Correcting for Mobile X-Band Weather Radar Tilt Using Solar Interference

Round 1

Reviewer 1 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

In general, the paper has improved significantly.  Congrats to the authors. I have comments annotated to the draft. The authors should introduce a short  paragraph on the scanning parametrization, and  the moments that are used. that were used for analysis. I can tell that it is SNR, but it should be introduced properly. Why don't you analyzed SNRv, which would give you beamsquint?

 

Comments for author File: Comments.pdf

Author Response

See attached pdf

Author Response File: Author Response.pdf

Reviewer 2 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

This is a second review, even if it records as a first. In the previous review, I considered that the manuscript was already in very good shape.

Now, the authors have put a lot of effort and done a great work addressing the comments and suggestions by the reviewers, and the reasonings, descriptions, figures and discussions in the current manuscript have improved notably.

I consider that this work is ready for publication, but I have made a few, very minor side comments on the pdf file attached.

Comments for author File: Comments.pdf

Author Response

See pdf.

Author Response File: Author Response.pdf

Reviewer 3 Report (Previous Reviewer 3)

Comments and Suggestions for Authors

This revised manuscript, combined, with the review response, answers most of my previous concerns.  I noted a few minor items that should be corrected.

1) The lower right plot in Figure 4 should have a y-axis label. 2) The sentence starting with "For example" in line 237 is a bit hard to read.  I think it's saying that if tilt is properly removed, particles would appear more oblate, based on their polarization signature. Please consider re-phrasing. 3) line 255, I'm not sure what's meant by an elliptical melting layer. 4) Please provide a reference that illustrates D, I, and y_o.

Author Response

See pdf.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Overall remarks:

 

This is a nice exercise to assess the pointing accuracy  using the sun as a reference applying standard approaches  from literature to a X-Band dualpol radar that is deployed in a measurement campaign. The resulting tilting is applied to the data in a postprocessing step.

What is completely missing is the fact that you operate a polarimetric radar.  So the assessment should be done for  both the H- and V-channel separately.  Did you do this? This will provide an assessment of the antenna characteristics (i.e. beam squint). The results from the raster scans and a detailed discussion is missing. This needs to be included. But when writing this, I suspect, that you actually cannot do rasterscans and save the data for analysis (it is a Leonardo system). This is not clear to the reader. Please clarify. Nevertheless the results from those raster scan need to discussed. Radar software from other manufactures allow to save raster scans in ODIM, and allow for continuous scheduling of those scans (see Frech et al, 2019). One additional recommendation should be that radar software must be capable be to do scheduled raster scans.

To assess the benefit of the tilt-correction (which is a correction that is applied only to the data, and not to level the tower) the authors make use the topography of the study area: The tilt correction is “transported” into the data through the beam blockage correction only for the whole study period.  Though the result  seem to go into the right direction, I wonder if other factors indirectly contribute here. E.g. how do deal with the potential increased weight of bright band effects due to the pbb correction and the impact on your qpe? Have you checked case studies?

This approach also assumes that the estimated  tilt is there from the beginning, and is constant. This in contradiction to Figure 3, where the tilts becomes apparent starting 2020-04. The QPE statistics of Figure 7 do not really reflect this in my opinion. In that respect the qpe approach is presented here is flawed in my opinion, unless I have missed s.th.

So I have doubts about the usability of the qpe approach and the corresponding statistics.

 

some comments as I go through the manuscript:

l. 101: how often do you do a noise sample? And how? At a fixed elevation?

l. 133: so you don’t use the timestamp to determine the expected solar  position? Why not. I think this the established method

l 141: curious to hear: were you able to verify the calibration?

l. 159: I don’t understand the methodology to assess the quality: the RMSE is calculated from which parameters exactly?

 

l. 162:  with 6 min scaning sequence and the given elevations, it  seems surprising that you see only 10 hits a day. Did you verify that this number is reasonable? Seems to be a too small number We have a comparable (C-Band) an we see at least see at least 20 hits a day. You may use the ray timsteps and the corresponding el and az of the sun  to determine the maximum expected number of solar hits in your scanning. But perhaps this is reasonable for the latitude.. but it  should be verified.

L 194: tilt assessment.

So   for the tilt assessment you use the data from you raster scans. From the text is not obvious to what data you are using. Please clarifiy

l. 195: elevation offset = elevation pointing bias?

 

L 221: bimodal distribution of solar hits: I’m irritated by the usage of bi-modal. Might be a misunderstanding, but please clarify what is meant by this  (do you see two modes?)

 

 

L 253: here you could use the timstamps of a sweep and analytically compute how many solar hits you should see theoretically (see above).  I assume you have ntp synchronized timestamps for you ray header?

L.270: how do you come up with this statement, that the inclination is significant enough to impact radar observations? Do you show this somewhere, or does it refer to the target accuracy?

 

Figure 4: lower right plot: shown are all points that go into the fit? Looks like data before filtering. If not I am a bit surprised that you can do a reasonable with fir eq. 1. Why don’t you show the biases for the first fit (since you show it for the RMSE)?

 

l. 281 where do you show the results from the rasterscans?

L 285: the online analysis is what exactly? Is this defined somewhere?

 

L 391: Is this a spatial correlation or the temporal correlation?

L. 398: A constant underestimation of precipitation amounts by the radar of >50% is really high. Have you tried using different Z-R-Relationships for QPE? What could be the reason for the strong underestimation?

The point of integrating the results to catchments is kind of lost if the precipitation amounts would not have any use in e.g. flow modelling. Please elaborate the reasoning

L. 437: don’t understand why there  is an increasing occurance of beamblockage. Please explain!

 

Figure 6: Could you add some information on the total amount of precipitation recorded during the campaing, or an amount of precipitation events?

The relatively high correlation and the high bias at the same time make me think that most of the time the recorded precipiation is 0.

      A more applicable measure for similarity may be the Nash-Sutcliffe Efficiency which can handle time series with many zeros and a few spikes better.

 

Reviewer 2 Report

Comments and Suggestions for Authors

I find the research presented in this manuscript very relevant in the context of weather radar calibration procedures, and I believe it will be of great interest to the weather radar scientific community. It extends, merges, and adapts two antenna pointing calibration methodologies developed for fixed C-band radars to mobile X-band radars.

I think that the authors have done great work not only in the scientific aspect but also in the writing and presentation of the manuscript. It is sincerely a pleasure to review a first version in such good shape.

I have found that most of my concerns were appropriately addressed in the text, and it is evident that the authors have thoroughly considered, reasoned, explored, and evaluated the relevant aspects of the research.

I have some suggestions and notes that I have written as side comments on the PDF document of the manuscript. This is the first time I am presenting a review like this, and I have done it this way not out of indolence, but because I hope it makes it easier for the authors to review and answer them. If there were any problems reading or understanding the comments, please do not hesitate to contact me immediately.

Comments for author File: Comments.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript discusses application of methods for estimating ground-based radar pointing parameters from reception of solar interference.  While likely of general interest to the ground-based radar community, the manuscript is particularly applicable to radars with temporary deployments and at high latitudes, where solar interference during winter time is less useful than at lower latitudes.  My own expertise is in airborne and spaceborne radars; hence, I do not have direct experience with the data and procedures being reported. Nevertheless, I was able to generally follow the manuscript and so conclude that it is well-organized and clearly written.  I have a few specific questions and comments below that will, I hope, assist in clarifying the presentation for general readers like myself.  My main concern with the conclusions is the rather large variability in the retrieved parameters over the duration of the data collection.  I would have guessed that physical changes, like shifting or sinking supports, would result in fairly steady changes in the retrieved parameters.  Hence, a bit more explanation of why the authors have high confidence would be good.

More specific comments and questions:

In line 21, is this value (0.1) an example?  I would guess that the required accuracy might depend on radar parameters and on other factors.

Line 32 mentions assessing the position of radars.  Should this be “pointing”?

Line 152: I'm a bit puzzled by the need to retrieve both beamwidths and pointing.  Aren't the beamwidths already pretty well-known, while the pointing varies with the installation?

Line 156:  1 dB seems like a pretty tight threshold.  Has this been used by others or was it experimentally determined?

Figure 2 caption: this seems to be the first use of the term "bullseye".  For clarity, it should be defined first in the text.  Is this the name being used with the first method?

Section 3.3:  I think there really needs to be a diagram showing the various angles and model parameters.  This could be a new drawing or perhaps referral to a diagram in a previous publication.  This would help clarify, for example, the difference between the estimates of this section versus the previous section.

Line 208: My understanding is that both techniques use the solar data.  However, the statement here refers to the first method as “solar disk”.  Does the tilt analysis also use the solar disk?

Line 319: as noted above, the assertion of high confidence could be better supported.

Line 335 and reference to Figure 5:  Is the correction applied here using the retrieved corrections for each period of time?  If so, isn't this result expected?  One might expect the errors to be slowly varying - would some sort of average correction applied over the full campaign improve results?

Line 419, regarding the catchments with lowest usable elevation below 2 km: The expectation here is that these catchments should not have problems with PBB and so wouldn’t change much with the correction?