Real-Time Calibration and Monitoring of Radar Reflectivity on Nationwide Dual-Polarization Weather Radar Network
Round 1
Reviewer 1 Report
Review: Real-Time Calibration and Monitoring of Radar Reflectivity on Nationwide Dual-Polarization Weather Radar Network by Lee et al.
Remote Sensing Manuscript # 1297399
Comments and questions:
Main comment: I think the authors have done very good work in the use of RCA and SC for calibrating the KMA radar network and I fully agree with the benefits of this work. The main issue with this manuscript is that the overall method needs to be better explained. I think a more detailed explanatory section near the beginning that provides the big picture would be extremely useful.
Other comments and questions:
Introduction:
Lines 36-42: “Even in a homogeneous……” This sentence needs to be split as it is too long. Splitting the sentence will improve readability and clarity.
Section 2
Line 104: “The KMA’s radar was mainly….” please include “network” after radar as you are discussing the KMA network and not a single radar.
Figure 1 caption: add “network” to end of first sentence. Please modify the second sentence to provide clarity. Perhaps this….. “The ten circles indicate the 240 km radius coverage of each radar within the KMA network.”
Is the same scanning strategy used for all 10 KMA network radars?
Lines 112-115: The text would be clearer if the quality-controlled reflectivity field was referred to by a unique identifier. You state that “ZH” represents unfiltered reflectivity (line 113), but then refer to “ZH after the quality control process” in line 114.
Table 1 and text: A single PRF is shown for each elevation, however Table 1 shows that a dual-PRF is used. Please explain “5:4” in Dual PRF column. For the antenna speed, please specify in degrees per second.
What is the maximum unambiguous range for the GSN radar? Is this the same for all 10 radars?
Section 3.4
Line 240: remove comma after radar
Line 241: for clarity, please change “unsatisfied” to “did not satisfy”
Table 2: Over what time period does this show the number of overlapping volumes? Is this for times when rain was occurring?
Lines 242-243: You state that the two radars with the largest number of overlapping volumes were matched for comparison. But wouldn’t this be dependent on the time period analyzed and the amount/coverage of rain in the KMA network?
Line 244: How was beam blockage corrected? Do you mean that rays with partial beam blockage were removed from the analysis?
Lines 243-248: You state that errors (calibration bias, beam blockage, and attenuation) were corrected before extraction of ZH for intercomparison. From lines 112-114 you state that ZH represents uncorrected reflectivity. The mixing of label ZH to describe both uncorrected and corrected reflectivity is confusing.
Section 4.1: The authors refer to the KMA weather radar as if this was a single radar (line 261 and other lines). Please clarify the text to indicate that results are from the network of operational radars.
Line 260 states that “ZH” was monitored for SC and GC. Lines 112-114 state that ZH represents uncorrected reflectivity. This is okay for GC, but SC uses corrected data as described earlier in the text.
Figure 6: RCA results from 8 radars are shown. Both Figure 1 and Table 2 indicate 10 radars in the network. (Figure 7 shows results from 10 radars.) “BRI” and “JNI” appear to be missing from Figure 6 and Table 3.
Figure 7: There are numerous questions about this figure.
1) Why are MYN and GSN the only radars to show the RCA bias (red dots)? 7 of the 8 radars indicated in Figure 6 show noticeable changes in RCA.
2) Why are there no vertical lines in SSP and GSN corresponding to P1, P2, and P3?
3) MYN RCA (red dot) at start of P2 indicates an RCA bias of +0.98 dB, while the MYN panel in Figure 6 shows RCA change of approximately 4 dB. This is very confusing.
4) GSN RCA (red dots) show RCA-determined bias of -5.30 dB and -1.95 dB. Neither of these points correspond to RCA changes seen in GSN panel of Figure 6.
Table 3: Was the mean RCA computed in linear units and then converted back to dB?
Line 268: Change “black” dot to “blue” dot to correspond to Figure 7.
Section 4.1.2
Line 292: Please explain the logic as to how dMedian (Equation 2) is used to determine an error in antenna elevation angle. I’m not asking for much here…..just a few sentences explaining the logic.
Figures 8a and 9a: Please correct typo “dMeian”
Lines 298-299: “The number of CGs in Fig 8c is similar to that in Fig 8a.” It is not clear what you are trying to say.
Lines 308-315: It makes sense that if the azimuth rotated slightly the dMedian values would change. So please explain why the RCA values do not change.
Lines 335-337: “The calibration bias (0.98 dB, red dot in Fig. 7) was estimated from ΔRCA (2.97 dB) even if the SC method was not applicable.” So if SC was not applicable, and change in RCA was 2.97, then how did you obtain a bias of 0.98 dB?
Lines 341-344: You state the change in RCA was approx. 4 dB, yet Figure 7 shows -5.3 dB with the red dot. Some of this text needs to be together with figures 6 and 7 for clarity. It just is not clear how you are combining the use of RCA and SC – further clarification is needed.
Section 4.2
Line 346: How is the “mean bias” computed? Is mean bias routinely computed on a daily basis as shown in Figure 10?
Figure 10: What are some mean bias numbers shown in black and some in red? Is this simply the magnitude of the bias (< 1 dB = black and > 1 dB = red)? The numbers 1-9 in the panels appear to correspond to Table 2 – is this correct?
Figure 10: In each panel, it is not clear the benefit of showing radar pairs when both radars are uncorrected for bias. The comparisons show one biased radar compared to another biased radar.
Lines 350-355: In order to have a bias comparison throughout the network, you need to start with one or two well-calibrated radars. Are you stating in lines 353-354 that GSN and JNI are your reference radars? The procedure in determining the mean bias between radar pairs is not clear. Are you adjusting for the bias in one radar compared to a reference radar, and then using that corrected radar to then compare with another?
Figure 11: Results show very good agreement after bias corrections have been applied, but the overall method of how the bias corrections are determined and pushed through the network is not clear. Are you starting with a reference radar that is known to be well-calibrated and then the bias checks are pushed through the entire network? Is RCA being used at this point or is this only from SC?
There was no discussion of calibrating differential reflectivity (Zdr). Do the KMA network radars perform vertical profile (90-degree elevation) PPI scans? If not, do the radars use an engineering approach for Zdr calibration?
===== End of review =====
Author Response
Please see the attachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
Real-Time Calibration and Monitoring of Radar Reflectivity on 2 Nationwide Dual-Polarization Weather Radar Network
Jeong-Eun Lee, Soohyun Kwon and Sung-Hwa Jung
Submitted to Remote Sensing, reference 1297399
General comments
This paper considers the network of 10 S-band dual polarized radars in South Korea and suggests to combine in this case several calibrating methods: adjustment from ground clutter, self-consistency from polarized data, and intercomparison. They point out that for such a combination no information external to the network itself is needed.
The emphasis is definitely on homogenising the network. The authors do conclude that with a short delay it is possible to bring substantially down the discrepancies likely to damage the calibration and hence the quality of the network data and monitoring abilities.
There are many specific remarks or questions listed below. Although the English is generally correct, I have found some statements and expressions difficult to understand. Most of the remarks aim at making the work both more accurate and easier to understand by the reader.
They are therefore minor. I suggest however to consider carefully the point 32 below.
Finally, in my opinion a small item is missing in this work: a comparison between (i) what seems to be achieved here in terms of calibration and (ii) what are the requirements on the side of the users. On figure 12, red lines bracketing a +or- 1dB are drawn; it would be interesting to know whether these lines simply help to show that the corrected calibrations do stay on the whole within this interval, or they are to be understood as a requirement set by users of the radar reflectivity.
Detailed comments
01 L029 I do not understand " the construction of its network "
02 L036-42 is a comma missing after "over the network"? otherwise this sentence is missing a subject.
03 L059-060 Is this a serious problem? I would have thought that poor calibration in periods where there are no precipitation echoes is not a major drawback.
04 L104: I assume you are referring to the network; hence the text ought to read: "The KMA's radars were mainly installed….". Otherwise please explain.
05 L111-112 is the 1° figure valid also for the angular resolution along elevation?
06 L122 in table 1, how can the antenna scan speed be given in m/s units?
L129-130: What I understand is as follows: GC calibration is only able to measure the difference between the actual "radar constant" (meaning the factor which aggregates every effect associated with the radar system) and some "baseline" radar constant.
Hence it is correct to state as you say that it is a relative calibration. But is not it true that, as soon as the baseline radar constant is known thanks to an absolute calibration, it is known once and for all, so that GC calibration becomes an absolute calibration? Please comment.
07 L158-159 This sentence is not completely clear. Do you mean that the 0,5 dBZ numerical resolution was inadequate? What is the nominal resolution then?
08 L166 figure 3: what is the use of the colour code? Also, the meaning (units) of the x and y coordinates ought to be indicated.
09 L166 figure 3: assuming that the thin black represents Jeju's coastline, it looks like the GSN radar is located over the ocean. Is this really the case?
10 L170-172 I wonder why you are quoting Hunziger et al here, since this reference is focused (as stated in your introduction) on extending the RCA approach to higher frequencies and RHI scans, and these issues do not seem relevant here. Also, I fail to understand the " with a known calibration value for a posteriori calibration" end of the sentence.
11 L173-174 Again I find difficult to understand this statement. RCA provides modifications of calibration with respect to a baseline. As long as the baseline is arbitrary, the calibration is indeed a relative one.
Actually, I had to wait until line 186 to understand: you are using the clutter data in a purely differential approach. Then the exact knowledge of a baseline is no longer an issue.
12 L201-202 I suggest that you indicate that this empirical relationship was obtained using disdrometer measurements over South Korea.
13 L207-208 This is probably not significant. However, in the paper by Kwon et al, the Z_H/K_DP relationship reads (Z=7.3x10^4 K^1,11) in the text, against (Z=7.3x10^4 K^1,1) within figure 3, and (Z=7.33x10^4 K^1,10) in the legend of figure 3. Was there a reason for selecting your choice?
14 L213-214 Figure 4: could you comment on the slope (0,80) in this scatter plot?
15 L221-222 Figure 4: please indicate the meaning of the dotted lines.
16 L241 Should not you replace "unsatisfied" by "do not satisfy"?
17 L241-242 How do you define an elementary overlapping volume? While it does not seem difficult to identify the overall overlapping volume, as illustrated on figure 5a, it seems trickier to define elementary overlapping volumes to be numbered in table 2.
18 L254 how are defined the "top" and "bottom" of the radar beams with respect to the properties of the antenna pattern?
19 L263-264 This is definitely not recommended! Please try to organize your text in such a way that the reader does not have to rely on explanations to be given later.
20 L272-273 Please explain why on figure 6 time series are shown for 8 radars, while there are 10 radars indicated on figure 1. On the other hand, the reader will find again 10 time series on figure 7!
21 L278-279 figure 7: What is the meaning of the vertical thick dotted lines? I first thought they corresponded to the limits between the identified P1/P2/P3 phases; then this rule does not apply for SSP nor GSN for example.
22 L288-289 please correct to "This will be proven in section 4.2"! However, perhaps it would be more logical to shift this somewhere on section 4.2 after the proof has been given.
23 L296-298 In addition, according to figure 8a, the RCA fluctuations during the P2 interval were considerably larger than during P1 or P3. This again seems to point to pointing problems along the elevation angle.
24 L306 "(c) June 25" ought to be replaced by "d) June 25"
25 L312-315 While I am ready to believe this conclusion, the PPI maps on figures 9b and 9c are not fully convincing. Maybe changing the scale for the colour code might bring some improvement.
26 L350-351 This sentence is puzzling. Is 4,591 a magic value?
27 L353-354 Sorry, I could not understand the meaning of this sentence.
28 L358-359 and 372 Please indicate the thresholds selected for using red fonts in the subplots. Also, what is the meaning of the numbers from (1) to (9) in the subplots?
29 L372 It would be better to remove "except for"
30 L374 Using "the" (end of line) imply that these events have already been mentioned. This is however not the case.
31 L408 add a "s" to "period"
32 L415-416 Although I have reread the text twice, I could not find anywhere mention of this " When the change in RCA exceeds 1.0 dB" condition. Therefore, unless I am mistaken (in which case accept my apologies), it is not possible to introduce this condition in the conclusions.
33 L434 typo on "intercompairson"
Author Response
Please see the attachment
Author Response File: Author Response.docx
Reviewer 3 Report
This research article discusses real-time calibration and monitoring of radar reflectivity for a dual-polarization weather radar network. This article combined three techniques 1) GC monitoring, 2) SC, and 3) intercomparison to calibrate S-band reflectivity data in real-time. This paper is well written, and the objective of the paper is very clear. The results are presented nicely with the appropriate figures. The findings from this article are helpful to other researchers in this field. Since this paper looks perfect, I recommend accepting this article in its present form.
Author Response
We are grateful for your consideration of this manuscript.
Round 2
Reviewer 1 Report
The clarity of this second version has been improved. The authors have addressed the majority of my questions - thank you.
