Integrated water vapor and liquid water path retrieval using a single-channel radiometer

Billault-Roux, Anne-Claire; Berne, Alexis

doi:https://doi.org/10.5194/amt-14-2749-2021

Articles | Volume 14, issue 4

https://doi.org/10.5194/amt-14-2749-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Winter weather research in complex terrain during ICE-POP...

https://doi.org/10.5194/amt-14-2749-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 14, issue 4

Research article

|

08 Apr 2021

Research article |

| 08 Apr 2021

Integrated water vapor and liquid water path retrieval using a single-channel radiometer

Anne-Claire Billault-Roux and Alexis Berne

Download

Final revised paper (published on 08 Apr 2021)
Preprint (discussion started on 17 Aug 2020)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'amt-2020-311-reviewer-comments', Anonymous Referee #1, 13 Sep 2020
- RC3: 'Review for AMT-2020-311: Integrated water vapor and liquid water path retrieval using a single-channel radiometer by Anne-Claire Billault-Roux and Alexis Berne', Anonymous Referee #3, 22 Sep 2020
  - AC2: 'Responses to the reviewers', Anne-Claire Billault–Roux, 07 Dec 2020
- AC1: 'Responses to the reviewers', Anne-Claire Billault–Roux, 07 Dec 2020
RC2: 'Comments to AMT-2020-311', Anonymous Referee #2, 15 Sep 2020
- AC3: 'Responses to the reviewers', Anne-Claire Billault–Roux, 07 Dec 2020

Peer-review completion

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

AR by Anne-Claire Billault–Roux on behalf of the Authors (07 Dec 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (08 Dec 2020) by GyuWon Lee

RR by Anonymous Referee #1 (26 Dec 2020)

RR by Anonymous Referee #3 (07 Jan 2021)

Suggestions for revision or reasons for rejection

General:

I like how the authors have improved the manuscript – specifically concerning a more detailed discussion of the results. I do have some remaining further general and specific points, which I would like the authors to bring forward more clearly.

For my point of view, the authors don’t make it clear enough, why 89 GHz TB impacts the LWP retrieval more than IWV compared to the non-radiometric observations. To that extent it is important to note that LWP and atmospheric water vapor (mainly IWV and to lesser extent the profile shape) are basically the only contributors to the TB signal at 89 GHz. Since you can’t retrieve two largely independent parameters (IWV and LWP) from only one measurement (89 GHz TB), you need to consider additional information that constrains the retrieval. The additional information the authors have (correctly) chosen (e.g. near-surface temperature & humidity, geographical location and altitude) is mostly correlated to IWV and hardly to LWP. This is the actual reason why LWP is more impacted by the 89 GHz TB: the additional information constrains the IWV and so the 89 GHz information can be used to retrieve the LWP. This needs to be made more clear. The authors, in contrast, also argument (see abstract, conclusions and within the main text) with a lower sensitivity of 89 GHz TB to IWV compared to LWP. This is a bit like comparing apples with oranges. If the authors would like to use this argument, they need to explicitly state the sensitivities (in TB/kgm-2) but these numbers then need to be related to the absolute variability of IWV, respectively LWP which again are not directly comparable.

Error analysis: could the authors explain why they use R2 and not just R? R2 is often termed “explained variance”. Also, I assume the RMSE is calculated in a way that includes the bias? Including information how RMSE is calculated would be great. I’m a bit puzzled about the “relative error” that is given for LWP and IWV. How was this calculated? E.g. the authors claim that the “relative error” is 7.2% for LWP values larger than 30 gm-2. But, if I look at Fig. 5, the RMSE is about 50 gm-2 at a target value of 100 gm-2 and about 100 gm-2 at a target value of 500 gm-2 which I would interpret as relative uncertainties of 50%, respectively 20%. Please clarify.

And last but not least: the presented IWV retrieval results should be briefly discussed in the context of ground-based GPS receivers, which are a world-wide standard methodology for deriving IWV.
Specific points:

2.1 Radiosonde data set: please state how high a radiosonde ascent had to be in order to be used for NN training

3.1 Cloud liquid model: I see the ERA5 LWP comparison critical because, if I understand correctly, you assume the reanalysis LWP to be bias free. Could you comment on this in the text?

3.2 Radiative Transfer model: For completeness, please state which liquid water absorption model you have used. Also, I find the third paragraph difficult to follow and too lengthy. How can a LWC value correspond to a LWP value? This is dependent on cloud vertical extent. Can’t you just justify a threshold in LWP above which with X% probability you’d expect significant precipitation leading to an additional drop size and DSD dependency of your TB?

5.1.2 LWP algorithm: the authors write: “Let us highlight that in the case of a linear regression one would expect the error to diverge when high-order polynomials are included. This is not the case here because of the non-linear behavior of the neural network.” I don’t understand this argumentation. Doesn’t the inclusion of higher order terms in a linear regression actually lead to a retrieval improvement if the original dependencies between observation and target variable are non-linear? And shouldn’t the NN actually capture this non-linearity without having to include higher-order terms?

5.2 Instrument calibration: I’m missing a discussion why in Fig. 7 there is practically no dependency on TB-offset (even if TB-offset = 5 K, corresponding to a delta LWP of say 250 gm-2? Please check...) for the TB-only retrieval, however very well on the TB + additional information retrievals. I’m pretty sure this has to do with the combined IWV and LWP dependency which I already commented on in the beginning.

Hide

ED: Publish subject to minor revisions (review by editor) (17 Jan 2021) by GyuWon Lee

AR by Anne-Claire Billault–Roux on behalf of the Authors (15 Feb 2021) Author's response Author's tracked changes Manuscript

ED: Publish as is (16 Feb 2021) by GyuWon Lee

AR by Anne-Claire Billault–Roux on behalf of the Authors (22 Feb 2021) Author's response Manuscript

Short summary

In the context of climate studies, understanding the role of clouds on a global and local scale is of paramount importance. One aspect is the quantification of cloud liquid water, which impacts the Earth’s radiative balance. This is routinely achieved with radiometers operating at different frequencies. In this study, we propose an approach that uses a single-frequency radiometer and that can be applied at any location to retrieve vertically integrated quantities of liquid water and water vapor.