Dear authors,
your paper is now nearly ready for publication. I think you have adequately addressed the reviewers' comments. I have just a couple of technical suggestions (see below).
Best regards,
Thomas von Clarmann

l 133: I think the sentence "This manuscrit contains a number of sections" does not contain any specific useful information. I suggest to start the paragraph with "In the Sections 2 and 3 of this paper, the full details
of the retrieval schemes and ... of ACE and MIPAS, respectively, are discussed."
l150: It would be advantageous to give a reference for the 3km altitude resolution of ACE-FTS; it is not always clear how the ACE altitude resolution is defined but with a quotation of a paper where this issue is discussed you can easily avoid to include a lengthy discussion of this in your paper.
l175: This sounds as if the altitude grid of the retrieval is the native grid of the tangent heights, and the interpolation to the 1km grid is kind of post-processing, correct? If so, it might be clearer to start in l 174 with "During the retrieval the state vector is sampled on an altitude grid co-inciding with the tangent altitudes of the measurement. The retrieved VMRs are then interpolated..."
l223 ff: I think it is correct what you have done here but the reader might miss how S_a changes during the sequential retrieval. I suggest: "For this work, the initial a priori estimate is taken from ... 2007); after each step of the sequential estimation x and S_a are updated according to the results of the preceding step." (Correct?)
l226: Is the MIPAS retrieval also performed on the tangent altitude grid? If so, please mention this here.
l236: I assume this setting of the diagonal elements refers only to the initial Sa, which will be updated during the sequential estimaltion. Correct? If so, then please add "the initial Sa diagonal elements".
l402: if you want to do me a favour, you could add a reference to my recent AMTD paper on this issue ("Smoothing error pitfalls"; current status: AMT in press)