The paper provides a very comprehensive analysis of the Z_DR biases during a decade-long period of continuous observations with the X-band polarimetric radar at the University of Bonn and suggests a new “reverse Z – Z_DR method” for absolute calibration of Z based on the measurements of calibrated Z_DR. In the first part of the manuscript, the long-term statistics of the Z_DR biases estimated using the traditional “bird-bathing” and QVP-based calibration methodologies are presented illustrating temporal trends and differences between the outputs of the two techniques. This is a very valuable part of the study that demonstrates realistic spans of the Z_DR biases which seem to be quite sensitive to the changes of the radar hardware and software. The message is also that the bird-bathing calibration methodology is not a golden standard at all.

I have a few questions / concerns regarding the reverse Z – Z_DR method.

• The Z – Z_DR dependence in rain is very nonlinear and it looks that the best results for Z(Z_DR) retrieval can be achieved only for Z_DR within the range between 0.5 – 2.0 dB where a slope of the Z_DR(Z) dependence is maximal. I doubt that any reliable estimate of Z is possible outside of this Z_DR range.
• Certain criteria should be satisfied to select appropriate values of Z_DR for retrieval of Z and for utilizing the self-consistency relationship. One of the requirements is Φ_DP < 30°. The bias of Z_DR caused by differential attenuation in rain at X band is equal to the product β Φ_DP where the factor β strongly increases with Z_DR and often assumed to be 0.05 dB/deg. This means that ΔZ_DR can be as high as 1.5 dB for Φ_DP = 30° that is not acceptable if correction for differential attenuation is not performed. Should the Φ_DP requirement be stricter? Was the Z_DR correction for differential attenuation performed in the study?
• The authors do not mention a situation with wet radome if it rains over the radar. Were these measurements excluded from the statistics? The problem is that Z bias associated with wet radome can be very high (up to 20 dB) and Z_DR can be heavily biased as well.

Table of Contents

_________________

1. AMT 2026-493

.. 1. General comments

.. 2. Abstract

.. 3. Introduction

.. 4. Radar data

.. 5. Calibration of Zdr and Zh

..... 1. Zdr calibration

..... 2. Zh calibration

.. 6. Conclusion

1 AMT 2026-493

==============

1.1 General comments

~~~~~~~~~~~~~~~~~~~~

  - The authors provide a rigouros calibration of Zdr and Zh of their

    multi-year X-band radar data set

  - They first calibrate Zdr with standard techniques (birdbath and

    quasi-vertical profiles in light rain)

  - They introduce the reverse Zdr-Zh technique to calibrate Zh based on

    the calibrated Zh

  - They compare their results with GPM satellite measurements and

    self-consistency Zh-Kdp methods

This is a valuable data set which is an important basis for  polarimetric radar research spanning multiple years.  I'm not too convinced about their reverse Zdr - Zh technique for various reasons, i.e., the radome attenuation and ambiguity They authors might address these concerns to further strengthen the applicability of the proposed method.  I also do not understand the differences that occur in birdbath scanning and QVP profiles in light rain in view of Zdr offsets.  It might be necessary to provide some further insight into this.

1.3 Introduction

~~~~~~~~~~~~~~~~

  - In your introduction, you also need to consider and discuss the

    following article: Figueras y ventura, 2012:

    <https://rmets.onlinelibrary.wiley.com/doi/epdf/10.1002/qj.1934>

    which contains relevant information similar to what you are doing

  - Sphere calibration: Joshil, 2022:

    <https://www.mdpi.com/2072-4292/14/15/3534>

  - line 51 / 52: virtually generated radar targets: wang 2025:

    <https://ieeexplore.ieee.org/document/11143545>; schneebeli, 2025:

    <https://amt.copernicus.org/articles/18/5157/2025/>

  - line 59: birdbath scan: Gorgucci 1999:

    <https://ieeexplore.ieee.org/document/739161>

  - line 64: "One major benefit and motivation using the QVP..." -> but

    this is also true for the birdbath scan

  - line 70: "the results were not reliable for the whole period..." ->

    Why?

1.4 Radar data

~~~~~~~~~~~~~~

  - line 103: rho_hv of 0.7 is terribly low. Why do you use such a low

    threshold? Rain is usually above 0.995 for well calibrated radars.

  - line 105: Phase offset processing is not really necessary for Kdp

    estimation.

  Table 1:

  - Radar description: a 0.2us transmit pulse leads to a range

    resolution o 30 m. So the range resolution you are indicating are

    averages? Or you are sampling at at 25m resolution?

  - STAR is not really a transmit mode

  - Staggering mode is a transmit mode from my point of view

  - What kind of transmitter (Magnetron I guess, but what's the power)?

  - Antenna gain = ?

  - What do you mean with high PRF? What's the low PRF then?

1.5 Calibration of Zdr and Zh

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1.5.1 Zdr calibration

---------------------

  - line 118: You are using the Probert-Jones Gaussian approximation for

    the antenna pattern, which can introduce errors of up to 1 dB

    compared to the usage of the full antenna pattern. Can this be

    justified?

  - line 120: You are using theta_h^H and theta_V^H but only G^H

    (similar for both directions). Why?

  - line 123: If your reference plane is directly behind the antenna,

    why do you introduce A_t an A_r?

  - line 136: I don't think that Zdr needs to be calibrated to 0.2 dB

    for QPE applications.

  - line 141: If you look from below, raindrops appear spherical for Zh

    and Zv not only in light rain

  - line 150 onwards: I don't understand why the birdbath scan should be

    influenced by the radar hardware, since it is supposed to

    out-calibrate such effects.

    - Also: Consists your hardware of rotary joints or similar equipment

      or is your transmitter / receiver attached to the antenna

      directly?

    - Could rotary joints be a source of elevation dependence?

    - In general: I don't fully understand the reasoning here

  - line 166: 10 valid values in the height dimension depends on how you

    interpolate your range data to a vertical grid

  - line 175: What is the standard deviation of the canting angle? 8 deg

    as stated in line 177?

  - Not sure if sufficient information is provided on how you performed

    the T-matrix calculations.

    - Are you using the code of Mishchenko?

    - Can more information on the processing be found in Chen et

      al. (2021)?

    - What model of the dielectric constant of water are you using?

  - Line 181: You should also take into account the findings of Zeyong

    2019: <https://ieeexplore.ieee.org/document/9025914>

    - I also recall that Figueras y Ventura made experiments with Zdr

      calibration in light rain, but I'm not sure if this is published

      somewhere.

  - Line 194: I'm not sure if this argumentation is valid, since (as

    mentioned) the birdbath scan is supposed to calibrate such hardware

    changes.

  - Figure 4: This figure is using a misleading color scale, since

    negative values range in an interval between [-0.2 and -1]

1.5.2 Zh calibration

--------------------

  - You ignore the effect of radome attenuation. What is the

    consequence?

  - From your curve in Figure 3, a Zh value between 34 and 36

    corresponds to a Zdr value between 1 and 1.75, which is a huge range

    for Zdr but a small one for Zh.

    - How can you make an accurate calibration from such a curve?

  - line 224 and Figure 6: This is not really a too convincing fit

  - line 241: For a comparison

  - line 276: 0.25 -0.5 dBZ means 0.25 to 0.5 dBZ?

  - Table 3: gap-filled means that the reverse Zdr-Zh algorithm has been

    applied?

    - I.e., upper row is reverse Zdr-Zh, lower row Kdp-Zh self

      consistency

  - line 295: I'm not convinced that the reverse Zdr-Zh algorithm

    outperforms the Kdp self consistency.

    - Kdp is independent of radome attenuation, which is not true for

      Zdr.

    - In addition the Zdr - Zh relation is ambiguous (as mentioned

      before)

    - Just because it fits better to the satellite is not a strong

      argument

  - If you compare to the satellite anyway, why not taking the offsets

    obtained from satellite comparisons directly?

1.6 Conclusion

~~~~~~~~~~~~~~

  - See comment above