Climate SPHINX: evaluating the impact of resolution and stochastic physics parameterisations in the EC-Earth global climate model

Davini, Paolo; von Hardenberg, Jost; Corti, Susanna; Christensen, Hannah M.; Juricke, Stephan; Subramanian, Aneesh; Watson, Peter A. G.; Weisheimer, Antje; Palmer, Tim N.

doi:https://doi.org/10.5194/gmd-10-1383-2017

Articles | Volume 10, issue 3

https://doi.org/10.5194/gmd-10-1383-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/gmd-10-1383-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 10, issue 3

Model experiment description paper

|

31 Mar 2017

Model experiment description paper |

| 31 Mar 2017

Climate SPHINX: evaluating the impact of resolution and stochastic physics parameterisations in the EC-Earth global climate model

Paolo Davini, Jost von Hardenberg, Susanna Corti, Hannah M. Christensen, Stephan Juricke, Aneesh Subramanian, Peter A. G. Watson, Antje Weisheimer, and Tim N. Palmer

Download

Final revised paper (published on 31 Mar 2017)
Preprint (discussion started on 23 Jun 2016)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

SC1: 'Executive Editor Comment on ""Climate SPHINX: evaluating the impact of resolution and stochastic physics parameterisations in climate simulations', Astrid Kerkweg, 04 Jul 2016
- AC1: 'Replies to Comment', Paolo Davini, 28 Nov 2016
RC1: 'Reviewer's report on', Anonymous Referee #1, 16 Sep 2016
- RC2: 'Supplement regarding RC1', Anonymous Referee #1, 19 Sep 2016
  - AC2: 'Replies to Reviewer', Paolo Davini, 28 Nov 2016
RC3: 'Comments on Climate SPHINX: evaluating the impact of resolution and stochastic physics parameterisations in climate simulations', Anonymous Referee #2, 24 Oct 2016
- AC3: 'Replies to Reviewer #2', Paolo Davini, 28 Nov 2016

Peer-review completion

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

AR by Paolo Davini on behalf of the Authors (28 Nov 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (12 Dec 2016) by Wilco Hazeleger

RR by Anonymous Referee #1 (23 Jan 2017)

Suggestions for revision or reasons for rejection

The manuscript reports about the existence of a large scale computational effort with the goal of assessing the effect of resolution and stochastic physics represented by ECMWF's stochastic perturbation tendency (SPPT) and stochastic kinetic energy back skater (SKEB) schemes. While this is an enterprise that would be useful to both the climate and NWP communities and could potentially become a benchmark for future developments of stochastic parameterizations, the presentation is rather dry and follows a narrow focus, it is more of a technical report rather than a scientific publication.

Main comments:

1)Page 3, Line 10:"A key aim of the study is to investigate the degree to which
stochastic parametrisation schemes can be used as a computationally cheaper alternative to increased model resolution". This is very noble and indeed it is perhaps the most valid argument why the community should invest more in stochastic modeling. It would been good if the paper ended by turning back and stating a few conclusions regarding this point in the light of their study. The most unbalanced part of this work is the fact that from the beginning the authors announce three types of runs: Present day, future climate projections, and coupled runs but in the end they only discuss (some aspects of) the results of the present day runs. I am wondering what is the scientific value of the work or what kind of publication is meant to be if it is not simply a technical report then, if the impact of these efforts are not determined and assessed?

2)Page 3, line 30: what is the the EU-H2020 PRIMAVERA project? A reference or some explanation is needed.

3) Page 4, line 20: This paragraph is poorly written. From what I can see IFS is used in both coupled and uncoupled runs. The only exception is that in the atmosphere-only runs, the authors use a tuned version of IFS. There must be a simpler and clearer way to say this.

4)Page 4, line 25: what " longitudes along the Equator" actually mean?

5)Eqn2: What does "tendency before perturbation" and what does "tendency after perturbation" actually mean? Shouldn't these just be "deterministic tendency " and "total tendency" (=deterministic plus stochastic tendency perturbation)?

6)Page 7, line 5: "...If the model were retuned for each run, it is not possible to determine whether it is changing the tuning parameters, changing the resolution, or including stochastic physics that is responsible ..." This is not necessary true. What would one say if the parametrizations are scale aware, i.e, each (tuning or not) parameter depends on the resolution?

7) Page 7, line 10: " Further tuning has been performed in order to produce a realistic Quasi-Biennial Oscillation (QBO) at higher resolutions." This seems to contradict what the authors say in previous paragraph. Or am I missing something?

8) Section 3.1, First paragraph: the quantitative description of the SPHINX project is already made in the previous sections. Why there is a need to repeat it here?

9) Page 8, line 14: "...active and three control members without the stochastic physics, " Add " without the stochastic physics"

10) Page 9, line 21: " Therefore the SST for the future scenario have been obtained as a combination of HadISST2.1.1 variability and the CMIP5 EC-Earth simulations ensemble mean." What kind of improvements are expected here when the "non stochastic and low-resolution" CMIP5 model results are used to set the boundary conditions for the new runs? What is the innovation here? What are we supposed to learn from such runs? Since nothing is reported in here about the FSA simulations I guess the reader is summoned to wait and see for what to come? Nonetheless, the fact that the boundary conditions were based on the "bad" models that are meant to be improved is questionable.

11)Page 9, line 26: "This provides for each grid point the average expected SST increase from the present day to the future period according to a GCM,..." Just out of curiosity, how big is the variance of the 10 CMIP5 members that are used to compute a future "mean GCM" SST?

12) There are way too many details in the paragraphs on how initial and boundary conditions for future climate scenarios are obtained. While it is useful, it lacks scientific background. It makes the paper more of an internal report than a scientific research paper. There is no doubt that the publication of this work (though not in its present form) will be useful and can potentially serve as a benchmark for future developments of stochastic parameterizations. However, the paper will be better if all the technical description of the data in section 3 (including figs 2,3,4) is put in an appendix and have the main text focus on the topic of the paper, i.e, the importance of stochastic physics and high resolution climate simulation. Similarly, I don't understand why a whole section (section4) is needed for "technical configuration". If the authors feel like this is important they could add in an appendix also.

13) Page 12, line 25: "slight --->slightly"

14)Page 12, line 29: "the stochastic physics" insert the. The authors are not possibly trying to say that all stochastic parameterizations would have the same performance or are they?

15)Page 13, line 5: stochastic parameterizations ---> the SPPT and SKEB stochastic parameterizations

16)Page 15, end of line 22: who are "they"?

Hide

RR by Anonymous Referee #2 (23 Jan 2017)

Suggestions for revision or reasons for rejection

This paper describes the model configuration and experiment setup of the
project Climate SPHINX (Stochastic Physics HIgh resolutioN eXperiments),
which explores the impact of stochastic physics in an ensemble of 30-year
climate integrations at five different atmospheric horizontal resolutions
(from 125km up to 16km) within the EC-Earth model. This suite of
simulations is impressive and the setup throughtful, in particular that of the
coupled simulations, and the description of the experimental details
excellent.

After the experimetal description in the first 10 pages of the
manuscript, preliminary results are summarized in the following 4 pages.
They are comprised of an analysis of Tropical rainfall variability, an
analysis of the Madden-Julian Oscillation variability and an analysis of
Mid-latitude atmospheric blocking variability. Frankly, I was
dissapointed by the evaluation of the simulations: while these are
important topics, the manuscript did not go far beyond merely stating the
results without a deeper analysis or understanding. I assume that the
authors kept the evaulation of the simulations to a minimum, so that they
can publish then elsewhere.

Given that the stated aim of GMD is "the public discussion of the
description, development, and evaluation of numerical models of the Earth
system and its components" and not focused on the scientific results per se,
I recommened acceptance with minor revisions. This judgment is based
on the fact that the model description and experimental setup are well
described and thought through.

Nevertheless, I encourage the authors to extend the evaluation of the
simulations somewhat and in the following I will suggest how the evaluation
section could be improved:

1. I would have liked to see some mean field comparison of temperature, wind,
precipitation and in particular their variabiliy.

Tropical Rain:
2. The manuscript interprets the difference between the TRMM and GPCP datasets
as observational uncertainty. The TRMM data should be added to the right panels
displaying the frequency fractions with regard to GPCP.

3. Since in the observational datasets the differences in the rainfall
rates of 40mm/day are larger than the differences between the model
simulations, a crude significance test would help with the
interpretation of the results. This could be as simple as dislaying
rainfall rates for the first and second half of the simulation period.

4. I would suggest a second figure showing of rainfall rates and frequency fractions
for the small rainfall rates only, so that the differences in the curves can be seen
better.

MJO
5. Figure 6: Please specificy how the patterns for the 4 phases of the
MJO are computed. Are they based on all data, or are they calculated for
the observations only? It seems the comparison of the frequency of
occurrence and the amplitude of the MJO should be done in a common basis
system. Otherwise a case must be made, why different basis system are
used and the patterns for each experiments (or at least some) shown and
the change in patterns discussed.

6. Please refer to Weisheimer et al. 2014, who report that in particular
SPPT improves MJO in System 4.

7. Are these results significant? Again a dot for each first half and
second half of the simulations would be sufficient.

Blocking
8. Dawson et al. 2015 is in the Bibliography, but I couldn't find a
reference in the text. This paper should be clearly cited in the blocking
section.

9. Other work has suggested a positive impact of stochastic physics on
blocking. Why is this not the case here? Maybe it only helps at
horizontal resolutions of less than T159? Is there really no benefit
even at resolutions of T159?

Hide

ED: Reconsider after major revisions (12 Dec 2016) by Wilco Hazeleger

ED: Publish subject to minor revisions (Editor review) (06 Feb 2017) by Wilco Hazeleger

AR by Paolo Davini on behalf of the Authors (17 Feb 2017) Author's response Manuscript

ED: Publish subject to technical corrections (26 Feb 2017) by Wilco Hazeleger

AR by Paolo Davini on behalf of the Authors (02 Mar 2017) Author's response Manuscript

Short summary

The Climate SPHINX project is a large set of more than 120 climate simulations run with the EC-Earth global climate. It explores the sensitivity of present-day and future climate to the model horizontal resolution (from 150 km up to 16 km) and to the introduction of two stochastic physics parameterisations. Results shows that the the stochastic schemes can represent a cheaper alternative to a resolution increase, especially for the representation of the tropical climate variability.