Stratospheric geoengineering impacts on El Niño/Southern Oscillation

Gabriel, C. J.; Robock, A.

doi:https://doi.org/10.5194/acp-15-11949-2015

Articles | Volume 15, issue 20

https://doi.org/10.5194/acp-15-11949-2015

© Author(s) 2015. This work is distributed under
the Creative Commons Attribution 3.0 License.

Special issue:

The Geoengineering Model Intercomparison Project (GeoMIP):...

https://doi.org/10.5194/acp-15-11949-2015

© Author(s) 2015. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 15, issue 20

Research article

|

27 Oct 2015

Research article |

| 27 Oct 2015

Stratospheric geoengineering impacts on El Niño/Southern Oscillation

C. J. Gabriel and A. Robock

Download

Final revised paper (published on 27 Oct 2015)
Preprint (discussion started on 27 Mar 2015)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC C1575: 'Review of "Stratospheric geoengineering impacts on el Nino / Southern Oscillation', Anonymous Referee #1, 14 Apr 2015
- AC C4964: 'Response to Referee #1', Corey Gabriel, 16 Jul 2015
RC C1626: 'Review of “Stratospheric geoengineering impacts on El Niño/Southern Oscillation” by C. J. Gabriel and A. Robock', Anonymous Referee #2, 16 Apr 2015
- AC C4965: 'Response to Referee #2', Corey Gabriel, 16 Jul 2015
RC C1661: 'Comments on Stratospheric geoengineering impacts on El Niño/Southern Oscillation', Anonymous Referee #3, 17 Apr 2015
- AC C4966: 'Response to Referee #3', Corey Gabriel, 16 Jul 2015

Peer-review completion

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

AR by Corey Gabriel on behalf of the Authors (16 Jul 2015) Author's response

ED: Referee Nomination & Report Request started (21 Jul 2015) by Lynn M. Russell

RR by Anonymous Referee #1 (27 Jul 2015)

Suggestions for revision or reasons for rejection

Whilst the authors have addressed a number of the more minor concerns satisfactorily, the article still has a number of fundamental flaws that were raised by myself and the other reviewers. The main improvements the authors have made have been to aggregate fewer experiments together and to apply a boot-strapping method to confirm that there are no statistically significant ENSO changes within their dataset. These changes and the others made to the manuscript, whilst useful do not address the following fundamental flaws.

Firstly, my previous comment 1) still stands. The authors have confirmed what I suspected, that the ENSO changes cannot be assessed with this dataset yet ignored the thrust of this criticism, i.e. what is the point of dedicating the entire results section to displaying a null result?

I suggested that other analysis be pursued in place of the many figures that show no change (Figures 3-9). Reviewer 2 made a similar suggestion that alternative analyses be made but I believe the authors misinterpreted our suggestions. Rather than suggesting that the authors use different analyses to assess ENSO, I was suggesting that the authors do different analyses of the Pacific Ocean in general, i.e. ones where concrete results would be forthcoming. The authors’ response that the SST response to SRM is well studied does not tally with my understanding of the literature; few studies assess the oceanic response to SRM, most do so only very briefly, and this is an area deserving of greater attention.

Secondly, in my previous points 2), 7) and 9) I noted that more should be done to assess and quantify what would be needed to detect changes in ENSO. These points were not addressed satisfactorily in the revision and as such the article would not help future researchers to progress on this issue. Beyond noting that it will be hard and that longer simulation times would help, little has been added. To make a useful contribution to this issue the authors should make a quantitative assessment of what would be required to detect changes in ENSO of various magnitudes.

In other words the article ought to have addressed a specific quantitative question along these lines: How long (or how many 100 year ensemble members) ought it to take to detect a 10% increase in ENSO frequency or average intensity in one model? Are the results different in different models? How do these figures relate to findings of observational studies?

In comment 8) I noted that there are longer pre-industrial simulations that the authors could use that could form the basis of such an assessment. This remains the case – all CMIP5 models ran extended pre-industrial spin-up simulations (i.e. the ones from which the historical simulations where launched) and these should be available on PCMDI. These would be a fantastic resource from which robust statistics of the behavior of ENSO in the models could be drawn and the authors ought to have utilized them.

Thirdly, the methodology of the piece remains unsatisfactory in general with many of the methods and analyses unexplained. Some example problems below:

I don’t believe the authors have adequately addressed the concern about treating results from models with different ENSO behavior as though they form part of the same set for their statistical tests. That Cai et al. 2015 used this method does not seem justification enough.

The approach used to generate the 90% confidence statements on page 31, line 1024 is unexplained. Are the authors using inter-member variance in their grand ensemble? Is it then appropriate to assume these are independent measures of some underlying sample?
Which years are used for 4xCO2? Is a linear fit appropriate?

Page 29, line 982: “we find reasonable agreement between all aggregated model data and the observations during all historical time intervals” – Not explained how this is found
Page 30, line 1001: “However, given only 90% confidence, the relatively large number of comparisons made and the agreement between cold event frequency and amplitude in those same comparisons, the result is most likely by chance.” – There is no justification for why this is the case.

In the discussion and conclusion section, the following statements are made which seem to indicate a misunderstanding of the nature of the statistical tests that have been applied:
• “We conclude that ENSO event frequency or amplitude in a geoengineered world will not differ from the observational record, historical model runs, or under global warming.” – surely the conclusion ought to be that any changes in ENSO are below the detection threshold of your approach (which has not been quantified).
• “The absence of a significant result, showing differences in ENSO variability between geoengineering and AGW may well be a valid finding.” – This seems to suggest that the changes were too small to detect in this study and may well be too small to detect in the real world but it’s unclear

Finally, the structure is very muddled with many of the sections containing text that seems wholly out of place. Additionally, where the authors have made efforts to revise their text, it reads as if they are responding to unasked questions (of course these are the reviewer comments but this would not be clear to the reader). The result is that the structure and the style of the piece are not of a quality suitable for publication.

Hide

RR by Anonymous Referee #2 (27 Jul 2015)

ED: Reconsider after major revisions (18 Aug 2015) by Lynn M. Russell

AR by Corey Gabriel on behalf of the Authors (11 Sep 2015) Author's response Manuscript

ED: Referee Nomination & Report Request started (16 Sep 2015) by Lynn M. Russell

RR by Anonymous Referee #2 (27 Sep 2015)

ED: Reconsider after minor revisions (Editor review) (12 Oct 2015) by Lynn M. Russell

AR by Corey Gabriel on behalf of the Authors (17 Oct 2015) Author's response

ED: Publish as is (20 Oct 2015) by Lynn M. Russell

AR by Corey Gabriel on behalf of the Authors (20 Oct 2015) Manuscript

Download

The requested paper has a corresponding corrigendum published. Please read the corrigendum first before downloading the article.

Article (1128 KB)
Full-text XML

Short summary

This is a first look at how the El Niño/Southern Oscillation (ENSO) might change under a regime of stratospheric geoengineering (GE). We find neither ENSO event frequency nor ENSO event amplitude will be different under various GE experiments than they would under unabated global warming. We additionally find substantial disagreement between models in depicting ENSO amplitude and frequency. Additionally, output from several GCMs is excluded from comparison due to unrealistic ENSO variability.