the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jan 2023
Calibrating Radar Wind Profiler Reflectivity Factor using Surface Disdrometer Observations
Abstract. This study uses surface disdrometer reflectivity factor estimates to calibrate the vertical and off-vertical pointing radar beams produced by an Ultra High Frequency (UHF) band radar wind profiler (RWP) deployed at the US Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) program Southern Great Plains (SGP) Central Facility in northern Oklahoma from April 2011 through July 2019. The methodology consists of five steps. First, the recorded Doppler velocity power spectra are adjusted to account for Nyquist velocity aliasing and coherent integration filtering effects. Second, the spectrum moments are calculated. The third step increases the signal-to-noise ratio (SNR) due to signal power leakage during the Fast Fourier Transform (FFT) calculation, which can exceed 20 dB during convective rain events. The fourth step determines the RWP calibration constant for one radar beam (called the “reference” beam) by comparing uncalibrated RWP reflectivity factors at 500 m above the ground to 1-min resolution surface disdrometer reflectivity factors. The last step uses the calibrated reference beam reflectivity factor to calibrate the other radar beams during precipitation. There are two key findings. The RWP sensitivity decreased approximately 3-to-4 dB/year as the hardware aged. This drift was slow enough that the reference calibration constant can be estimated over 3-month intervals using episodic rain events. Calibrated moments are available on the DOE ARM data archive and Python processing code is available on a public GitHub repository.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2925 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2925 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1405', Anonymous Referee #1, 12 Feb 2023
Review of:
“Calibrating Radar Wind Profiler Reflectivity Factor using Surface Disdrometer Observations”,
submitted to the Atmospheric Measurement Techniques (AMT) Journal.This manuscript presents a methodology to calibrate radar wind profiler reflectivity factor using disdrometer data collected during rain events.
The authors found that the ageing of the hardware over time greatly impacts the RWP sensibility as it produces smaller values of SNR. The authors proposed compensating for this by varying the calibration constant and several other methods clearly described and evaluated using long-term data sets.It is great to see that the data sets and the code are available (or will be in the near future), thus increasing the reproducibility of the proposed methodology.
I reckon this manuscript totally fits the scope of the AMT journal, and it is an excellent contribution to the radar community.
I have minor comments and suggestions that may help improve the paper.
1. The Spectrum adjustment methods and the adjustments due to Nyquist velocity aliasing, coherent integration filtering, and increased noise power will benefit from a diagram. A flowchart describing the critical aspects of the proposed algorithm could help depict the method more clearly.
2. The sensitivity of the radar plays an important role in the calibration procedure explained in this manuscript. If the RWP cannot collect data during rain events, I reckon this method will not be valid for calibration. I suggest the authors add a discussion regarding this or if some other meteorological target could be used instead.
Citation: https://doi.org/10.5194/egusphere-2022-1405-RC1 - AC1: 'Reply on RC1', Christopher Williams, 21 Mar 2023
-
RC2: 'Comment on egusphere-2022-1405', Anonymous Referee #2, 27 Feb 2023
The manuscript suggests a very interesting methodology to calibrate UHF radar wind profiler data with the help of surface distrometer observations. Getting radar reflectivity values from the hitherto uncalibrated power measurements would clearly increase the usefulness these instruments. This approach furthermore provides an opportunity for a long-term monitoring of the hardware status, which offers very welcome insights as nicely demonstrated by the authors. There are a couple of points that the should be addressed in a revised version of the manuscript:
As the authors describe, it is known that excessive broadening of Doppler spectra during precipitation leads with neccessity to an incorrect estimation of the noise level by the Hildebrand and Sekhon (1974) method, because this broadening often affects the complete so-called "full-scale" velocity interval (which is bounded by the Nyquist velocity), so that the prerequisites inherent in the noise estimation algorithm are violated. Of course it is possible to address this issue by adjusting the radar settings to allow for a much wider Nyquist limit, in particular by reducing the number of coherent integrations and increasing the length of the time series for the FFT. However, this can not be done a-posteriori.
The authors then assert that the decrease of the signal-to-noise ratio (due to the spectral broadening described above) is due to signal power leakage of the Discrete Fourier Transform (FFT) calculation. Spectral leakage due to the finite extent of time series (in the case of RWP coherently integrated I/Q data) is a well-known fact in Fourier transforms and it is usually controlled through the multiplication of the time series with some kind of window function. However, the authors unfortunately do not provide information on what particular window function was used in the ARM RWP systems. The Vaisala LAP-3000 RWP traditionally use a "von Hann" (or Hanning) window since this operation was quite easy to apply in frequency domain, but other options were possible in later versions of the Vaisala software. The authors should provide this additional detail.
However, spectral leakage of the Fourier transform is not the only explanation for the broad Doppler spectra observed with RWP during precipitation: Tests with a similar radar have shown that the broadening of the Doppler spectra is rather independent of the window function used in RWP signal processing, at least as long as a reasonable selection like the Hanning window is made. This observation is in contradiction to the assertion that the broadening is due to spectral processing. Given the dynamic range of the RWP receiver it therefore appears to be more likely that the broadening is caused by signal contributions from the antenna sidelobes. The authors should comment on this alternative explanation, even though it has no consequences for the correction methodology presented.
A few other minor remarks:
Line 16/17: "The third step increases the signal-to-noise ratio (SNR) due to signal power leakage during the Fast Fourier Transform (FFT) calculation" - should be reformulated for more clarity, as the broadening (regardless whether this is due to the antenna sidelobes, or due to spectral leakage) leads to an increase of the noise level and thus a decrease of SNR.
Line 34: "At the radar measurement level, radars measure the return signal power as a function of range" - suggest a different wording
Line 37 "every radar subsection" - suggest radar hardware component
Line 39 "pole mounted corner reflector calibrations" - this method is impractical for the rather large and fixed phased array antennas of the RWP and does not need to be mentioned here
Line 57/58: "to account for radio frequency interference (RFI) that sporadically increases noise power estimates" - The effects of RFI remain largely unclear. Therefore the authors should provide more details on the characteristics of this particular RFI contamination. Not every RFI signal increases the noise power.
Line 96 (Table 1 Pertinent RWP operating parameters): For sake of completeness the authors should also provide an short overview of the RWP signal processing algorithms used. Was there any kind of time-domain nonlinear filtering applied ? Was the spectral integration done using Merritts (1995) statistical averaging method? Which moment estimation algorithm was used - single peak picking or multiple peak picking?
Line 243 "The impact of the TDA low-pass filter can be mitigated by applying a correction factor.." The authors could perhaps add a reference to Wilfong et al. (1999) "Optimal Generation of Radar Wind Profiler Spectra" which further discusses the TDA filter characteristics.
Line 277/278: "As the signal power magnitude increases, the FFT leakage causes the spectrum noise power to increase above the noise power produced by other radar noise sources". This statement is rather awkward and should be reformulated, also in view of the remarks made above.
Line 312 plus multiple other occasions: "Parson’s correlation coefficient" should be corrected to Pearson’s correlation coefficient
Line 645: The given link "https://github.com/ChristopherRWilliams/rwp/Python/spectra" is incorrect. The correct link is obviously https://github.com/ChristopherRWilliams/RWP_Python_moments
Citation: https://doi.org/10.5194/egusphere-2022-1405-RC2 - AC2: 'Reply on RC2', Christopher Williams, 21 Mar 2023
-
RC3: 'Comment on egusphere-2022-1405', Anonymous Referee #3, 06 Mar 2023
Williams et al present a methodology for long-term calibration of UHF radar wind profiles using collocated surface disdrometer observations. They apply their method to an 8-year dataset collected at ARMs Southern Great Planes site. The approach is technically sound and the manuscript is well written, hence I recommend the study for publication after very few minor corrections. The open availability of source code and data already during the discussion phase has to be highlighted positively.
Comments
- L32: It should be stated more clearly, that the assumption of negligible attenuation in rain only holds for the rather long wavelength of the radar wind profiler.
- L42f: The list lacks methodologies, where ground-based radars are cross calibrated with collocated ground-based radars, e.g. Hogan et al 2000, Williams 2012, Kneifel et al 2015, and Radenz et al 2018; Also, for the disdrometer comparison, the more recent work of Myagkov 2020 could be cited.
- L93: As the scope of the paper is quite technical, it might interest the reader what components where changed following the two hardware failures.
- L199/Fig 3: At first it was confusing, how the algorithm decides, which of the two peaks is the valid one. Only in Sec 3.1.3 L265f, it becomes clear, that the moments of both peaks are calculated and the ‘correct’ one is only selected later. Please clarify in the text.
- L232: The definition of coherence seems quite vague. It should be stated, that the phase difference of the signals has to vary slowly enough.
- Fig 5: It seems that the noise oscillates with season before 2017, but ceases to do so after the hardware change. Have you looked into that issue and is it related to hardware temperature stability?
- L311: Is -1min time lag typical? Could you identify any dependence on horizontal wind?
- Fig 8: It would help to understand the case study, if the 2DVD derived drop size distribution and rain rate would be shown in the figure as well.
Also the limits of the colormap should be similar in Fig 8a and Fig 9c. - L392: At least the mean relative offset and the standard deviation should also be given for the wind mode beams. One would suspect, that Beam V shows the least offset, as it is pointed vertically as well? Do the offsets change over time?
- L418: From Fig 12 the impression arises, that the calibration was stable until early 2013, then changing rapidly until 2014, afterwards being more stable until mid-2015. This would give a change of about 13dB/year, but other years being more stable.
- Code availability: For long-term availability of the source code, please consider also submitting it to a permanent storage, such as zenodo.
Technical issues
- L103: “Pulse duration () [ns]” unnecessary set of brackets
- L533: “[…] on the DOE ARM archive as a PI Product […]”
References
- Myagkov, A., Kneifel, S., and Rose, T.: Evaluation of the reflectivity calibration of W-band radars based on observations in rain, Atmos. Tech., 13, 5799–5825, https://doi.org/10.5194/amt-13-5799-2020, 2020.
- Williams, C. R.: Vertical Air Motion Retrieved from Dual-Frequency Profiler Observations, J. Atmos. Ocean. Tech., 29, 1471–1480, https://doi.org/10.1175/JTECH-D-11-00176.1, 2012
- Radenz, M., Bühl, J., Lehmann, V., Görsdorf, U., and Leinweber, R.: Combining cloud radar and radar wind profiler for a value added estimate of vertical air motion and particle terminal velocity within clouds, Atmos. Tech., 11, 5925–5940, https://doi.org/10.5194/amt-11-5925-2018, 2018.
- Hogan, R. J., Illingworth, A. J., and Sauvageot, H.: Measuring Crystal Size in Cirrus Using 35- and 94-GHz Radars, J. Atmos. Ocean. Tech., 17, 27–37, https://doi.org/10.1175/1520-0426(2000)017<0027:MCSICU>2.0.CO;2, 2000
- Kneifel, S., von Lerber, A., Tiira, J., Moisseev, D., Kollias, P., and Leinonen, J.: Observed Relations between Snowfall Microphysics and Triple-Frequency Radar Measurements, J. Geophys. Res.-Atmos., 120, 6034–6055, https://doi.org/10.1002/2015JD023156, 2015
Citation: https://doi.org/10.5194/egusphere-2022-1405-RC3 - AC3: 'Reply on RC3', Christopher Williams, 21 Mar 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1405', Anonymous Referee #1, 12 Feb 2023
Review of:
“Calibrating Radar Wind Profiler Reflectivity Factor using Surface Disdrometer Observations”,
submitted to the Atmospheric Measurement Techniques (AMT) Journal.This manuscript presents a methodology to calibrate radar wind profiler reflectivity factor using disdrometer data collected during rain events.
The authors found that the ageing of the hardware over time greatly impacts the RWP sensibility as it produces smaller values of SNR. The authors proposed compensating for this by varying the calibration constant and several other methods clearly described and evaluated using long-term data sets.It is great to see that the data sets and the code are available (or will be in the near future), thus increasing the reproducibility of the proposed methodology.
I reckon this manuscript totally fits the scope of the AMT journal, and it is an excellent contribution to the radar community.
I have minor comments and suggestions that may help improve the paper.
1. The Spectrum adjustment methods and the adjustments due to Nyquist velocity aliasing, coherent integration filtering, and increased noise power will benefit from a diagram. A flowchart describing the critical aspects of the proposed algorithm could help depict the method more clearly.
2. The sensitivity of the radar plays an important role in the calibration procedure explained in this manuscript. If the RWP cannot collect data during rain events, I reckon this method will not be valid for calibration. I suggest the authors add a discussion regarding this or if some other meteorological target could be used instead.
Citation: https://doi.org/10.5194/egusphere-2022-1405-RC1 - AC1: 'Reply on RC1', Christopher Williams, 21 Mar 2023
-
RC2: 'Comment on egusphere-2022-1405', Anonymous Referee #2, 27 Feb 2023
The manuscript suggests a very interesting methodology to calibrate UHF radar wind profiler data with the help of surface distrometer observations. Getting radar reflectivity values from the hitherto uncalibrated power measurements would clearly increase the usefulness these instruments. This approach furthermore provides an opportunity for a long-term monitoring of the hardware status, which offers very welcome insights as nicely demonstrated by the authors. There are a couple of points that the should be addressed in a revised version of the manuscript:
As the authors describe, it is known that excessive broadening of Doppler spectra during precipitation leads with neccessity to an incorrect estimation of the noise level by the Hildebrand and Sekhon (1974) method, because this broadening often affects the complete so-called "full-scale" velocity interval (which is bounded by the Nyquist velocity), so that the prerequisites inherent in the noise estimation algorithm are violated. Of course it is possible to address this issue by adjusting the radar settings to allow for a much wider Nyquist limit, in particular by reducing the number of coherent integrations and increasing the length of the time series for the FFT. However, this can not be done a-posteriori.
The authors then assert that the decrease of the signal-to-noise ratio (due to the spectral broadening described above) is due to signal power leakage of the Discrete Fourier Transform (FFT) calculation. Spectral leakage due to the finite extent of time series (in the case of RWP coherently integrated I/Q data) is a well-known fact in Fourier transforms and it is usually controlled through the multiplication of the time series with some kind of window function. However, the authors unfortunately do not provide information on what particular window function was used in the ARM RWP systems. The Vaisala LAP-3000 RWP traditionally use a "von Hann" (or Hanning) window since this operation was quite easy to apply in frequency domain, but other options were possible in later versions of the Vaisala software. The authors should provide this additional detail.
However, spectral leakage of the Fourier transform is not the only explanation for the broad Doppler spectra observed with RWP during precipitation: Tests with a similar radar have shown that the broadening of the Doppler spectra is rather independent of the window function used in RWP signal processing, at least as long as a reasonable selection like the Hanning window is made. This observation is in contradiction to the assertion that the broadening is due to spectral processing. Given the dynamic range of the RWP receiver it therefore appears to be more likely that the broadening is caused by signal contributions from the antenna sidelobes. The authors should comment on this alternative explanation, even though it has no consequences for the correction methodology presented.
A few other minor remarks:
Line 16/17: "The third step increases the signal-to-noise ratio (SNR) due to signal power leakage during the Fast Fourier Transform (FFT) calculation" - should be reformulated for more clarity, as the broadening (regardless whether this is due to the antenna sidelobes, or due to spectral leakage) leads to an increase of the noise level and thus a decrease of SNR.
Line 34: "At the radar measurement level, radars measure the return signal power as a function of range" - suggest a different wording
Line 37 "every radar subsection" - suggest radar hardware component
Line 39 "pole mounted corner reflector calibrations" - this method is impractical for the rather large and fixed phased array antennas of the RWP and does not need to be mentioned here
Line 57/58: "to account for radio frequency interference (RFI) that sporadically increases noise power estimates" - The effects of RFI remain largely unclear. Therefore the authors should provide more details on the characteristics of this particular RFI contamination. Not every RFI signal increases the noise power.
Line 96 (Table 1 Pertinent RWP operating parameters): For sake of completeness the authors should also provide an short overview of the RWP signal processing algorithms used. Was there any kind of time-domain nonlinear filtering applied ? Was the spectral integration done using Merritts (1995) statistical averaging method? Which moment estimation algorithm was used - single peak picking or multiple peak picking?
Line 243 "The impact of the TDA low-pass filter can be mitigated by applying a correction factor.." The authors could perhaps add a reference to Wilfong et al. (1999) "Optimal Generation of Radar Wind Profiler Spectra" which further discusses the TDA filter characteristics.
Line 277/278: "As the signal power magnitude increases, the FFT leakage causes the spectrum noise power to increase above the noise power produced by other radar noise sources". This statement is rather awkward and should be reformulated, also in view of the remarks made above.
Line 312 plus multiple other occasions: "Parson’s correlation coefficient" should be corrected to Pearson’s correlation coefficient
Line 645: The given link "https://github.com/ChristopherRWilliams/rwp/Python/spectra" is incorrect. The correct link is obviously https://github.com/ChristopherRWilliams/RWP_Python_moments
Citation: https://doi.org/10.5194/egusphere-2022-1405-RC2 - AC2: 'Reply on RC2', Christopher Williams, 21 Mar 2023
-
RC3: 'Comment on egusphere-2022-1405', Anonymous Referee #3, 06 Mar 2023
Williams et al present a methodology for long-term calibration of UHF radar wind profiles using collocated surface disdrometer observations. They apply their method to an 8-year dataset collected at ARMs Southern Great Planes site. The approach is technically sound and the manuscript is well written, hence I recommend the study for publication after very few minor corrections. The open availability of source code and data already during the discussion phase has to be highlighted positively.
Comments
- L32: It should be stated more clearly, that the assumption of negligible attenuation in rain only holds for the rather long wavelength of the radar wind profiler.
- L42f: The list lacks methodologies, where ground-based radars are cross calibrated with collocated ground-based radars, e.g. Hogan et al 2000, Williams 2012, Kneifel et al 2015, and Radenz et al 2018; Also, for the disdrometer comparison, the more recent work of Myagkov 2020 could be cited.
- L93: As the scope of the paper is quite technical, it might interest the reader what components where changed following the two hardware failures.
- L199/Fig 3: At first it was confusing, how the algorithm decides, which of the two peaks is the valid one. Only in Sec 3.1.3 L265f, it becomes clear, that the moments of both peaks are calculated and the ‘correct’ one is only selected later. Please clarify in the text.
- L232: The definition of coherence seems quite vague. It should be stated, that the phase difference of the signals has to vary slowly enough.
- Fig 5: It seems that the noise oscillates with season before 2017, but ceases to do so after the hardware change. Have you looked into that issue and is it related to hardware temperature stability?
- L311: Is -1min time lag typical? Could you identify any dependence on horizontal wind?
- Fig 8: It would help to understand the case study, if the 2DVD derived drop size distribution and rain rate would be shown in the figure as well.
Also the limits of the colormap should be similar in Fig 8a and Fig 9c. - L392: At least the mean relative offset and the standard deviation should also be given for the wind mode beams. One would suspect, that Beam V shows the least offset, as it is pointed vertically as well? Do the offsets change over time?
- L418: From Fig 12 the impression arises, that the calibration was stable until early 2013, then changing rapidly until 2014, afterwards being more stable until mid-2015. This would give a change of about 13dB/year, but other years being more stable.
- Code availability: For long-term availability of the source code, please consider also submitting it to a permanent storage, such as zenodo.
Technical issues
- L103: “Pulse duration () [ns]” unnecessary set of brackets
- L533: “[…] on the DOE ARM archive as a PI Product […]”
References
- Myagkov, A., Kneifel, S., and Rose, T.: Evaluation of the reflectivity calibration of W-band radars based on observations in rain, Atmos. Tech., 13, 5799–5825, https://doi.org/10.5194/amt-13-5799-2020, 2020.
- Williams, C. R.: Vertical Air Motion Retrieved from Dual-Frequency Profiler Observations, J. Atmos. Ocean. Tech., 29, 1471–1480, https://doi.org/10.1175/JTECH-D-11-00176.1, 2012
- Radenz, M., Bühl, J., Lehmann, V., Görsdorf, U., and Leinweber, R.: Combining cloud radar and radar wind profiler for a value added estimate of vertical air motion and particle terminal velocity within clouds, Atmos. Tech., 11, 5925–5940, https://doi.org/10.5194/amt-11-5925-2018, 2018.
- Hogan, R. J., Illingworth, A. J., and Sauvageot, H.: Measuring Crystal Size in Cirrus Using 35- and 94-GHz Radars, J. Atmos. Ocean. Tech., 17, 27–37, https://doi.org/10.1175/1520-0426(2000)017<0027:MCSICU>2.0.CO;2, 2000
- Kneifel, S., von Lerber, A., Tiira, J., Moisseev, D., Kollias, P., and Leinonen, J.: Observed Relations between Snowfall Microphysics and Triple-Frequency Radar Measurements, J. Geophys. Res.-Atmos., 120, 6034–6055, https://doi.org/10.1002/2015JD023156, 2015
Citation: https://doi.org/10.5194/egusphere-2022-1405-RC3 - AC3: 'Reply on RC3', Christopher Williams, 21 Mar 2023
Peer review completion
Journal article(s) based on this preprint
Model code and software
RWP Python Moment Christopher Williams https://github.com/ChristopherRWilliams/RWP_Python_moments/
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 240 | 98 | 19 | 357 | 3 | 3 |
- HTML: 240
- PDF: 98
- XML: 19
- Total: 357
- BibTeX: 3
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 160 | 45 |
| Germany | 2 | 55 | 15 |
| China | 3 | 26 | 7 |
| Russia | 4 | 12 | 3 |
| France | 5 | 11 | 3 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 160
Christopher R. Williams
Joshua Barrio
Paul E. Johnston
Paytsar Muradyan
Scott E. Giangrande
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2925 KB) - Metadata XML