Dear Nanna,

Thank you for the revised version of the paper entitled "Response of the Large-Scale Subglacial Drainage System of North East Greenland to Surface Elevation Changes" By Karlsson and Dahl-Jensen. I have received your response to the referees and the revised manuscript. Both referees find the paper good to excellent and subject to minor revisions, which you took into account in the revised manuscript. I still found that some clarification is needed, especially with respect to the spinup phase of the model, in demand of referee 1. I list the remarks below:

"Firstly, if we initialise the model from present day topography using the inverted sliding parameter, the model will need a relaxation time in order for the numerical scheme to reflect the surface topography." I don't understand this, because inversion will lead to a surface topography that is in equilibrium with the flow field. So in absence of any perturbation, the ice sheet will remain in a steady state that equals the observed topography. If you apply a perturbation, it will be a perturbation starting from that state. No relaxation times involved. However, your method uses an approximate inversion, followed by an interpolation on a different grid. That is not the same and should be made clear. Then it is understandable that a relaxation of the ice sheet will follow.

AND

"In order to achieve this, we wish to start with an ice-sheet configuration that is not influenced by basal sliding." See previous remark. You should be more specific in addressing the problem. First outline what the benefits and drawbacks of an initialization by inversion is. It does not mean that there is no influence of basal sliding. It gives you a spatial distribution of basal sliding in accord with the observed topography, which is not the same thing as your approach. However, I understand that you would like to make things as simple as possible using approximate values of basal friction. You have to clearly point out why that is important, even though your initial state is not necessarily corresponding to the observed state. From there it should automatically become clear that the interpolation of the $k_s$ values can be done, because in principle, any inversion is only valid for the given spatial resolution on which you wish to work.

If you make these clarifications, then I may accept your manuscript for publication in The Crysophere.

Thank you very much for submitting to TC.

Kind regards,

Frank Pattyn
Editor TC.