|I reevaluated the manuscript submitted by Lewis et al. entitled "Recent Precipitation Decrease Across the Western Greenland Ice Sheet Percolation Zone". The structure of the paper improved and through removal of several parts the reader gets less distracted by numerous presented uncertainty values and parameters. I can support publication after some minor - mostly technical corrections.|
However, scientifically, I still have concerns about Fig. 11 and conclusions derived from it. According to my understanding, a causal relationship in between trends in SMB and average melt rates is not necessarily the case. A strong accumulation year followed by a strong melt season could still result in average SMB values. The trend in SMB would be unaffected but the average melt rate increased. Since a causality between average melt and trends in SMB is not present, a linear regression for the given significance level does not allow interpretations such as the ones being presented. In addition, only 1/3 of the points are within the confidence bounds, while e.g. strong melt and a strong negative trend in SMB can occur (C1) same as low melt and an even stronger negative trend in SMB (C8). Same occurs for average trends. Melt rates for roughly a trend of -6*10^-3 m w.e. a^-2 are within a range of 0 - 0.11 m w.e. per year in melt. I am quite skeptical concerning the statements in L557-564 as well. I cannot see a confirmation of the hypothesis that percolation and refreezing is enhancing the negative accumulation trends. I know, it would be very intuitive but the cores you present do not show this. For a trend "above" -0.006 m w.e. a^-2, 4 cores result in an average melt rate below 0.04 m w.e. per year and 4 cores are situated at or above 0.08 m w.e. melt per year. I recommend to remove Figure 11 and the corresponding text in L557-564. Especially, since the following lines are contradictory to the statement that melt influences trends within the 20a period.
Some technical details:
Please present significant digits and be consistent with it. Several occurrences of statements with different levels of accuracy are placed even within given ranges (e.g. L502). I don't see the necessity to present sub mm accuracies especially while accuracy levels do not allow for such precision.
You should differentiate in between accumulation and accumulation rates. Within the latter part of the manuscript (Section 3), you almost exclusively use the term accumulation for given accumulation rates. Be consistent!
Please carefully correct typos, edits and missing links:
L270-275: 4 sentences in a row start exactly equally
L101 link missing
I recommend to include more often percentage values especially for trends and errors. This facilitates the assessments of errors and trends especially for accumulation rates.
L340 missing unit for Delta-h
L383 I disagree with higher accumulation rates in the SW. I would rather identify higher rates in the central parts of the transects.
L399 ...firn are within error of those... should be ... are within uncertainty ranges of those...?
I don't see the benefit of the plotted annual accumulation rates in Fig. 5. As far as I remember, those rates are not quoted in the manuscript either.
L341 give values for average firn density
Additional note: A recent TCD manuscript presents values for snow densities at Dye-2 for spring 2015 and the following years. In case you want to simplify your core assessment for Dye-2 consider using those.
Heilig, A., Eisen, O., Schneebeli, M., MacFerrin, M., Stevens, C. M., Vandecrux, B., and Steffen, K.: Spatial and temporal variability of snow accumulation for the South-Western Greenland Ice Sheet, The Cryosphere Discuss., https://doi.org/10.5194/tc-2019-184, in review, 2019.