Dear Authors,

After two reviews that somewhat differ in their recommendations, and after the issues raised by reviewer #2 have been responded to by you, I suggest that the manuscript should be published if the two (technical) modifications mentioned below have been made.

1) The modifications in response to the 2nd comment of reviewer #2 (in the revised manuscript version 2, page 6 line 26 to page 7 line 2) seem misplaced, I would suggest moving this to the end of page 7, or to the end of section 2.1. It makes more sense to discuss the computational cost after mentioning the 25 000 iterations.

2) Also I agree with the 2nd reviewer that a caveat such as high computational cost should be clearly and openly discussed. So I suggest including a statement how the number of iterations compares with typical inversions with fixed error structure and parameters (e.g. for the NHB case).