NEMO Developer's Committee meeting

The meeting took place by videoconference on Wed. 8 January 2020.

Participants:

• Chairperson: Mike Bell (also Met-Office Consortium expert)
• External experts: David Marshall, Olivier Aumont, Frédéric Dupont
• Consortium Experts : Rachid Benshila (CNRS), Julein Le Sommer (CNRS), Jérôme Chanut (MOI), Momme butenschon (CMCC), Pier Guiseppe Fogli (CMCC)
• NEMO Officers: Sébastien Masson (CNRS), Dorotea Iovino (CMCC), Andrew Coward(NOC), Pierre Mathiot (Met-Office), Guillaume Samson (MOI)
• Working groups leaders: Mike Bell (Kernel and HPC), Jérôme Chanut (AGRIF), Sea-ice (Martin Vancopennolle and Ed Bockley), TOP (O. Aumont & Tomas Lovato), Sébastien Masson (Air-sea interactions)
• NEMO Project manager (Claire Lévy)
• Permanent Invitees System Team members and Developers: Stefania Ciliberti, Emanuela Clementi, Oriol Tinto, Gurvan Madec, Amy Young

Absentees: Joel Hirschi (Consortium expert for NOC), George Nurser (Consortium expert for NOC), Gilles Garric (Consortium expert for MOI), Helene Hewitt(Consortium expert for Met-Office), Dan Lea (leader of data interface WG), and the following External experts: Paolo Oddo, Joanna Staneva, Jean-Marc Molines,Laurent Brodeau

Discuss draft Terms of Reference for Scientific Advisory Committee

Mike Bell invited comments on the proposed Terms of Reference (ToR) drafted by elaborated by the NDC subgroup​ for the NEMO Scientific Advisory Committee that will be distinct from the Developers Committee (subject to Steering Committee approval).
Questions arose on why biogeochemistry is declared out of scope? Consensus was that the interface to biogeochemistry (i.e. TOP) is firmly within scope. The PISCES component itself is not used by all projects using NEMO, but is part of NEMO scope as stated in the Consortium Agreement. Olivier Aumont raises the point that there are some important interactions of biogeochemistry with other components of the system for developments.
Data Assimilation remains out of scope of NEMO and its SAC but at least some interfaces to data assimilation are within scope. The ToR should include a mechanism for reviewing what is in and out of scope in case needs change.
Similarly, other components such as stochastic closures may increase in importance and the ToRs? should provide some flexibility or risk excluding new areas.
The ToR should not be overly proscriptive on the length and number of terms members can serve (the word "normally" can help that). A limited number of terms is a good idea to ensure fresh input. It may be better to shorten the terms and allow more terms. Varying lengths of appointment is recommended, especially for initial appointments, so that not too many members change at the same future stage.
Action: NDC subgroup to revise Terms Of Reference accordingly before sending it to Steering Committee endorsement.

Summary of progress in 2019

Claire Lévy presents a summary of NEMO development progress in 2019​ . Highlights include:

• Significant preparatory work for the Kernel evolution towards a 2-level time-stepping scheme.
• Work on vertical interpolation with AGRIF to allow AGRIF nests to use different vertical coordinates.
• Implementation of the Atmospheric Boundary Layer model and work on Air-sea fluxes and bulk formulae
• Reworking of tidal harmonics and tidal forcing options

In all 39 development branches were created in 2019.
The system team successfully closed at least as many tickets as were opened.
6 papers were published in the NEMO GMD special edition
Developments remain on track for a 4.2 release end 2020/early 2021

Lessons learned should mean a realistic workplan for 2020

Comments on the NEMO 2020 work plan: Working Group leaders then other members of the committee

Kernel Working group (Mike Bell)

Changes to enable the new time-stepping scheme have remained a priority and will continue apace in 2020. A nwe staff member has been recruited who will work on the IMMERSE action to develop horizontal pressure gradient schemes in 2020. ALE functionalities will continue to be explored in 2020 following work by Jerome and Alex Megann in 2019. All this work is in line with the current NEMO Development Strategy (NDS).

HPC Working Group (Mike Bell)

Fruitful discussions were held during the 2019 merge party and a method for implementing extended haloes has been agreed which brings together the work carried out in Paris and by CMCC. The agreed approach will be implemented in 2020. Basic DO-loop macros will be introduced early in 2020. A MetO WP item will extend this to enable 2D tiling. Oriol (BSC) and Sam (ECMWF) will implement a mixed precision capability. Exploratory work has been carried out into adaptation for GPUs (Mirek (MetO) and Andrew Porter (STFC)). The approach used so far has no impact on the code. All this work is in line with the current NDS.

AGRIF Working Group (Jérôme Chanut)

Jérôme Chanut presents 2020 AGRIF goals​ which includes:

• Further changes to enable vertical refinement.
• Adapting AGRIF to RK3 time-stepping
• Introducing more MPP flexibility

In addition, a CMEMS AGRIF development project is ending and it is important not to lose these developments which include:

• A New nesting tool based on DOMAINcfg
• E-W cyclic and North-fold AGRIF domains
• An update to the AGRIF external library (also allows for overlapping grids)

There are dependencies between this work and some of the Kernel work which may mean the overall WP is too ambitious. Needs some coordination (or at least an agreed calendar).
Action: Andrew and Dave S to investigate possibility of completing KERNEL-04_Storkey_Coward_RK3_stage3 by early Summer.

This pressure can be relieved if AGRIF does not need to be working with RK3 by the end of the year. Julien stated the IMMERSE requirement which is for a demonstration configuration with AGRIF nests by the end of the year – but the time-stepping to be used is unspecified. Claire asked if implementing the new AGRIF options for LF first and then RK3 later would mean a duplication of effort? The answer is not too much.

Sea-ice Working group (Martin Vancopennolle)

2020 goals include:

• EAP rheology
• VP rheology
• Evaluation if lagrangian drfiters
• Updating the NDS for SI3
• SI3 reference manual

2019 included lots of distractions elsewhere so that the work scheduled has not been done, neither developments nor documentation, but both are expected to be picked up again in 2020. The Sea-ice advection is still being worked on by Clement.

Action: Martin and Ed to add the actions into the 2020 proposed workplan and to check the existing ones

Air-Sea interaction WG (Sébastien Masson)

2020 goals include: Further work on OSMOSIS (including optimisation by Simon). It was asked whether the WG had developed a longer-term view (up date to their strategy)?

Biogeochemistry WG (Olivier Aumont)

2020 goals include:

• Accounting for biological fluxes between ice and ocean (requires coordination with Sea-ice WG).
• Compatibility of AGRIF with passive tracers
• Common light penetration algorithm
• Further work on coarsening. The approach here is to use OASIS to couple the BGCM to the physics but this work (by Eric) relies on Clement Bricaud’s coarsening code.

Rachid Benshila reported that the planned update to the AGRIF library should enable the possibility of using AGRIF (via a Grand-mother grid) to perform coarsening but there are no current plans to explore this option.
Given these plans involve three WGs (AGRIF/Sea-ice/BGC) the chair suggested a multi-group discussion may be useful later in the year .
Action: Olivier Aumont

Data interfaces action

2020 goals include:

• Cleaning of IOM. Simon has proposed this following some of the changes he had to make for the tidal harmonics work in 2019. The IOM code is due for a refresh and Sebastien is aware of the plans.
• NEMO to OOPS/JEDI interface for NEMOVAR. This was generally felt to be out of scope as a shared action but could be investigated as a specific action by the MetO.

Verification and validation WG (Mike Bell)

2019 saw much discussion and the generation of useful ideas. Work has started on writing a roadmap based on these discussions and this will continue throughout 2020

Physical Process WG (Julie Le Sommer)

Rather than a single long-term group, this WG’s purpose is to foster small, short-lived groups based on specific interests. Two candidates identified in 2019 are:

• Tides and fast barotropic motions. Fred Dupont has already taken the initiative and formed a group. Members include Jerome, Enda (MetO) and Simon
• Eddy closures for non-eddying simulations

GEOMETRIC is the front-runner as far as NEMO is concerned on the latter but some NDC members are also part of a US initiative which will explore a wide range of eddy closures for eddy permitting and non-eddying simulations. Timing may be critical and NEMO cannot afford to react too slowly in this field. A GEOMETRIC publication is imminent. The code exists but is not yet in the trunk or any official release. George Nurser is to work on a GEOMETRIC test-case in 2020.

How to keep NEMO live 10 years from now

Most points are covered in the presentation by Mike Bell​ but usability is an obvious omission. Alternatives such as MITgcm are still viewed as easier to configure and use. Much of this comes down to availability of reliable documentation and test-cases. Action: Work on documentation to be stated more clearly in 2020 work plan, especially concerning sea-ice . More generally, to be discussed in NDC subgroup and returned to later

Messages to Steering Committee

This year’s message will include:

• Summary of progress
• A request for each member to prepare an assessment of how IMMERSE resources add to the 1 FTE of the institution
• A request to separate the Scientific Advisory Committee from the Developers Committee
• Proposed changes to the NDC (in particular, creation of the V&V and Physical Processes WGs)

AOB (none)

Next meeting to be scheduled in 6 months