The manuscript by Gacitúa et al. describes a novel ship-based intercomparison of sea surface temperature measurements using infrared and microwave radiometers.  This is, to my knowledge, the first evaluation of such a deployment and the work has implications for future validation of satellite SST measurements and intercomparison of skin and subskin measurements.  As such, I believe the manuscript can be an important contribution to the literature after revisions that, in particular, help to place the use of the microwave radiometer in better context of all the potential validation approaches and more directly speak to the challenges associated with variations in the incidence angle.  My suggested revisions are largely related to presentation and context, but could involve some additional computations associated with incidence angle variations.

The conclusions speak toward understanding the intricate relationships between TIR and PMW measurements.  This is indeed an important end goal and justification for the type of work included in the study.  To help get to that point, I think there could be more discussion up front and throughout on the overall context and approach to such studies.  How will passive microwave radiometers be used and fit into an overall validation and research strategy?  Given the measurement uncertainties involved (that are highlighted in this study), how suitable is a microwave radiometer for ultimate validation of subskin temperatures and differences from the skin temperature measured by an infrared radiometer?  While a skin temperature measurement cannot practically be obtained by other than radiometric approaches, the same is not necessarily true for the subskin temperature.  Line 437 speaks to the potential importance of a coincident measurement of the subskin temperature - this is indeed a critical point and I think this should be addressed up front.  The manuscript could touch on, in at least a cursory manner, the relative tradeoffs between different measurements – a physical measurement is potentially more accurate but it can be challenging to get at the precisely desired depth without disturbing the measurement.  When and for what purpose would you potentially want to use the different measurements?

Moreover, I think somehow the paper could touch a bit more on the differences between validating the general form of the retrieval algorithm, vs specific measurements from the deployed radiometers.  I appreciate that these are closely intertwined, but the uncertainty analysis could potentially speak a bit more explicitly to these differences.  Are we really ready to directly evaluate the performance of a specific radiometer or are we still very early on in validating a concept?  If the ultimate goal were to validate a spaceborne microwave radiometer, would that be better done by looking at directly intercomparing the brightness temperature rather than the retrieved temperature, thereby eliminating uncertainty relative to the retrieval algorithm?  Managing expectations like this could help further establish the unique significance of the potential study.

Another more overarching point that could benefit from some increased treatment is the effect of sensor incidence angle.  Line 416 in the conclusions speaks to the presented assessment speaking to the importance of accurately accounting for the incidence angle, but I don’t feel this is done to the potential degree required.  There is some discussion near line 396 but this is significant and could use more up front and overall discussion.  The error of 0.2 discussed is actually smaller than some of the variations observed in practice.  The manuscript states (near line 230) that effects based on the IR angle were explored but not utilized.  How did the results compare?  While the angle measured by the infrared sensor obviously can’t be used directly, was the offset in incidence angles a constant that could possibly be accounted for?  Additionally, the results are presented in terms of the standard deviation of the incidence angle apparently computed over the entire deployment but in Fig 9 it is clear that over some periods the variation is much greater.  What is the expected uncertainty under these specific periods?  Is there some potential correlation between the observed errors/differences and these periods with greater angular variations? (the results in the figure might suggest that there is).  Line 416 in the conclusions speaks to the presented assessment speaking to the importance of accurately accounting for the incidence angle, but I don’t feel this is done to the potential degree required.

Other minor comments

Line 9 – “stable”:  While I understand what is meant here, I wonder if another word might be used in place of “stable” as it also often is used to refer to atmospheric conditions related to the slope of the temperature profile.

Line 12:  Since the previous sentence speaks of both skin and subskin temperature, I think it is important to be clear as to what quantity (presumably the microwave) the error values refer to.

Lines 24-25, “brightness temperature…emitted from”:  “representative of” might be more appropriate that “emitted from” as radiation is “emitted” but not brightness temperature.

Line 43: O’carroll -> O’Carroll

Line 51:  Isn’t what is novel here the use of the passive microwave radiometer?  Thermal IR instruments have been deployed extensively in the past.

Section 2.1:  I think it is important to mention that the ISAR is just one such instrument.  Other instruments provide radiometric measurements of the skin temperature including even more sophisticated instruments like the M-AERI.  It could be useful to cite previous studies showing positive intercomparison results between other IR radi  ometers.

Line 80:  The Hoyer reference could be improved.  From the listed website there are further directories and then multiple papers which are hard to align with what is referred to here.  The most seemingly closely related effort is not available through this website.  Is there any other information on this radiometer available?

Line 90:  While there is a reference here, given the importance of these steps, it might be helpful to the reader to explicitly say a bit more about the different steps.

Line 93:  Range of datasets “was” -> were.  Also line 97, data was -> data were

Line 95:  SLSTR definition later (line 106) should be moved to this first use.  What is the source of these data?  I see acknowledgement and reference of other data sets but I might have missed the source here.

Line 136:  Is there any possible direct calibration possible for the microwave brightness temperature/radiometric measurement?  Is there any corresponding absolute calibration?

Line 199:  “raw clean” seems to me to be a bit of a misnomer – instead perhaps “raw data with XX removed”

Section 4.2:  I wonder if it would be worth commenting on the potential use of detrending prior to the running average.  While the sky variability shown in Fig. 5 is longer term, is there any potential influence over the longer 60-min averaging time?

Line 219 and prior:  What is the effective emissivity at these microwave frequencies?  I believe there are significant differences from 1.  Given the longer-term variation of the sky temperature shown in Fig. 5, is there any potential for the influence of reflective effects?  How are these handled?  If the microwave data are independent of cloud liquid water and water vapor, what is the source of the longer-term sky variation in Figure 5?

Fig 5b:  Given the occurrence of greater incidence angle variations, is it worth expanding the x-axis here to account for the full range of variations seen?

Line 230 and nearby:  As touched on in my more general comments, could a constant offset in angle be applied?  Given the large sensitivity to incidence angle, I believe more justification of why this approach is sufficient is required earlier on.

Line 248:  I appreciate the limitation associated with the quality of the ERA5 winds, but I am less clear on use of the roll angle to justify setting winds to zero.  Doesn’t wind have an effect on roughness that can affect the measured brightness temperature?

Table 3:  Tying in to a broader discussion on incidence angle, given the range of incidence angle variations shown in Figure 9, for instance, is a value of 0.2 degrees for the incidence angle sensitivity truly sufficient?  How are temporal variations accounted for?

Line 330:  The difference here, as I understand it, is near 0.3 K – is this really “not significant”?

Line 388:  The grammar of this sentence could be improved.

Line 396:  Again tied to a potential broader discussion of incidence angle – Fig 9 suggests even larger variations in the incidence angle than 0.2.  How much larger does the effect get for these larger angular variations?

Line 410:  specially refurbishment -> specially refurbished?

Line 51: Not sure I understand why this is a first use of FRM TIR instruments as IR radiometers have been used for a long time and after the FRM4ST project many will have been checked against the SI. Do you mean the first FRM TIR against MW? Please make clear what you mean.

Line 76: EMIRAD instrument: Any comment regarding possible sideline contamination?

Please use uncertainty and error appropriately (see, for example, "the guide to the expression of uncertainty in measurement" for definitions). Examples:

Figure 5: Should be “instrument uncertainty” not “instrument error”

Line 180: Section 4.1. This section derive the MW instrument noise but there is nowhere where the instrument systematic uncertainties are discussed. Are these available anywhere? They will, of course, increase the total MW uncertainties.

Line 188, 194, 196: Should be “instrument(al) uncertainty”

Line 238: Observation uncertainty not error

Line 250: What fitting process was used (LSQ?)

Line 300: Section 5.3: Are there correlations between the MW/IR SST differences and elements of the retrieval function e.g. wind speed, incidence angle etc.?