Developer's Committee meeting Tuesday 16 January 2018

Last edition: Wikinfo(changed_ts)? by Wikinfo(changed_by)?

1. Meeting Minutes
   1. Introduction
   2. Discussion on 2017 annual report
   3. Report of activities and 2018 perspectives for Working Groups
   4. NDS document final approval
   5. Information on important calls or projects
   6. Information on XIOS and OASIS: possible evolutions, convergences, sustainable resources for development and support
   7. Presentation on open development practices elsewhere : Alistair Adcroft
   8. Discussion on 2018 proposed Workplan
   9. Conclusions
2. List of participants
3. Connection information:

Meeting Minutes

Chairperson: Julien Le Sommer

Introduction

As decided during last meeting, the NEMO Developer's Committee subgroup has been initiated in order to build a reduced agenda focusing on the points for which a discussion during the meeting is useful. This subgroup met in December 2017, minutes available here

Discussion on 2017 annual report

The draft of 2017 annual report is available as input to the discussion here

Brief overview of the main streams of the year:

• Goal 1 (short term): Bring the 3_6 release to its final stable state in time for CMIP6
• Goal 2 (mid term): Implement new features as prerequisites for next version 4.0, including important developments, major simplifications and

robustness changes

• Goal 3 (long term): Contribute to prepare, discuss and update of NEMO development Strategy document (3 days meeting in April in Barcelona)
• A major new feature: the new sea-ice framework, in place of LIM2 and LIM3
• Report to be completed with discussions during this meeting on Working groups
• Additional work in 2017 to prepare and organise Enlarged DevCom? meeting in Barcelona in April

Round table for each of the Consortium and external experts :

• Met-Office (H. Hewitt): Very happy with the work done in 2017 and the report. Confirms that CMIP6 diagnostics did take more time than expected
• CMCC (P.G. Folgi): Also happy with work done. Points important and good work on simplification, AGRIF and HPC
• Mercator (Y. Drillet): Also happy although Mercator is not concerned by CMIP6
• CNRS (R. Benshila): Happy also, and especially to see that AGRIF is alive in NEMO and that this working group has been working and discussing, meaning that there is now more than J. Chanut on this part. Happy also to see that active work is on going on numerics
• NOC and CMCC Consortium experts not attending

Comments and suggestions of external experts :

• D. Marshall, L. Brodeau and J.M. Molines : not really much to add to previous comments.

The Developer's Committee congratulates the NEMO System Team for the amount and quality of the work done in 2017.

Report of activities and 2018 perspectives for Working Groups

10. Le Sommer reminds that the work expected from Working groups is somewhere in between the long term (NEMO development strategy document) and the clearly defined actions of the workplan. Working group are expected to address some questions on a yearly basis. Report and discussion on Working group should aim at clearly define those questions: those answered in 2017 and scheduled for 2018.

AGRIF

Add J. Chanut slides, or their content Discussion:

• AGRIF working with TOP is a functionality that has been dropped for now. This should be reconsidered.
• Usefulness of test cases (vortex used widely for AGRIF). Should become systematic
• Back port (better) status of AGRIF in the trunk now to 3_6_STABLE? Would need a large amount of work on LIM3 and vvl if is was to be done. No member of DevCom? advocates for this back port.
• Nesting tools allow setting up a NEMO configuration with AGRIF zoom. It’s update in order to work with the future release is scheduled, which is indeed an important and priority task, since for now people willing to use AGRIF are stopped by configuration problems. It is indeed critical to have a tool for this. Why doing it with Nesting tools, rather than for example with SIREN?

Answer:

• As a start because Canada is using it now, Nesting tools was the simplest way to go
• For the future (where a number of AGRIF zooms can be expected) using directly the AGRIF library is seen by WG as the best way to go, since it is more flexible.

• AGRIF WG is more active which is good news, but one of its active members TimGraham?, just left. Will his work time on AGRIF be replaced (question to Mt-Office and System team)? No answer at this point. Must be on the agenda of next June’s meeting

Recommendations of Developer’s Committee to AGRIF WG:

• The WG now has a clear view on the priorities for NEMO-AGRIF development
• Ensure the availability of a tool to set up NEMO configurations with AGRIF is a high priority for 2018

Air sea interactions

Presentation available here: here

Discussion:

• Raises the questions on where the coming developments are done at first and why (in the “trunk”, i.e. the coming release, or in 3_6_STABLE)? Discussions to be continued here, within WG and System Team
• D. Marshall: the merging of vertical mixing (TKE GLS and OSMOSIS)is not mature for now, at least concerning OSMOSIS. Should maybe be envisaged later on

Recommendations of Developer’s Committee to Air-sea interactions WG:

1. Add list of members of this new WG in minutes of this meeting
2. Clarify for next meeting (mid 2018) schedule of ALBATROS project contributions, their status (starting release, status of validation, of documentation) in the next 2 years.
3. Build a list of extra work (not in ALBATROS) and possible PIs for it, like sea-ice tiling with this new air-sea interface
4. Rationalise what should be strategy of connection of this new air sea interface to vertical physics in ocean

Configuration Manager

Add Stefania’s presentation content Discussion *Question on status of J. Harle work (had announced some pynemo config tools): on hold, no new for now

• For 2018, emphasis on Nesting tools for AGRIF is a priority for the Consortium
• There is quite an amount of work going on with 4 different tools, but no general frame
• Future implementations need discussions an decisions
• There will be a tutorial on SIREN at Mercator in 2018

Recommendations of Developer’s Committee to Configuration Manager WG:

Since the Developer’s Committee now has a subgroup to discuss things in advance and prepare agenda of meetings, the Devcom asks

1. For the next 6 months, this WG is on hold, and decision to be made during next DevCom? meeting (mid 2018)
2. For each of the four existing tools, complete a brief report, for April 2018, using the template sent by subgroup in March. List of reports : SIREN (Julien Paul), Nesting tools (J. Chanut), BDY tools (Stefania Ciliberti), pynemo (J. Harle). Same request to Test case WG on the tools to set up a test case configuration (S. Flavoni)
3. Since the tools for setting up an AGRIF configuration are the priority of Consortium for now, move the SIREN Actions of 2018 Work plan from “Shared actions”, to “Specific actions for Mercator”

• Informations from subgroup on other Working groups acivities and their 2018 agenda

HPC WG

This Working Group is closely following all the development and exploratory work going on simultaneously in this field. Important to note than the WG not only included the Consortium experts but also a number of other major partners including BSC, and computer vendors 2017

• Intra-node performance
• Real vs peak performance
  • Investigation of a benchmark configuration on different HPC systems
  • Vectorization improvement
• Hybrid parallelization
  • OpenMP shared memory approach
  • Fine grained (loop-level) approach à execution time gain ~5-6%
  • Coarse grained approach (tested on vertical physics) à intra- node parallel efficiency gain ~9%
• I/O performance improvement

2018

• Intra-node performance cont’
• Scalability improvement
  • Reducing the communication overhead
• Performance portability on heterogeneous architectures
• Reduced precision

Recommendations of Developer’s Committee to HPC WG:

Kernel (new) WG

Proposed by Mike Bell, see attached document, and List of experts contacted for now by Mike Bell:

Jérôme Chanut Andrew Coward Alex Megann Dave Storkey Florian Lemarié Laurent Debreu David Marshall Julien Le Sommer Adrian New Gurvan Madec Rachid Benshila Anne-Marie Tréguier

Recommendations of Developer’s Committee to kernel WG:

• Creation of this new WG is approved
• Is Mike Bell going to lead both HPC and Kernel WG?

Data assimilation and interface, modelling uncertainties WG

• WG is not really alive since 2016 (leave of its leader) , but subject is indeed important
• Next relevant meeting will be a workshop which Mercator-Ocean-International (MOI) is organising in late May or early June and that the scope of the workshop is defined in the MOI workplan for this year as to:

Organize a workshop with NEMOVAR and NEMO/Assim communities on data assimilation issues, setting the scene for creating a European infrastructure in marine data assimilation. This will include an assessment of the strengths, weaknesses as well as long term evolution plans and future commonalities (e.g. ensemble approaches) of the NEMOVar, NEMO/Assim and SAM tools.

• Some feedbacks on the outcomes of this meeting are expected for next DevCom? meeting mid 2018

Top discussion group

2017

• Increase TOP robustness with bug-fixes to code wrt BCs, damping, tracers integration;
• Code improvement for offline simulations;

2018

• Add TOP chapter to NEMO user manual and create a quick user guide on the NEMO forge documentation
• Develop a general scheme to simulate tracers vertical sinking;
• Development of the generalized tracer sinking scheme is the first issue identified in the NDS (10.2.1) whereas documentation (reference manual and wiki pages) has become a mandatory task before the next release (July 2018)

Recommendations of Developer’s Committee to TOP WG:

• AGRIF compatibility with TOP is a priority to be adressed by WG
• Documentation is indeed a high priority for the coming release of NEMO

Robustness and test cases

2017

• Methodology and tools to set up a test case have been designed (github and python notebook)
• A number of institutions of the consortium did implement at least one test case:
  • Overflow (CNRS – Simona)
  • Lock exchange (CNRS- Simona)
  • Wetting and drying (NOC – Andrew)
  • ISOMIP (Met-Office – Pierre)
  • Baroclinic vortex (Mercator - Jérôme)
  • SAS biperiodic box (CNRS - Clément R.)

2018:

• Each builder of an existing test case to update it on 2018 trunk, so as compliant with github and python notebooks
• A common github to be set up
• Choose the test cases the NEMO System team wants to use as validation tests (-> action of WG) and implement them in the validation script (SETTE)
• The discussion on criteria for acceptance of new test cases does not appear as a priority task for WG in 2018: already a lot to be done.

Recommendations of Developer’s Committee to Test cases WG:

• For next meeting mid 2018, WG to present the list of available and up to date test cases with a proposed list (to be discussed) on those entering the NEMO test bed.

Sea-ice

Add content of Martin’s slides

Recommendations of Developer’s Committee to sea-ice WG:

• Presentation is expected for next DevCom? meeting mid 2018 to address: distribution of fundings, new developments and developers, evaluation process and documentation

NDS document final approval

The document (attached to this page) has its final content and no red light from Sterring Committee. It is now important to check the final form for a public release. DevCom? members are requested to make a final check on the document form if possible and sent suggestions to Claire Lévy before January 23rd, in order to make the public release as soon as possible now.

Information on important calls or projects

Information on CMEMS calls now open at Mercator given by Yann Drillet

Some new calls have been opened very recently and are closely related to NEMO development. These calls aim to improve NEMO for forecasting centres. See list of calls : ​https://www.mercator-ocean.fr/en/mercator-ocean/tenders/ and dead-line for proposal submission is in March 2018. These calls are open to all and in particular to NEMO consortium.

Information on the proposal now being build by NEMO consortium given by Julien Le Sommer

NEMO consortium to answer Call SPACE-03-EO-2018: Copernicus evolution - preparing for the next generation of Copernicus Marine Service ocean models: ​http://ec.europa.eu/research/participants/portal/desktop/en/opportunities/h2020/topics/lc-space-03-eo-2018.html.

The PI of the proposal is Julien Le Sommer and a Writing Group nominated by NEMO Steering Committee is at work to write the proposal. Dead-line: 6 March 2018

Information on XIOS and OASIS: possible evolutions, convergences, sustainable resources for development and support

XIOS (Input/output component used in NEMO, developped at IPSL) and OASIS (coupler, developped at CERFACS) are two major features needed by NEMO. They are prerequisites for NEMO (as for other components of Earth System Models). It is therefore critical to be sure of sustainable resources for their development and support. Question has been asked by the subgroup to their leader (Yann Meurdesoif for XIOS, and Sophie Valcke for OASIS).

XIOS

Resources where up to now ensured by IS-ENES2 and the ANR Convergence projects (as stated in the answer from IPSL in 2013), both ended now. There is no answer for the near future, so as no governance for now.

OASIS

• Governance set up in 2017: First meeting of Users Group and Advisory board
• Concerning the sustainable resources, Sophie Valcke answered:

We stress here that for the 2018-2021 period the funding of these developments will come from CNRS (3/4 FTE), Cerfacs (1/2 FTE), the ESiWACE CoE (6 pms total), and possibly IS-ENES3 EU project ). The commitment of CNRS and Cerfacs can be considered stable: in practice, it ensures active user support to the community and the maintenance of the actual code with the possibility of minor upgrades. Additional developments will require additional funding and the only source of such additional funding seems to be EU research infrastructure projects (like IS-ENES3) or French national projects (like a CONVERGENCE follow-up). Other funding models were analysed and proposed to the community but none was considered practical. Such efforts have been successful for the past 20 years and, even if there is obviously no guarantee for the future, Cerfacs commits to keep on seeking such additional EU or national funding.

Presentation on open development practices elsewhere : Alistair Adcroft

• Presentation of A. Adcroft
• Discussion

Recommendations of Developer’s Committee:

• Continue this discussion and comparison within System Team and with Alistair
• Results to be presented by System Team at next DevCom? meeting

Discussion on 2018 proposed Workplan

DevCom? approves the proposed work plan so as its priorities on validation and documentation in the perspective of the new NEMO release announcement by mid 2018 Recommendations of Developer’s Committee:

• Following decision for Config Manager, move Config Manager development actions from Shared actions to specific Mercator ocean action
• Approval on priority to Validation and documentation actions to be produced during first semester 2018

Conclusions

Message to prepare for Steering Committee

• Annual report and WP2018 approved
• XIOS : suggestion to write a letter from SC to IPSL : sustainable support and no answer to CMCC problems with very hogh number of cores
• Make SC aware on difficulties about version used in operation versus version used for developments
• Need to endorse new NDS now and make it public
• AGRIF resources with Tim Graham’s departure is a question to be solved
• Different projects need contact points (see question addressed to Mercator)

List of participants

Name	Role	Institution
Claire Lévy	Project manager	CNRS LOCEAN
Dorotea Iovino	CMCC NEMO officer	CMCC Bologna
Andrew Coward	NEMO officer	NOC Southampton
Jérôme Chanut	AGRIF WG leader	Mercator Océan
Martin Vancoppenolle	SIWG co-lead	CNRS LOCEAN
Rachid Benshila	CNRS consortium expert	CNRS -OMP/LEGOS
David Marshall	External expert	CNRS LOCEAN 4ème étage
Clément Bricaud	Mercator Officer	Mercator Océan
Emanuela Clementi	NEMO Officer & WAVE WG lader	INGV
Pier Giuseppe Fogli	CMCC Consortium Expert	CMCC Bologna
Silvia Mocavero	NEMO ST & HPC-WG member	CMCC Lecce
julien Paul	NEMO ST	Mercator Océan
Helene Hewitt	Met-O Consortium Expert	Met-Office
Pierre Mathiot	NEMO officer	Met-Office
Nicolas MARTIN	NEMO ST	LOCEAN
Yann Drillet	Mercator Consortium expert	Mercator Ocean
Gurvan Madec	...	CNRS LOCEAN
Stefania Ciliberti	NEMO ST & CM WG Leader	CMCC Lecce
Tomas Lovato	NEMO ST - TOP WG leader	CMCC Bologna
Simona Flavoni	NEMO ST - CNRS Officer - Test cases WG leader	CNRS LOCEAN
Oriol Tinto	NEMO ST contributor	BSC
Julien Le Sommer	CNRS Consortium expert	CNRS LGGE
Olivier Aumont	TOP WG leader	CNRS LOCEAN
Jean-Marc Molines	External expert	CNRS LGGE
Laurent Brodeau	External expert	Ocean Next
Clément Rousset	NEMO ST	CNRS LOCEAN
Christian Ethé	NEMO ST	CNRS LOCEAN

Excused: Mike Bell (Met-Office Consortium expert), Francesco Trotta (INGV Consortium expert), Joel Hirschi (NOC consortium expert)

Absentees: Frédéric Dupont (Regional modelling external expert), Paolo Oddo (Regional modelling external expert), Joanna Staneva(Shelf modelling external expert), George Nurser (NOC consortium expert)

Connection information:

If possible, use a video-conference system. Otherwise, you must have a way to turn off your microphone in order to avoid echoing problems

To join the conference from your system:

Connection to the conference
Connection from a single terminal (Windows, OS X, tablet, smartphone ...)	http://desktop.visio.renater.fr/scopia?ID=723550***9188&autojoin
Installation guide of Scopia Desktop	doc_scopia_desktop-fr.pdf
IP	194.214.202.146
GDS	+33 (0)9 88 83 00 07 723550
SIP	sip:723550@195.98.238.109
H.323	h323:723550@mgmt.visio.renater.fr
Conference number	723550 (end by #)
Access code	9188 (end by #)
Conference
Title	NEMO Developers Committee
Begin	2018-01-16 12:45 Europe/Paris
Duration	04:30
Number of estimated connections	20
A test room is available to validate the configuration and the equipment compatibility.

room 9999
PIN 0000
How to connect: https://renavisio.renater.fr/user_guide