the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Higher-Order Calibration on WindRAD scatterometer winds
Zhen Li
Ad Stoffelen
Anton Verhoef
Abstract. WindRAD (Wind Radar) is a dual-frequency rotating fan-beam scatterometer instrument on the FY-3E (Fengyun-3E) satellite. Scatterometers are generally calibrated using the linear NOC (NWP Ocean Calibration) method, to control the main gain factor of the radar. While WindRad is stable, the complex geometry, the design of the instrument, and the rotating antenna make the backscatter (σ°) distributions persistently non-linear, hence NOC is insufficient. Therefore, a higher-order calibration method is proposed, called HOC. The CDF (Cumulative Distribution Function) matching technique is employed to match the CDF of measured σ°s to simulated σ°s. HOC removes the non-linearities for each incidence angle. However, it is not constructed to remove the anomalous harmonic azimuth dependencies caused by the antenna rotation. These azimuth dependencies are reduced by NOCant (NOC as a function of incidence angle and relative antenna azimuth angle). Therefore, the combination of HOC&NOCant is implemented to correct both anomalous σ° amplitude and azimuth variations. The wind retrieval performance is evaluated with NOCant, HOC, and HOC&NOCant combined. The wind statistics and the cone distance metric both show that HOC&NOCant achieves the optimal winds for C-band and Ku-band. The calibrations have been tested on two operational input data versions; HOC works well on both data versions and HOC&NOCant can achieve the optimal wind performance for both data versions. This confirms the usefulness of HOC calibration in the case of non-linear instrument gain anomalies.
- Preprint
(4776 KB) - Metadata XML
- BibTeX
- EndNote
Zhen Li et al.
Status: closed
-
RC1: 'Comment on amt-2023-112', Anonymous Referee #1, 16 Jul 2023
This manuscript proposes a higher-order calibration method for the WindRAD, with the objective of mitigating the non-linear characteristics of the radar measured sigma0, and in turn deriving high-quality winds. The methodology is well described, and the results are quite promising. I think the manuscript may draw common interests from the ocean surface wind community. However, a few minor questions need to be addressed before publication.
- Do you use all of the sea surface data in the analysis of Figures 4 and 8? I would like to know how you deal with the quality control and the negative sigma0s.
- The schematic illustration of HOC in Fig. 3 is fine. However, how do you calibrate each particular sigma0 value according the bias derived from Fig. 3? Is it done in linear space or in dB? Again, how do you deal with the negative sigma0s?
- The azimuth-dependent bias in Figures 13 and 18 is quite suspicious. Does it exist in both versions of data? What could be the reason for the azimuth-dependent bias? Moreover, are the HOC and NOCant results the same for both versions of data?
- It would be nice to see an illustration or a table on the HOC and the NOCant results.
- Page 3, line 68: “correcting for air mass density” should be “correcting for the effect of air mass density”
Citation: https://doi.org/10.5194/amt-2023-112-RC1 -
AC1: 'Reply on RC1', Zhen Li, 15 Aug 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-112/amt-2023-112-AC1-supplement.pdf
-
RC2: 'Comment on amt-2023-112', Raj Kumar, 17 Jul 2023
This work is mainly to calibrate Fan Beam Scatterometer (WindRAD) data using proposed nonlinear calibration technique HOC. Manuscript is well written, however in my view it can be conscised more.
I have a few comments as following, which may require the explaination.
At many places in the manuscript, it has been mentioned that issues have been corrected in v2opr, so It is not clear, why V2opr has not been used for whole analysis of sigma0 to apply HOC based calibration, instead of using v1opr which has few drawbacks.
Line 127: Authors mention that nonlinearity can’t be corrected by NOCant, however plots in the figure suggest that NOCant also may correct the data. One need to check the difference in both corrections.
Line 163: It will be also good to know the differences in wind retrieval for NOCint and HOC
Figure 13a&b: What can be the reasons for HH & VV showing opposite behavior for low incidence angle.
Figure 13 c&d: After applying HOC correction, the correction doesn’t seem to be uniform for all azimuth angles. For both HH&VV, correction seems to improve for azimuth directions above 180 degrees.
Figure 14 suggest that NOCant doesn’t improve the wind retrieval quality as directions retrieval accuracies are almost same. For wind speed also, there is only marginal improvement. Does it mean that even NOCint also would have performed in similar way.
Figure 15: It is not clear, why HOC should be higher for sweet spot (centre WVCs). Normally at these WVCs, SD should be low.
Line 214: Authors have made a good point that MLE is lowest for HOC with NOCant, whereas the difference between NOCant alone and HOC combined with NOCant is almost same. What can be the reason that it is unlike C band. The nonlinearlty in the sigma0 should be frequency dependent for a particular angle (or range of angles) only.
Line 220: Couldn’t understand the mirroring effect between ascending and ascending. It will be advisable to explain it.
Citation: https://doi.org/10.5194/amt-2023-112-RC2 -
AC2: 'Reply on RC2', Zhen Li, 15 Aug 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-112/amt-2023-112-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Zhen Li, 15 Aug 2023
Status: closed
-
RC1: 'Comment on amt-2023-112', Anonymous Referee #1, 16 Jul 2023
This manuscript proposes a higher-order calibration method for the WindRAD, with the objective of mitigating the non-linear characteristics of the radar measured sigma0, and in turn deriving high-quality winds. The methodology is well described, and the results are quite promising. I think the manuscript may draw common interests from the ocean surface wind community. However, a few minor questions need to be addressed before publication.
- Do you use all of the sea surface data in the analysis of Figures 4 and 8? I would like to know how you deal with the quality control and the negative sigma0s.
- The schematic illustration of HOC in Fig. 3 is fine. However, how do you calibrate each particular sigma0 value according the bias derived from Fig. 3? Is it done in linear space or in dB? Again, how do you deal with the negative sigma0s?
- The azimuth-dependent bias in Figures 13 and 18 is quite suspicious. Does it exist in both versions of data? What could be the reason for the azimuth-dependent bias? Moreover, are the HOC and NOCant results the same for both versions of data?
- It would be nice to see an illustration or a table on the HOC and the NOCant results.
- Page 3, line 68: “correcting for air mass density” should be “correcting for the effect of air mass density”
Citation: https://doi.org/10.5194/amt-2023-112-RC1 -
AC1: 'Reply on RC1', Zhen Li, 15 Aug 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-112/amt-2023-112-AC1-supplement.pdf
-
RC2: 'Comment on amt-2023-112', Raj Kumar, 17 Jul 2023
This work is mainly to calibrate Fan Beam Scatterometer (WindRAD) data using proposed nonlinear calibration technique HOC. Manuscript is well written, however in my view it can be conscised more.
I have a few comments as following, which may require the explaination.
At many places in the manuscript, it has been mentioned that issues have been corrected in v2opr, so It is not clear, why V2opr has not been used for whole analysis of sigma0 to apply HOC based calibration, instead of using v1opr which has few drawbacks.
Line 127: Authors mention that nonlinearity can’t be corrected by NOCant, however plots in the figure suggest that NOCant also may correct the data. One need to check the difference in both corrections.
Line 163: It will be also good to know the differences in wind retrieval for NOCint and HOC
Figure 13a&b: What can be the reasons for HH & VV showing opposite behavior for low incidence angle.
Figure 13 c&d: After applying HOC correction, the correction doesn’t seem to be uniform for all azimuth angles. For both HH&VV, correction seems to improve for azimuth directions above 180 degrees.
Figure 14 suggest that NOCant doesn’t improve the wind retrieval quality as directions retrieval accuracies are almost same. For wind speed also, there is only marginal improvement. Does it mean that even NOCint also would have performed in similar way.
Figure 15: It is not clear, why HOC should be higher for sweet spot (centre WVCs). Normally at these WVCs, SD should be low.
Line 214: Authors have made a good point that MLE is lowest for HOC with NOCant, whereas the difference between NOCant alone and HOC combined with NOCant is almost same. What can be the reason that it is unlike C band. The nonlinearlty in the sigma0 should be frequency dependent for a particular angle (or range of angles) only.
Line 220: Couldn’t understand the mirroring effect between ascending and ascending. It will be advisable to explain it.
Citation: https://doi.org/10.5194/amt-2023-112-RC2 -
AC2: 'Reply on RC2', Zhen Li, 15 Aug 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-112/amt-2023-112-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Zhen Li, 15 Aug 2023
Zhen Li et al.
Zhen Li et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 220 | 42 | 18 | 280 | 9 | 9 |
- HTML: 220
- PDF: 42
- XML: 18
- Total: 280
- BibTeX: 9
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1