scientific discussion of our paper. We hereby address the comment point by point:



1) We agree that a thorough study including measurements like impurities or also densities would be interesting. We already showed differences and similarities in the stratigraphic noise of water isotopes (Münch et al., 2016, 2017) as well as densities (Laepple et al., 2016). However, this specific study is a quantification of stratigraphic noise in water isotope records across space which is relevant for using ice cores as a climate archive. Also, this study is not about diffusion itself. In our view, a more detailed assessment of diffusion is also not relevant for our conclusions, as we do not expect it to significantly differ between the cores at each site.

2) There might be a misunderstanding - the surface topography is characterized by variations on different spatial scales. For the stratigraphic noise with a decorrelation length of several meters (Münch et al., 2016), we are interested in the local snow height variations e.g., surface roughness, in our case on a scale of < 60 m. The slope inclinations are assessed across scales of 10 km. They are therefore two different quantities.

Furthermore, the surface roughness, as the standard deviation (SD) of the snow heights, is a statistical measure of the variation within a sample and does not give direct information on the amplitude or max/min values, e.g., sastrugi heights. Assuming a sine wave, the amplitude would be 2*√2 = 2.8 times larger than 1x SD. As we are looking at minimum and maximum values within a transect, additional stochastic variations are expected, which can result in an even higher peak to peak amplitude relative to the SD’s. Therefore, the SD’s of < 10 cm do not contradict the measured sastrugi heights. We still consider the SD to be the more robust metric for snow height variations related to stratigraphic noise as estimates of ranges are statistically less stable.

3) As pointed out by M. Frezzotti, the surface roughness varies throughout the year. In this study, the focus is on the relative differences in space. As all measurements were conducted at the same point in time, this still allows intersite comparisons.

However, the question remains whether the measurements at this single season are representative for the surface variations influencing the stratigraphic noise over time. We found that the past surface roughnesses based on isotope isochrones are similar to the surface roughness, even though they originate from different seasons. We therefore assume that the summer surface roughness is still a relevant site characteristic and could be a predictor for the overall stratigraphic noise.

We also fully agree that there are strong variations in accumulation at small spatial scales. In this study, we do not neglect, but on the contrary, analyse these spatial and temporal variations by assessing stratigraphic noise = the noise caused by local variations in accumulation and wind-drift on a scale of < 60 m.

4) We agree that the assignment of isochrones in these highly variable records is affected by some degree of uncertainty - such that it was only done at two sites. At these two sites, as correctly stated by M. Frezzotti, the isotopic values representing an isochrone do not match in their value. Some of them differ in several permil - with a max. overall difference of 6.4 permil and a mean SD of 1.6 permil. These differences are in a range which we expect within at least the same season (e.g., summer or winter). Also and maybe more importantly, we do not expect the isochrones to have the same values in the snow cores, due to isotopic diffusion. The thickness of a specific layer will affect the amplitude reduction and thus lead to variations in the isotopic composition even inside isochrones.

We will carefully consider the comment of M. Frezzotti when integrating the reviews. We will e.g., try to avoid misunderstandings by rechecking the description of the data and analysis.

Literature:

Laepple, T., M. Hörhold, T. Münch, J. Freitag, A. Wegner, and S. Kipfstuhl.: Layering of surface snow and firn at Kohnen Station, Antarctica: Noise or seasonal signal?, J. Geophys. Res. Earth Surf., 121, doi:10.1002/2016JF003919, 2016

Münch, T., Kipfstuhl, S., Freitag, J., Meyer, H., and Laepple, T.: Regional climate signal vs. lo- cal noise: A two-dimensional view of water iso- topes in Antarctic firn at Kohnen Station, Dron- ning Maud Land. Climate of the Past, 12(7):1565– 1581, 2016

Münch, T., Kipfstuhl, S., Freitag, J., Meyer, H., and Laepple, T.: Constraints on post-depositional isotope modifications in East Antarctic firn from analysing temporal changes of isotope profiles, The Cryosphere, 11, 2175–2188, https://doi.org/10.5194/tc-11-2175-2017, 2017.

We thank the anonymous referee for giving valuable feedback and very constructive suggestions to improve the quality of our work.

Referee summary

This paper analyzed the stratigraphic noise across a 120 km transect on East Antarctic Plateau. At seven sites they extracted five 1 m snow cores with a 10 m interprofile spacing and measured the oxygen isotope records. The measured values were used to estimate signal to noise ratios as a measure of stratigraphic noise. The goal of this paper is to provide sampling strategies guide for high resolution isotope records.

This paper is well written, and the subject is appropriate for Cryosphere. However, I have still some minor comments which should be addressed before publication.

1. General comment
   
   It would be interesting to compare the isotopes sampling with the snow properties (e.g. density or grain size) to see whether there is a connection to the stratigraphic noise. Did you also measure snow properties (like density) for each side and snow core?
   
   
   
   We agree that it would be interesting to compare snow densities with the isotope records, e.g., at the sites where we observed glazed surfaces. Unfortunately, we do not have any density measurements of the snow cores. Such a comparison was, however, previously carried out by Laepple et al. (2016) at the Kohnen Snow Trenches. In this study, no significant correlations were observed between isotope and density profiles, except for a relationship of the mean profiles related to a seasonal cycle in both parameters.
   
   
   
   
   
   
2. Main Comments
   1. I'm a bit curious about the quality of the provided results because of the low sampling numbers and locations. The author mentions two times (line 191 and 232) that the hypothesis is based on low number of samplings. Could you provide some sentences about how robust your hypothesis/assumption are and what is approx. the minimum sampling numbers and locations to make your hypothesis more robust?
      
      
      
      Thanks for the good suggestion. Currently, we are providing uncertainty ranges for the SNR estimates that take into account the number of samplings. For the potential relationships between SNRs and the environmental properties we state if the p-value of a correlation is below or above 0.05. We also generally recommend taking a higher number of records or longer records to make the SNR estimates more robust. Based on your suggestion we will 1.) include a paragraph on how the uncertainty depends on the number and length of the cores. Within the medium range of correlation values found at the site (near gaussian distribution), we can expect that an increase of the number of snow cores or the length of cores by factor n will reduce the standard error of the correlations by 1/√n.
      
      2.) include a recommendation for the design of future studies in the discussion.
      
      
      
      
   2. Line 69-70: Could you elaborate a bit more why you only took one line profiles perpendicular to the dominant large scale wind direction and not more lines parallel to the large scale wind direction? I would expect that also having profiles parallel to the wind field would help to show that the impact from other wind direction can be excluded.
      
      
      
      We agree with the reviewer, that ideally, taking profiles at different directions relative to the wind would be useful to study the influence of the wind direction.
      
      Here, time constraints in the expedition didn’t allow to systematically study this aspect and we therefore decided to use the same sampling direction as we used in earlier studies on stratigraphic noise (Münch et al., 2016, Laepple et al., 2016, Münch et al., 2017) to allow a comparison. Across the region covered in this study, including the area around Kohnen Station, the main wind direction is similar.
      
      Furthermore, snow dunes form parallel to the main wind direction. Thus, in order to measure the same amount of dunes, a shorter measuring distance is required if the measurements are taken perpendicular to the wind direction compared to if they are taken parallel to the wind direction. We will clarify this in the revised version.
      
      
      
      
   3. Figure 2: Could you provide some information whether you corrected your snow core data? Does the isotope signal correspond to the actual snow depth?
      
      
      
      The depth in the snow-profiles correspond to the actual snow depth. The method of using thin carbon tubes for sampling is well established (e.g. Schaller et al., 2016). In some cases, in the process of pushing the tube into the snow, extracting the snow core, cutting and handling, minor errors can occur, which is while we expect the depth scale uncertainty to be < 2 cm. We have such a description in the method section, which we will elaborate on.
   4. Line 215: Could you elaborate this more? What kind of signal could the D24 isotope profiles represent instead?
      
      
      
      At location D24, the isotope profiles are rather constant through depth (with a low variance) except for two layers (one depleted and one enriched) that appear as strong anomalies at the same depth in all cores. These ‘flat’ isochrones contrast the high surface roughness at D24 (that is strongest across all sites). Further, the profiles are not consistent to the expectation of a continuous preserved climate record that should contain 5-7 seasonal cycles. This anomalous behavior is also reflected at the surface: here, we observe strong spatial variability with a mix of glazed snow fields and patches characterized by high sastrugi - a feature that was (visually) unique at this site compared to the other sites.
      
      We thus speculate that D24 represents a discontinuous record in which the two anomalies / isochrones probably represent a summer and a winter layer from any time within the 5-7 years of accumulation.
      
      The SNR defined in this study is the ratio of the signal shared between the cores and the local variations; therefore, the estimated SNR is high for this site, driven by the two anomalies. However, if our interpretation is right, the record could not be used for seasonally to annually resolved climate reconstructions. We will refine our text  to make sure that our description is understandable.
      
      
      
      
      
      
3. Minor comments
   1. Line 101: Could you provide information about what kind of interpolation you used? Linear interpolation?
      
      
      
      Yes, we used linear interpolation. As the isotope profiles are smooth on the cm scale (due to isotopic diffusion), the details of the interpolation method do not affect our results. The confidence intervals and significance values are also not affected as they are based on bootstrapping (and do not rely on the number of datapoints that is artificially increased by the interpolation). We will clarify this in the revised text.
   2. Figure 3 caption: what is several km?
      
      
      
      Thank you for pointing this out. We will include the distance between locations. 
      
      
      
      
      
      
4. General comment from the authors
   
   
   
   To underline the significance of the results, we will include a discussion section on the relationship between the SNRs and the ability to recover a climate signal at a certain time resolution; an isotope dataset with an increased SNR will result in the recovery of a climate signal of a higher resolution, which we expect can be achieved by using our sampling recommendations.

Literature

Laepple, T., M. Hörhold, T. Münch, J. Freitag, A. Wegner, and S. Kipfstuhl (2016). Layering of surface snow and firn at Kohnen Station, Antarctica: Noise or seasonal signal?. J. Geophys. Res. Earth Surf., 121, doi:10.1002/2016JF003919.

Münch, T., Kipfstuhl, S., Freitag, J., Meyer, H., & Laepple, T. (2016). Regional climate signal vs. local noise: a two-dimensional view of water isotopes in Antarctic firn at Kohnen Station, Dronning Maud Land. Climate of the Past, 12(7), 1565-1581.

Münch, T., Kipfstuhl, S., Freitag, J., Meyer, H., & Laepple, T. (2017). Constraints on post-depositional isotope modifications in East Antarctic firn from analysing temporal changes of isotope profiles. The Cryosphere, 11(5), 2175-2188.

Schaller, C. F., Freitag, J., Kipfstuhl, S., Laepple, T., Steen-Larsen, H. C., & Eisen, O. (2016). A representative density profile of the North Greenland snowpack. The Cryosphere, 10(5), 1991–2002. https://doi.org/10.5194/tc-10-1991-2016