Mitigating Radome Induced Bias in X-Band Weather Radar Polarimetric moments using Adaptive DFT Algorithm

Padmanabhan, Thiruvengadam; Lesage, Guillaume; Ramanamahefa, Ambinintsoa Volatiana; Van Baelen, Joël

doi:https://doi.org/10.5194/amt-2024-117

Preprints

https://doi.org/10.5194/amt-2024-117

Preprints

22 Aug 2024

| 22 Aug 2024

Status: a revised version of this preprint is currently under review for the journal AMT.

Mitigating Radome Induced Bias in X-Band Weather Radar Polarimetric moments using Adaptive DFT Algorithm

Thiruvengadam Padmanabhan, Guillaume Lesage, Ambinintsoa Volatiana Ramanamahefa, and Joël Van Baelen

Abstract. In recent years, the application of compact and cost-effective deployable X-band polarimetric radars has gained in popularity, particularly in regions with complex terrain. The deployable radars generally use a radome constructed by joining multiple panels using metallic threads to facilitate easy transportation. As a part of the ESPOIRS project, Laboratoire de l’Atmosphère et des Cyclones has acquired an X-band meteorological radar with four panel radome configuration. In this study, we investigated the effect of the radome on the measured polarimetric variables, particularly differential reflectivity and differential phase. Our observations reveal that the metallic threads connecting the radome panels introduce power loss at vertical polarization, leading to a positive bias in the differential reflectivity values. To address the spatial variability bias observed in differential reflectivity and differential phase, we have developed a novel algorithm based on the Discrete Fourier Transform. The algorithm's performance was tested during an intense heavy rainfall event caused by the Batsirai cyclone on Reunion Island. The comparative and joint histogram analysis demonstrates the algorithm's effectiveness in correcting the spatial bias in the polarimetric variables.

Received: 06 Jul 2024 – Discussion started: 22 Aug 2024

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Thiruvengadam Padmanabhan, Guillaume Lesage, Ambinintsoa Volatiana Ramanamahefa, and Joël Van Baelen

Status: final response (author comments only)

RC1:
'Comment on amt-2024-117', Anonymous Referee #1, 14 Oct 2024

It was my pleasure to review the paper entitled “Mitigating Radome Induced Bias in X-Band Weather Radar Polarimetric moments using Adaptive DFT Algorithm”. This study proposed a new method based on the Discrete Fourier Transform to correct the spatial variability of the ZDR and PHIDP offset for X-band radars. This paper may be interesting for the operational radar community. However, I think major revision is needed before publication. Please see my comments and suggestions bellow.
Major comments:
1) With the same method, the author tried to correct the “azimuth” variability of ZDR and PHIDP on PPI scan and to correct “vertical” variability on a RHI scan.
Figure 1 Phidp on PPI scans.
Figure 2 Some ZDR profiles on RHI scans in a sub-figure.
Figure 3 Illustration of the method based on a RHI scan of ZDR
Figure 4 Application of the method for PHIDP and ZDR on PPI scans
Figure 5 Application of the method for PHIDP on RHI scan.
According to my knowledge, it is too ambiguous to mix all these contents in the same investigation and attribute all of them to the radar radome.
For the RHI scan, it is well known that ZDR depends on the elevation angle (see Bechini et al. 2008: Differential Reflectivity Calibration for Operational Radars). If elevation angle is equal to 90°, the ZDR is expected to be around 0 dB, while ZDR is much larger than 0 dB for the elevation angle equal to 0°. This is totally due to the geometry of raindrops shape, but nothing related to the radome. Secondly, to my knowledge, the ZDR (or Phidp) difference at different elevations can be also related to the rotary joints of antenna for some radars.
For the PPI scan, the impact of radome is easier to understand (from the figure 1 of the paper). However, when the radome is wet, in the past many studies showed the ZDR bias on PPI scan depending on the direction of wind.
So the physical explications of the ZDR and Phidp biases can be very different for the PPI and RHI scans. In my opinions, it will be better to deal with them separately in different sessions.
2) The author used a moving window on the DC components of the DFT to filter the bias. A Dc component (k=0) is simply the averaged all values of ZDR (or PHIDP) along a radar ray (I called it mean_ZDR_ray or mean_PHIDP_ray). So the main idea of the paper is to assume the median of mean_ZDR_ray (or mean_PHIDP_ray) should not be biased. Hence we can use these medians for corrections. The author should carefully deal with the following questions in details:
- 2.1 How the proposed method can correct an absolute bias? It seems to me that the method can only reduce the spatial variability in ZDR and PHIDP. If there is a global and constant bias (e.g. 1 dB) induced by radome at all directions (all azimuth directions and all elevations), how the median of a moving window can determine and then filter this constant bias?
- 2.2 The averaged ZDR (or PHIDP) values along a radar ray are impacted by the attenuation as well, particularly for a X-band radar during a tropical cyclone event. We can get very negative ZDR and high PHIDP on some radars rays due to the attenuation. And we see often very large ZDR in the bright band. And these values can have large impact on your median of the moving window. The conventional method takes only “first and short” rain segment (close to radar) to calculate the averaged ZDR and PHIDP offset to avoid the impact of attenuation. Please explain the advantages of your method compared to the conventional method.
Some other comments:
- Line 222: In the figure 2, the author illustrates some RHI observed from the X-Band radar. The horizontal axis is marked as range direction in km. I am surprised that the author found the reflectivity equal to 20 dBZ (or 22 dBZ) in the figure 2c at a range of 200 km with elevation angle equal to 31° (or 36°). The height of the radar beam at 200 km is equal to 200 km x tan (31°) = 120 km! 20 dBZ at 120 km of the atmosphere is not impossible for X-band radar to my knowledge.
- Line 77: The author showed the bias induced by radome on PPI occurring at approximately 355, 85, 175 and 265 degrees (Line 71). Later, the author selected the RHIs at 28, 300, 45 and 76 degrees of azimuth and showed the illustrations in Figure 2. Please justify the selection of these RHIs at 28, 300, 45 and 76 degrees of azimuth? Why the author don’t show a RHI at 355 degrees (or at 85, or at 175, or at 265 degrees) of azimuth which is impacted by radome according to the Figure 1?
- Line 124 “The threshold value is determined based on the moving median window of …” The author should explain the width of this moving median window (how many azimuths/elevations points are used to calculate the median). This is a fundamental parameter of the method. If it is possible, I hope to see why the author selected such width?
- Line 125 : Missing a space between the "of" and "F[θ, 0]"
- Line 133 : "the zeroth frequency values that surpass the threshold are multiplied by the calculated correction factor, while the remaining values are left uncorrected." Please express "the remaining values are left uncorrected" by mathematics expressions. If I understand well, it should be
F ' [θ, k] = F [θ, k] if k <> 0
- Line 138 : In the equation (6), I think the F[θ, k] should be the power spectrum after the filter of the correction factor (A,B). But in the equation (1), the author uses the same F[θ, k] to represent the power spectrum before the filter. Same F[θ, k] in the (6) and (1) leads to confusion.
- Line 239. In Figure 6, the author forgets the color palette to indicate the Count or Frequency of the joint histograms of ZDR-ZH.

Citation: https://doi.org/10.5194/amt-2024-117-RC1
- AC1: 'Reply on RC1', Thiruvengadam PADMANABHAN, 25 Nov 2024
  
  Comment 1:
  Comment 1.1:
  With the same method, the author tried to correct the “azimuth” variability of ZDR and PHIDP on PPI scan and to correct “vertical” variability on a RHI scan.
  Figure 1 Phidp on PPI scans; Figure 2 Some ZDR profiles on RHI scans in a sub-figure; Figure 3 Illustration of the method based on a RHI scan of ZDR; Figure 4 Application of the method for PHIDP and ZDR on PPI scans; Figure 5 Application of the method for PHIDP on RHI scan.
  According to my knowledge, it is too ambiguous to mix all these contents in the same investigation and attribute all of them to the radar radome.
  For the RHI scan, it is well known that ZDR depends on the elevation angle (see Bechini et al. 2008: Differential Reflectivity Calibration for Operational Radars). If elevation angle is equal to 90°, the ZDR is expected to be around 0 dB, while ZDR is much larger than 0 dB for the elevation angle equal to 0°. This is totally due to the geometry of raindrops shape, but nothing related to the radome.
  Response 1.1:
  Thank you for your insightful comments and for referencing the Bechini et al. (2008) study. You are correct that ZDR depends on the elevation angle due to the geometry of raindrop shapes:- the ZDR decreases with increasing elevation angle. However, our study detected an anomalous increase in ZDR at certain elevation angles in our RHI scans, which suggests the presence of bias at specific elevation angles. We believe these biases are introduced by the radome joints, affecting the signal differently at various elevation angles.
  Our correction algorithm is specifically designed to address these bias at specific elevation and azimuth angles. It does not alter the expected property of ZDR variation due to raindrop geometry. In Figure A (attached), we present the average ZDR observed across different elevation angles in stratiform rainfall from the 300° RHI scan measurements between the 1 km and 2 km range near the radar over 10 consecutive time steps, focusing on uniform precipitation below the bright band. The 300° RHI scan angle was chosen because it exhibits minimal ground clutter at close range, due to the absence of significant orography in this direction. Prior to the correction, the ZDR profile showed an unexpected increase at certain elevation angles, deviating from the theoretical decrease expected from raindrop shape considerations alone. After applying our correction algorithm, these anomalies are effectively removed, resulting in a ZDR profile that aligns with the theoretical expectations Fig A(c) (i.e., decreasing consistently with increasing elevation angles). This correction enhances the accuracy of ZDR measurements, thereby fulfilling the criteria necessary to apply the Bechini et al. (2008) algorithm.
  Regarding the potential ambiguity in mixing azimuthal variability on PPI scans and vertical variability on RHI scans, our intention was to comprehensively investigate the radome joints’ specific impact on our radar measurements in both horizontal and vertical scanning modes.