An improved hydrometeor detection method for millimeter-wavelength cloud radar

Ge, Jinming; Zhu, Zeen; Zheng, Chuang; Xie, Hailing; Zhou, Tian; Huang, Jianping; Fu, Qiang

doi:https://doi.org/10.5194/acp-17-9035-2017

Articles | Volume 17, issue 14

https://doi.org/10.5194/acp-17-9035-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/acp-17-9035-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 17, issue 14

Research article

|

27 Jul 2017

Research article |

| 27 Jul 2017

An improved hydrometeor detection method for millimeter-wavelength cloud radar

Jinming Ge, Zeen Zhu, Chuang Zheng, Hailing Xie, Tian Zhou, Jianping Huang, and Qiang Fu

Download

Final revised paper (published on 27 Jul 2017)
Preprint (discussion started on 08 Dec 2016)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review', Anonymous Referee #1, 11 Jan 2017
- AC1: 'Response to Anonymous Referee #1', Jinming Ge, 04 Mar 2017
RC2: 'Review: “An Improved Hydrometeor Detection Method for Millimeter-Wavelength Cloud radar” by Ge et al.', Anonymous Referee #2, 13 Jan 2017
- AC2: 'Response to Anonymous Referee #2', Jinming Ge, 04 Mar 2017
RC3: 'Review', Anonymous Referee #3, 15 Jan 2017
- AC3: 'Response to Anonymous Referee #3', Jinming Ge, 04 Mar 2017

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Jinming Ge on behalf of the Authors (05 Mar 2017) Author's response Manuscript

ED: Referee Nomination & Report Request started (12 Mar 2017) by Xiaohong Liu

RR by Anonymous Referee #1 (26 Mar 2017)

Suggestions for revision or reasons for rejection

The manuscript is in need of some thorough editing: there are numerous typos
and poor grammar that makes the text difficult to follow at times. As
mentioned below, the text describing figures is often confusing, making it
difficult to determine what results are actually being shown. Many of my
original comments have not been adequately addressed by the authors:

I again encourage the authors not to compare to the old MMCR cloud
mask. ARM's new KAZR has new processing algorithms. Therefore, the authors
are comparing to the cloud mask used for the old MMCR but applied to the KAZR
instead. This limits the usefulness of their results. Several times the
authors mention that their method is an improvement over ARM's operation
method. If the authors keep their analysis as is, statements that claim
improvement over ARM's operation method need to be removed since ARM does not
produce data with the MMCR algorithms applied to the KAZR.

The authors have not made a convincing argument that the increased detection
around 1.5 and 2km in Figure 7 in the cloud mask is dust. Their argument is
that the MPL backscatter (what are the units for the MPL backscatter?) and
depolarization is large where they detect dust with the KAZR. However, there
appears to be plenty of pixels with just as large backscatter and
depolarization that is not detected in the KAZR mask. To me, these KAZR
detections appear to be false positives and are therefore undesirable. It
would be clearer to show the PDFs of MPL aerosol backscatter and
depolarization for both the pixels identified in the KAZR cloud mask and
those not. If these PDFs don't differ significantly, then the authors need to
revise their approach to avoid these false positives.

The extended comparison to the MPL is helpful for understanding the accuracy
of the author's method. However, it appears in Figure 9 that comparisons are
made to the MPL detection of both cloud and aerosols. In this case, any
increased detection in the lowest several kilometers will also be detected by
the MPL since aerosol is always present there. Above those altitudes, the
results are not good with about 20% of the cloud mask being false positives!
Figure 9 would be more useful for assessing accuracy if the fraction of the
increased detections identified by the MPL as cloud was shown.

Figure 9 also does not show the opposite error: false negatives:
i.e. detected by the MPL but not the KAZR cloud mask. In lieu of Figure 9 it
would be more helpful to just show a confusion matrix (similar to what the
authors present in Table 1) for the MPL cloud mask compared to both the KAZR
cloud mask both with and without the noise reduction step. That way the
change in both error rates could be assessed.

I cannot tell the line styles apart in the top panel of Figure 8. I'm also
unsure what is being plotted there as the caption mentions a "confusion
matrix" but no matrix is given. It is also not clear what the lines in the
bottom panels of Figure 8 are normalized to. It would be most useful to show
the comparison of cloud occurrence profiles (i.e. number of cloudy pixels /
total number of pixels), but that doesn't appear to be what is plotted here.

Hide

ED: Reconsider after major revisions (26 Mar 2017) by Xiaohong Liu

AR by Jinming Ge on behalf of the Authors (06 May 2017) Author's response Manuscript

ED: Referee Nomination & Report Request started (22 May 2017) by Xiaohong Liu

ED: Publish subject to technical corrections (28 Jun 2017) by Xiaohong Liu

AR by Jinming Ge on behalf of the Authors (03 Jul 2017) Manuscript

Short summary

A modified method with a new noise reduction scheme that can reduce the noise distribution to a narrow range is proposed to distinguish clouds and other hydrometeors from noise and recognize more features with weak signal in cloud radar observations. It was found that our method has significant advantages in reducing the rates of both failed negative and false positive hydrometeor identifications in simulated clouds and recognizing clouds with weak signal from our cloud radar observations.