Causal dependences between the coupled ocean–atmosphere dynamics over the tropical Pacific, the North Pacific and the North Atlantic

Vannitsem, Stéphane; Ekelmans, Pierre

doi:https://doi.org/10.5194/esd-9-1063-2018

Articles | Volume 9, issue 3

https://doi.org/10.5194/esd-9-1063-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/esd-9-1063-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 9, issue 3

Research article

|

27 Aug 2018

Research article |

| 27 Aug 2018

Causal dependences between the coupled ocean–atmosphere dynamics over the tropical Pacific, the North Pacific and the North Atlantic

Stéphane Vannitsem and Pierre Ekelmans

Download

Final revised paper (published on 27 Aug 2018)
Preprint (discussion started on 15 Jan 2018)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Referee comment', Anonymous Referee #1, 27 Jan 2018
- AC1: 'Response to the Reviewer's comments', Stéphane Vannitsem, 30 Mar 2018
RC2: 'Review of Causal dependences between the coupled ocean-atmosphere dynamics over the Tropical Pacific, the North Pacific and the North Atlantic', Anonymous Referee #2, 11 Mar 2018
- AC2: 'Response to the Reviewer's comments', Stéphane Vannitsem, 30 Mar 2018

Peer-review completion

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

ED: Reconsider after major revisions (17 Apr 2018) by Rui A. P. Perdigão

AR by Stéphane Vannitsem on behalf of the Authors (11 May 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (22 May 2018) by Rui A. P. Perdigão

RR by Anonymous Referee #3 (13 Jun 2018)

Suggestions for revision or reasons for rejection

- Generic comments

The author responses and revision have failed to adequately address the most critical concerns. On the contrary. By refusing to properly tackle important limitations of the work with counter-argumentation devoid of merit, the authors preclude any chance for me to recommend publication of this manuscript.

- Causality

It is appalling that not even with the clear detailed explanations from the editor and one reviewer the authors could take the dignifying step to actually amend their work. Rather, the authors took an arrogant posture, insisting on the falsehood that CCM captures causality (it does not) by blindly following physically questionable studies without fundamental proof and at odds with physical evidence.

Scientifically apt readers would never be confused by framing the study around dynamic codependence, because that is what CCM honestly measures, and the concept is familiar to scientists with proper understanding of physics.

Confusion would rather arise from claiming to study causal links when no causality is truly investigated here. The editor is right and the title is indeed misleading.

The authors must understand that insisting on claiming causal links in this study is completely unacceptable in scientific terms as it entails fraudulent use of the "causal" term.

The authors further invoke Granger causality, which is purely statistical (no physics in there) to assess inferential ability between normally distributed random variables. Granger was an inspirational person, with a Nobel Prize in Economics (not Physics!) whose work is so misused as if it would have anything remotely to do with physics. It does not.

Causality, as everybody with some background in Physics should know, is essentially about cause-effect relationships, something that the CCM method does not provide, and that the authors seem completely misinformed about.

As noted in the previous reports, shared membership on a dynamical system proves no causality in the physical sense. There is no point in presenting a "clear definition of causality" that is completely wrong. So the authors have lost their time in investing in CCM for causal investigations.

- Correlations and MSE as validators

The stubborn use of deprecated correlation-based verification metrics in weather forecasting does not make them appropriate for the present problem. Validation via simple linear regressions and simplistic mean square error metrics is extremely crude, irrespective of how popular is still is among some applied fields languishing for decades behind approaches that are already standard in advanced disciplines.

Much more sophisticated metrics e.g. in nonlinear statistics are already available and widely used across a diverse scientific spectrum. The work would have had a lot to gain in robustness and credibility by improving their practices in this regard. The used metrics are simply unacceptable.

- Euclidean metrics

The mapping between the system orbits in phase space and an Euclidean tangent space is not valid unless the appropriate smoothness conditions are observed e.g. if the dynamics are an actual manifold - which is not the case in this data-based analysis. The author's arguments naïvely defending the use of Euclidean metrics are without substance and make one wonder whether the authors even know what a manifold is.

Again here, the classic references used by the authors talk about phase space manifolds without any regard to how to map the discrete records e.g. of data to the continuous (and even more critically) smoothly differentiable structure that is required to form a manifold.

- Phase and state spaces

The authors corrected the manuscript in this regard, albeit reluctantly.
I would have rested my case in that sense. However, the authors still believe that they are right and the editor has "forced" and "insisted" on it.

Reading the editorial comments, I saw no such insistence, but rather a clean request for correction that is perfectly natural in a professional setting. I would have been happy to have such constructive and helpful remarks to improve my work if I were the authors.

Yet the authors responded in a most unprofessional manner - to both the editor and the reviewer 2.

Well, I will reiterate their points: The notion that state and phase terminology can be used interchangeably as argued by the authors is completely wrong, and it is not because it appears in some classic books that that makes it correct. Faith in the books cited by the authors and subsequent ill-informed literature is not science: it is religion.

Especially so when all the rigorous literature in Physics says otherwise. Just open any proper Physics textbook, reference book or dictionary in Physics to see what reviewer 2 and the editor mean. I stress the term: Physics. Not literature in applied mathematics, nonlinear geophysics, chemistry, biology or other applied fields with shaky understanding of fundamental physics.

Taking the tendencies on the states will not solve the problem: a N-dimensional state space will be replaced by a N-dimensional conjugate space, but neither is the 2N-dimensional phase space where BOTH states and conjugate momenta reside.
So reviewer 2 and the editor are still right and the authors still wrong.

- Attitude

I must say that it is outrageous how the authors responded to the raised concerns, dismissing the most serious concerns without due reflection, and taking a hostile defensive attitude towards both reviewer 2 and the editor, who had made perfectly reasonable and wise remarks.

The authors should have paid more attention to the raised concerns, which were very pertinent and thoroughly stated. Until the authors would actually address the most fundamental problems of their work, I could never recommend publication of this manuscript at Earth System Dynamics.

Earth System Dynamics is not a journal of statistical applications. Dynamics call for a physical basis. And neither Granger causality nor the CCM kinematic-geometric assessment have anything to do with physical causality.

Overall, this manuscript is fundamentally playing a disservice to serious science.
It is inadmissible to further litter the literature with poor science that had its chance to be corrected, but ended up not being so out of sheer and pure arrogance.

My final advice: Prove each and every step of the formulation in a physically sound manner, and revisit the pertinent geophysical problem with proper causal inference tools.

As I do not believe in the authors' willingness to take any serious recommendation on board, I hereby recommend rejection. The authors can then proceed to a more gullible venue.

My apologies to the Journal for the strongly reprimanding tone, but frankly from the authors’ attitude, that is the only language that the authors will understand.

Hide

ED: Publish subject to minor revisions (review by editor) (09 Jul 2018) by Rui A. P. Perdigão

AR by Stéphane Vannitsem on behalf of the Authors (26 Jul 2018) Author's response Manuscript

ED: Publish as is (07 Aug 2018) by Rui A. P. Perdigão

Short summary

The El Niño–Southern Oscillation phenomenon is a slow dynamics present in the coupled ocean–atmosphere tropical Pacific system which has important teleconnections with the northern extratropics. These teleconnections are usually believed to be the source of an enhanced predictability in the northern extratropics at seasonal to decadal timescales. This question is challenged by investigating the causality between these regions using an advanced technique known as convergent cross mapping.