I am copying the comments of the reviewer below.
My opinion is that it would be nessecary to try to address them especially P7L29 and figure 5.
------------
My main content concerns with the previous version were (1) establishing the consistency between ATSR-2 and AATSR records such that they can be treated as a single data set for trend analyses, and (2) the split of material between this paper and the companion Part II. In this revision the authors have addressed those concerns, by adding substantial new material comparing ATSR-2 and AATSR retrievals, and by reorganising the later sections which had some overlap with Part II. My other comments have also been addressed. I appreciate their efforts and think that the balance between Parts I and II is better now (and also look forward to reviewing the eventual revised Part II). I therefore recommend publication after technical corrections, for the following small points:

Abstract – I would still like if this could be shorter (one long paragraph) but will leave this up to the Editor as it is more a matter of personal preference.

P7L29: The text should clarify whether ‘pixel by pixel’ here for this part of the analysis refers to level 2 retrieval pixels (which I think are 0.1 degree) or level 3 1 degree grid cells. (I think either is fine for this type of consistency analysis.)

Figure 5: ATSR-2 and AATSR were on the same orbit track, 30 minutes apart. So I would expect the number of matchups with AERONET across the August-December 2002 period to be very similar, since aerosol conditions and cloud cover should in general be similar. But this figure gives 223 matchups for ATSR-2, and more than double that (536) for AATSR. This suggests to me that something else is going on. Is ATSR-2 systematically only providing retrievals about half the time that AATSR is? If so, why is that? ATSR-2 had a narrow-swath mode for the solar channels due to bandwidth limitations related to the scatterometer also on ERS-2, but that was largely operating only over the ocean, so my guess is that this may not be the prime factor, since most AERONET sites are over the continents. Unless it happens that most AERONET sites during this month were near the coast. Or could there be a bug in the code somewhere here? I would appreciate if the authors could check this and add a sentence or two with the explanation for the difference, otherwise other readers may like me be a bit surprised and confused about this as well.

Figure 11: Since the x-axis is labelled by season, I don’t think this figure needs the symbols coloured differently for each season as well. It seems a little redundant/distracting, especially since two of the colours overlap with the lines denoting individual data sets (e.g. blue means ADV, but also means DJF for both ADV and MODIS). I would rather just see ADV in all blue and MODIS in all magenta.

best regards