Dear authors,
thanks you for submitting a revised version of your manuscript. You adequately addressed the issues raised by the reviewers. However, one question remains, which concerns the Monte Carlo simulations used to derive confidence intervals for the timing of seasonal maxima and minima (Section 2.4.4). Ideally, such a simulation would start from an error free seasonal cycle, add pseudo-random noise, and do statistics about the timing of the extrema of the noisy cycle.
How did you do it? I guess you started from the GRACE-based seasonal cycle (which already contains noise) and then added additional pseudo-random noise.
Then the question would be how the actual GRACE noise affects the results. For example, in Fig. 11b I would be concerned that the maximum at February is pronounced due to a little random spike and that this random spike causes the Monte-Carlo-based 50% confidence interval to be so narrow.

Typos:
p. 47 averge --> average
p. 20 line 20 ransect -->transect

I ask you to consider the comments above for the final version of the manuscript.

Regards,
Olaf Eisen

Dear authors,
thank you for revising the manuscript and redoing the calculations to remediate the error and I am pleased to accept your manuscript for publication.
Regards,
Olaf Eisen