Parameterization of wind evolution using lidar

Chen, Yiyin; Schlipf, David; Cheng, Po Wen

doi:https://doi.org/10.5194/wes-6-61-2021

Articles | Volume 6, issue 1

https://doi.org/10.5194/wes-6-61-2021

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

https://doi.org/10.5194/wes-6-61-2021

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

Articles | Volume 6, issue 1

Research article

|

12 Jan 2021

Research article |

| 12 Jan 2021

Parameterization of wind evolution using lidar

Yiyin Chen, David Schlipf, and Po Wen Cheng

Download

Final revised paper (published on 12 Jan 2021)
Preprint (discussion started on 17 Mar 2020)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Referee comment', Felix Kelberlau, 19 Mar 2020
RC2: 'Referee comment', Anonymous Referee #2, 20 Apr 2020
RC3: 'review of ``Parameterization of Wind Evolution using Lidar'' : major revisions needed', Mark Kelly, 20 Apr 2020
AC1: 'Author final response', Yiyin Chen, 24 Jun 2020

Peer-review completion

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

AR by Yiyin Chen on behalf of the Authors (23 Jul 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (19 Aug 2020) by Jakob Mann

RR by Mark Kelly (07 Sep 2020)

Suggestions for revision or reasons for rejection

This draft is much improved, after many changes addressing the 3 reviewers' comments.
A few comments remain:

l.39: mention that Ri is a stability measure, somehow--this is the first time that stability comes up.

l.47 replace "adequately" with perhaps 'thoroughly' and/or 'directly'.

Regarding this reviewer's comment about the \Delta t dependence already inherent in a, from the equations you use, where the review includes a derivation of such dependence: your reply unfortunately appears to indicate some confusion around the implied dependence. You respond with an explanation that "removing" the \Delta t in dimensionless frequency gives a new dependence of (\Delta t)^(0.51). But the implied dependence is still like eqn.(4) in this reviewer's previous review; shouldn't such "removal" give a dependence that looks like (\Delta t)*(\Delta t)^(-0.49) = (\Delta t)^(0.51), according to the review's eqn.(4)?. If you have an implied dependence which is clear, then why "guess" at the exponent later in section 4.2? At very least, a clearer restatement of your response seems to be needed, and included in the article.

l.299-302: eqn.(18) uses Taylor's hypothesis to replace the spatial correlation with temporal correlation. But since this is potentially equivalent to no evolution (the spatial and temporal spectra differ, no?), then you have a conundrum here...using eqn.18 implies ignoring evolution. I think you need a different argument for picking only one of L or T; at any rate your argument in these lines doesn't quite hold. Why not include the text from your response ("from measured data, we have no alternatives"), and an argument as to how L and T are similar?

l.305: "L might contain uncertainties" ... ---your eqn.18 is not correct, if there is evolution; see above comment.

l.313-316: if the amount of sampling noise in mu3 or mu4 is very large, how is it "still worth investigating"? ML can't magically connect things if there is too much noise per trend/pattern; perhaps you can rephrase in terms of expected uncertainty, or at least in terms of (re-)confirming the expectation found in e.g. Lenschow et al.(1994). In the end the results appear to show that mu3 and mu4 are maybe carrying useful information; perhaps that could somehow be leveraged (or compare to the estimated uncertainty).

l.378-380: in your response you clarify that "β can also be understood as the weight vector of h(x). But we’ve defined w as a weight vector before, we wanted to avoid using the same word in case reader might confuse these two different processes". Why not incorporate this in the updated draft version, to help the reader understand?

l.385-386: I understand that within the machine-learning context you are using the ML term 'characteristic length scale', but this is not a ML journal. I re-iterate my statement about lines 289-296 in v3; why not help the reader (with my something like my previous comment in parenthesis, or a footnote)?

Some of the references are still lacking information, such as conference name, journal, or year. Further, some in-text citations are also lacking the year, and in sum cases initials are included in these citations; i.e., they do not conform to the referencing standard used here, and are not consistent throughout the paper.

Hide

ED: Publish subject to minor revisions (review by editor) (12 Sep 2020) by Jakob Mann

AR by Yiyin Chen on behalf of the Authors (05 Oct 2020) Author's response Manuscript

ED: Publish subject to technical corrections (26 Oct 2020) by Jakob Mann

ED: Publish subject to technical corrections (28 Oct 2020) by Carlo L. Bottasso (Chief editor)

AR by Yiyin Chen on behalf of the Authors (28 Oct 2020) Manuscript

Short summary

Wind evolution is currently of high interest, mainly due to the development of lidar-assisted wind turbine control (LAC). Moreover, 4D stochastic wind field simulations can be made possible by integrating wind evolution into 3D simulations to provide a more realistic simulation environment for LAC. Motivated by these factors, we investigate the potential of Gaussian process regression in the parameterization of a two-parameter wind evolution model using data of two nacelle-mounted lidars.