|Second round of review for the paper "Detection of reflections in GNSS radio occultation measurements using phase matching", given by |
Thomas Sievert, et al.
Now the paper is more clear. Almost all the issues raised in the previews review have been addressed. But it should be made more clear that
A) the red spike you show in the mod(U(a)) plots by processing both the simulated and the real occs is not always placed at amin (since you truncated the signal) but very close to amin (being the range of impact heights characterizing the reflected ray in the last part of the occultation not varying too much with SLTA [demonstrated clearly for both the cases in Fig 1, where it varies of hundreds of meters with SLTA variations of 20 to 40 km)]. Thus it should be made more clear that what you are showing is the contribution of the reflected ray at the point where you truncated the signal.
Could you please thus specify (with a Table or within the figures title or legend or caption) for each figure from Fig 3 to Fig 12 at which impact parameter the peak of the red line is placed for both the simulated and real U(a) (by visual inspection I guess is possible to locate it easily) and how far it is from the amin (for which direct ray and reflected ray join)? I guess that amin can be easily computed by iterating Eq 2,3 and 4.
B) Another point that is not fully clear to me is that you stated several times that you can apply the Phase Matching technique as it is. If I well understood, in Annex A you are showing that the time derivative of the optical path length S(t,a) is always the same when direct and reflected rays are considered. I agree that *if* it is the same, you can apply the PM as it is. But this is not very well demonstrated. And this is the crucial point of your paper. At page 15, #1-2, you are writing simply that "Taking the time derivative of this expression (A22) leads to the very same expression as Eq A14". I think that the sentence does not demonstrate anything, unless the reader wants to spend time in deriving expr A14 by himself. Please provide further clearer mathematical evidence (not difficult at all, but this will help the reader a lot)
Moreover, Eq A24 (the derivative of the PM phase function) does not provide the bending angle for reflected rays (Eq A21). Signs are wrong.
Having these points in mind (both well addressed in the paper), will make the entire paper definitely clear and easier to be understood.
I've few more editorial/technical points that I'd like to see addressed in the paper before its publication in this ATM Special Issue.
Page 2, #5: ...It has since than been employed...
Page 2, #29: The sentence is bad written. The subject is amin, which does not correspond to a surface refractivity of 330 N units. I'd write that the ray whose impact parameter is amin is tangent to the Earth's surface (I don't think that making reference to the value of the surface refractivity).
Page 3, #2: ... well to be used for reflected rays as well...
Page 3, #5: ... we used data from two real ...
Page 3, #20: after equations you should start with lower case letters. Check troughout all teh paper.
Page 4, #3: I'd use "centre of curvature coordinates" instead of "translation data"
Page 4, #7-#12: could you rephrase a bit this sentence? It is quite confusing.
Page 5, around #10: in my view it is here that you should better specify what I'm asking in my point A) before.
Figure 2: In Figure 1 you are showing PM amplitudes, direct/reflected SLTA(ha) for two particular simulated occultations. Could you use the same simulations when you produce/show Figure 2?
Page 6, #9: I cannot see any "negative" peak in Fig 8 - 12 as described. There are spikes or almost delta fadings in arg(U).
Page 13, Eq A10 or the following text. I guess you are referring to wavenumber in vacuum (k0) also in the Equation.
Page 15, beginning: here is where you should address the answer to my comment B)
Page 15, Eq A24: wrong signs?