Atmospheric condition identification in multivariate data through a  metric for total variation

Hamilton, Nicholas

doi:https://doi.org/10.5194/amt-13-1019-2020

Articles | Volume 13, issue 2

https://doi.org/10.5194/amt-13-1019-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-13-1019-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 2

Research article

|

03 Mar 2020

Research article |

| 03 Mar 2020

Atmospheric condition identification in multivariate data through a metric for total variation

Nicholas Hamilton

Download

Final revised paper (published on 03 Mar 2020)
Preprint (discussion started on 11 Sep 2019)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review', Ásta Hannesdóttir, 12 Nov 2019
- AC1: 'Response to Comments from Reviewer 1', Nicholas Hamilton, 24 Dec 2019
RC2: 'Review (discussion phase)', Mark Kelly, 25 Nov 2019
- AC2: 'Response to Comments from Reviewer 2', Nicholas Hamilton, 24 Dec 2019

Peer-review completion

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

AR by Nicholas Hamilton on behalf of the Authors (24 Dec 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (06 Jan 2020) by Laura Bianco

RR by Ásta Hannesdóttir (19 Jan 2020)

RR by Anonymous Referee #2 (21 Jan 2020)

Suggestions for revision or reasons for rejection

The article has improved somewhat upon revision, though there are still some things that need to be addressed; I list them here.
First, the reason that the Hilbert-transform (Hristov) method was suggested, was because it does _not_ require a periodic reference signal. The author has now included references to this method and application, but however incorrectly followed with "Previous use of the reference-signal method (Kelly et al., 2009) required the use of a periodic reference signal, which does not lend itself easily to the detection of non-periodic atmospheric events".

The author gives a nice response to the reviewer's question about the difference between _V_ and similar measures found in the literature. He responds including distinguishing between total variability, overall variability, and summative or pooled variance; he also writes about generalized variance and its relation. However, the revision (end of §1, pg.3) doesn't include such, only mentioning that V "is closely related to the generalized variance as per Wilks (1932); Sengupta (2004), but is distinct in that it is applied to a collection of variables rather than a multi-dimensional vector." To help the reader understand, this response should be better integrated into the article (some is on p.8).

The addition in section 3 (lines 5-10, p.5) is somewhat problematic.
Line 5: need close-parens after 'Fig.2'.
Pg.5, line6: the added text "Steady-state wake models..." should be precedded by 'E.g.,'. Why introduce this sentence? It's a bit awkward, and also some 'high-fidelity models' do not assume unchanged speed or direction.
Following that sentence, the author states that effective validation requires stationarity or that "observations represent a dynamic event of interest", but simple measures to deal with nonstationarity (e.g. detrending) are also sometimes used.

p.7, line 24: is it literally normalization by "respective span" (max minus min), or by respective variance?
p.8, line 20: '(' should preceed 'Ruxton'.
p.16, l.18-20: the DLC's of the IEC 61400-1 do not fit actual events, but are mostly calibrated statistically for the sake of loads, based on expected joint occurence of various events and scenarios (e.g. Hannesdottir & Kelly, Dimitrov et al).

Hide

ED: Publish subject to technical corrections (21 Jan 2020) by Laura Bianco

AR by Nicholas Hamilton on behalf of the Authors (30 Jan 2020) Author's response Manuscript

Short summary

The identification of atmospheric conditions within a multivariable atmospheric data set is an important step in validating emerging and existing models used to simulate wind plant flows and operational strategies. The total variation approach developed here offers a method founded in tested mathematical metrics and can be used to identify and characterize periods corresponding to quiescent conditions or specific events of interest for study or wind energy development.