This paper uses routine aircraft-based temperature and humidity measurements from the AMDAR and WVSS-II programs, respectively, to evaluate the operational thermodynamic retrievals from the IASI instruments aboard Metop-A and Metop-B.  They find that the retrievals from the two satellites have almost identical biases relative to the aircraft obs.  Furthermore, the evaluation of the satellite retrievals by day/night, viewing angle off nadir, and season show very small relative differences. 

I find this paper very interesting, and very relevant for this journal.  It is well written, the figures are clear and informative, and the conclusions generally well supported by the writing.  I believe that this manuscript only needs a small amount of work to make it suitable for publication in EGUsphere.  I have one “major” (really not too major) and several minor comments.

Major comment:

Does the IASI team independently retrieve RH and SH (i.e., retrieve the two variables separately)?  It seems that they must, because otherwise the mean biases don’t make sense to me.  I have two examples that I’ll use to illustrate my confusion:

• Fig 6: at 205 K, the bias in SH is about 0 (or slightly negative) and the bias in T is positive.  For a given SH, a positive error in T would result in a decrease in RH, and similarly, decreasing SH should result in a decrease in RH.  But for the temperature bin, the RH bias is +3 RH%.  I know we are looking at means here, but the tails of the distribution (for that T) are even worse.  This is inconsistent
• In Fig 2 at 1000 mb, the bias in T is zero and the bias in q is -0.8 – which seems to lead to a bias in RH of about 2 %RH. I did some back-of-the-envelope calculations, and for a range of T (from 10 to 35 C), and starting from two points (roughly RH=80% and RH=40%) and assuming that the aircraft data were about 0.8 g/kg drier than IASI (to give the bias shown in Fig 2), the resulting difference in RH ranges from -10 %rh to -2% RH.  Since there is a year of observations from 2017 in this figure, I think that the mean RH bias should be much later than the 2% shown in Fig 2 if the q and RH retrievals from IASI are consistent with each other (I think the mean RH bias should be closer to -5 %rh).  (see the attached figure on the next page)

Anyway, I think the authors need to be more explicit about the consistency between the IASI retrieval of q vs RH (or if one is derived from the other), and if they are independently retrieved, to spend more time discussing the implications of this.

Minor comments:

• Line 41: twice-daily revisits by a single satellite – this should be clarified
• Line 93: “first constituted with of the order of 10^8 of real IASI” – this is very awkward, and should be rewritten
• Line 149: a laser diode hygrometer
• Line 177: there is also significant displacement (and perhaps even more) for descents. The way this was written suggests that only ascents (not descents) were analyzed here, which I don’t think was true.  Please update
• Line 273: does the shape of the vertical profile matter to the IASI retrieval? For example, in the daytime, the water vapor specific humidity is often pretty constant in the convective boundary layer, but that isn’t true at night.  This should at least be added to the text as a possible explanation for the small day/night differences in the bias
• Caption of Fig 1 and Fig 5: you state “2019” when I am pretty sure you mean 2017
• Caption of Fig 2: you state “AMDAR-minus-IASI” when all of the other results are the reverse. I believe this is a typo, based upon the results shown later

I thought this paper was very well written and an excellent research topic. The paper is clear, consise and well thought out. This paper points out the continued need for aircraft based observations even with thie abundence of satellite observations which is an important concept for the global weather community to recognize. 

A few minor comments:

1. Line 224 spelling issue at the start of the sentence "Iit"

2. Figure 1, 5 both say 2019but paper states 2017

3.Line 149. WVSS-II sensor, a tunable diode laser absorption spectrometer.

Wondering if any of the biases with WVSS-II could be off due to the accuracy range of the sensor which is  +/- 50ppmv or +/- 5% of reading, whichever is greater.

Dear authors,

please see in the attachment the review I prepared some weeks ago. I am not sure, if this has been already sent to you via Copernicus. My apologies, if in the meantime you have already addressed some of my comments.

Best regards,

Matthias Schneider