Comments on “Forward operator for polarimetric radio occultation measurements” by Hotta et al.

The manuscript presents an interesting exploration of the ability of an NWP system to estimate observed RO polarization observations, a precondition to assimilate them. The authors find qualitatively good forward estimates, with an ability to identify the effects of rain and snow. Although the agreement is still not accurate in the quantitative sense, this can be interpreted as both the uncertainty of the detailed hydrometeor field, and a margin associated with the oblateness, and probably the tilt, of hydrometeors, quantities with large uncertainty.

Some suggestions are added here:

L23-24: “radio waves travel through oblate objects”

The principle is true both with oblate and prolate objects (hydrometeors may be both). Also, a large fraction of the effect happens because radio waves travel next to such objects (not just through). I suggest a minor adaptation of the sentence, for instance “radio waves travel through a medium containing ellipsoidal objects”.

L92: “to avoid negative values”: Would it be inappropriate to have negative values? I understand that it may be desirable to maintain the monotonicity of the interpolated field, and not contaminate it with a spurious wavy interpolation, but please correct me if I am wrong (or better, specify in the paper): is the differential phase shift necessarily positive? I would guess that it could be positive or negative, as a function of the oblateness/prolateness of hydrometeors, and one could have both along a line of sight. At least, I understand that atmospheric snow/ice can appear prolate in polarimetric radar. Please elaborate/clarify.

L95: “tangent point drift is not crucial for the regular RO observations”. It looks oddly expressed. It was found to be important, albeit it is indeed a small fraction. I agree that the difference must be much larger for hydrometeors. Also, L95 “presumably due to the weak horizontal gradient”. The word “presumably” also looks odd, as it is quite established that for this very fact one gets a perceivably less accurate, although still reasonable, result if the drift is ignored. Consider rephrasing.

L122: “axis ratio of the ice particles” Since this is finally applied to water also, should it not “axis ratio of the particles”, thus including water droplets? Also, it looks strange that the water shape vs rain rate relationship being relatively well established, it turns out to be less developed in the paper.

Given the importance of the ice vs liquid water phase that is being featured in the manuscript, could some approximate indicator of the height of the freezing point be added to some figures (notably the panels in Fig 1). To some extent, it is visible by the level where the rain signal becomes non-zero in each panel (in the range 2-6 km). It that curve happened indeed to be a reasonable indicator, you may mention it in the caption.

Sec 4.2: Sensitivity to displacement. A 10 km displacement is introduced. It is however not mentioned whether this is a good/inaccurate estimation of the geographic accuracy of IFS. Presumably, it was selected because this is order of magnitude the position error of AR and TC features within IFS at short lead time (say, about 6h). Is that so? Please comment whether these figures are indeed representative of IFS’s accuracy.

See attached PDF for review comments.