NEMO in coupled mode, towards IPCC AR5 and others

Friday January 09 2009 11:04:38 am

One of the main streams of NEMO developments for the coming months concerns its coupling capabilities with other Earth System Models components. A first step will probably mainly concern the IPCC AR5, but all projects intending to use NEMO in coupled mode, including regional or coastal configurations, could be interested by this new forum.

What could (should?) be done and shared?
- A new coupling interface is now available within NEMO. It will allow a clear technical and scientific coupling interface for each component. All the progress for each component can now be added and shared within the NEMO framework. The coming developments could be listed and shared, with their improvements.
- A new OI Manager will be included by the end of February. It will allow easier additions of diagnostics and outputs. Since a list of requested and suggested outputs is underway for AR5, these outputs and diagnostics could be developed much faster if organized within the community.

Tentative schedule
- January 15th: beta version of NEMOV3.1 including the coupling interface (with LIM2) + bug fixes since NEMO V3
- End of January: NEMOV3.1
- End of February: beta version of NEMOV3.2 including the new IO manager for output
- End of March: NEMOV3.2 (corresponding, hopefully, to AR5 reference tag)
We are strongly suggesting that all coupled configurations use the new coupling interface and that all new IO developments are based on the new IO manager.

This forum, what for?
The aim of this first forum is mainly to list the resources of project/people using NEMO in coupled mode and avoid duplication of work. Based on this fist list and according to your responses, we will setup other forums on more specific tasks (for example: coupling with other LIM3, AR5 diagnostics…)

If you feel concerned, please respond and add the following information:
Project name:
Description:
Institution(s):
Person in charge:
Configuration(s) to be used (NEMO components, NEMO configuration, other ESM components)
Envisaged developments (ex: coupling with LIM3, list of new diagnostics...)
Envisaged experiments (ex: climatological runs, selection of IPCC scenarios..)
Envisaged topics for future forums:
Human resources:
Others:

Thursday January 15 2009 04:55:24 pm

Project name: IPCC AR5
Description: Climate Modelling
Institution(s): CERFACS and Météo-France/CNRM
Person in charge: Eric Maisonnave, David Salas
Configuration(s) to be used (NEMO components, NEMO configuration, other ESM components) :
OPA - ORCA1
Envisaged developments (ex: coupling with LIM3, list of new diagnostics...) GELATO ice model
Envisaged experiments (ex: climatological runs, selection of IPCC scenarios..) : IPCC AR5 decadal and long-term runs
Envisaged topics for future forums:
Human resources:
Others:

Wednesday February 18 2009 10:59:15 am

If you are using NEMO V3.1_beta, you should know that a bug was found in sbccpl in the computation of the surface current. see
https://forge.ipsl.jussieu.fr/nemo/changeset/1308
Note that NEMO V3.1 should be announced at the end of the week (after committing bugfixes on LIM3).

for information: we received the new output manager and will test it in the coming weeks.

seb

Wednesday April 08 2009 11:24:06 am

Dear All,

I have got several reactions indicating that a meeting on coupling NEMO to atmosphere models would be very much appreciated. The reactions strongly favoured to have the meeting in Toulouse, following the OASIS user meeting. CERFACS (thanks to Eric Maisonnave) is able to provide a meeting room for that day. Thus:

Meeting on technical issues in using NEMO in climate models
CERFACS, Toulouse
27 May 2009 (following the OASIS user meeting)

People who don't plan to come to the Oasis user meeting will have to register to this meeting, to be able to inform security to let them enter the Meteo France area. Please let me and Eric (maisonna@cerfacs.fr) know the name(s) of people who will attend the coupling meeting as soon as possible.

Format of the meeting

Aim is to discuss (technical) problems encountered in using NEMO as the ocean component in coupled climate models. To kick-off the discussion I would like one representative of each participating group to provide a short presentation of problems encountered, solutions found (or not found), etc. Please also send me a short description of problems that you would like to be discussed not later than 17 May. I will try to sort them, hoping to be able to give some a priori structure to the discussion.
Any other suggestions concerning the format of the meeting are most welcome!

Best regards,
Andreas

Tuesday June 02 2009 04:03:03 pm

Minutes: Meeting on NEMO coupling issues
CERFACS (Toulouse), 27.05.2009

About 15 NEMO-users met at CERFACS on Wednesday, 27.05.2009, to discuss problems encountered when using NEMO as the ocean component in a coupled climate model. The discussion was structured along questions and remarks submitted by the participants before the meeting. This document gives a short summary of the discussions.

Coupling in general
1.Coupling fields: Consistent definition, usage, documentation within the coupled model (Uwe Fladrich, SMHI)
It is difficult to identify which NEMO fields corresponds to which atmosphere field during coupling. this has been improved in NEMOv3, but the documentation is still missing. Sebastien Masson presents a short overview of the new coupling interface in NEMOv3. This interface is valid for LIM2 only. The interface to LIM3 needs still to be prepared. He reminds everyone to write to the Nemo system team mailing list (nemo_st@locean-ipsl.upmc.fr) in case of any questions, not to his personal e-mail address. some pressure might help to give the LIM3 interface a higher priority.
2.Vector rotation: via OASIS or explicitly in NEMO/other component models (Richard Hill, UKMO)
Richard describes that they do vector explicitly rotation within NEMO as the method provided within OASIS does not work as the numbers of u- and v-rows differ.
3.It would be very helpful if someone from NEMO and LIM group can give a brief introduction the required interface of NEMO/LIM to an atmospheric model and between water and sea ice (Shuting Yang, DMI)
See point 1 above.
4.I could do a short presentation about the new coupling interface available in NEMO V3 (Sebastien Masson, IPSL)
Done; see point 1 above.

OASIS
1.Oasis interpolation issues: experience with different methods; active development (SMHI)
One problem is runoff: In EC-Earth it seems that part of the runoff in interpolated to grid points that are land in NEMO.
Recommendations:
- use MOZAIC with ocean points corresponding to river mouths selected by hand
- change land-sea mask of atm. mod. to better fit that of NEMO
- use TRIP; needs definition of ocean points corresponding to river mouths
2.Multi-category ice thickness classes (UKMO, SMHI, IPSL)
Martin and Sebastien point out that work on coupling ice-categories has still to start. The proper way would be to add respective surface classes in the atm model. the idea was raised to work with average fluxes and redistribute them according to thickness classes.
It is not completely obvious whether this a job for NEMO or for OASIS people.
LIM3 still needs to be tested in several ORCA configurations.
3.Support for OASIS4 - any plans/timescales (UKMO)
OASIS4 is available on request, and the OASIS team is prepared to help. It is also possible to ask for dedicated help within IS-ENES.
4.Support for multiple instances of OASIS3 (UKMO)
Possible to use different OASIS processes for different fields ('pseudo-parallelism') – see OASIS3 user guide.
5.Use of OASIS restart files - are they really necessary? (UKMO)
The user can decide whether he wants the restart information in his model or in OASIS.
6.conservative regridding not conservative (KNMI)
The first-order conservative regridding method used in OASIS is not conservative if the grid-cells have curved boundaries. The solution is to use MOZAIC. The OASIS group is willing to incorporate weight-generation for grid-cells with curved boundaries into OASIS.
7.CONSERV method of SCRIP does not work properly (CMCC)
same as 6.
8.Domain decomposition following longitudes does not work (CMCC)
Antonella explains that they solved the problem. It is not present any more in NEMOv3. Sebastien asks everyone to submit model improvements and extensions to the NEMO team for possible incorporation in the code.

Model biases
1.warm bias round Antarctica (CERFACS, DMI, KNMI)
Warm bias in Southern Ocean up to 6 K in EC-Earth. Smaller in ORCA05/025 runs as well as in forced runs. Reasons unknown, but all IPCC models have problems (both with SST and sea ice). Two suggestions that might reduce problems:
- set namelist parameter nn_itau to 1 (only NEMO3)
- if using partial-step configuration, use energy and enstrophy conserving advection scheme (namelist parameter ln_dynvor_een=.true.)
We (KNNI) tried both, but the impact is small.
It was suggested that atmospheric resolution might play a role, too. Models with higher resolution produce less SLP bias, probably improving their winds.
2.The sea ice cover in SH far less than observed (DMI, KNMI)
probably related to 1.
3.SST-biases in the northern North Atlantic: connected to unrealistic sinking regions in the model? (DMI)
probably yes: wrong position of Gulf Stream (GS) in coarse models -> storm tracks too zonal -> THC too weak -> NAtl too cold -> wrong sinking regions -> wrong GS position

Miscellaneous
1.General experience and know-how on Nemo3.1 and/or LIM3 in standalone and coupled runs (SMHI)
nothing to say
2.NEMO on Earth Simulator (Front. Res. Cent.)
large decrease in turn-around time when migrating from ES to ES2 (SX-9); obviously related to I/O. Probable reason: blocksize (much) too small, would need to be optimized. Optimization of netCDF library may also help.
3.Spin-up of the coupled model and collapse of THC (IPSL)
(Near) collapse of THC dependent on model resolution; ORCA025 seems better than ORCA05.
in DMI's 60-year simulation no total breakdown -> spin-up state from which the run starts from reasonable.
4.initializing seasonal forecasts (CERFACS)
nothing to say
5.tips and tricks (KNMI)
- if d(SSH)/dt < 0: sublimation of snow on ice done twice (atm model and LIM – namelist parameter parsub)
- closed freshwater budget: what about enclosed seas (e.g., Caspian)?
- NEMOv3.2 will have improved I/O
- binary compatibility for different decompositions is desirable
- would be desirable to change decomposition without the need to re-compile
- for physics do not rely on namelist on NEMO website
- new ORCA1 files with smoothed coastline are available on NOCS website

General recommendations
- questions to NEMO team to NEMO system team mailing list: nemo_st@locean-ipsl.upmc.fr.
- create a 'ticket' with your question on http://forge.ipsl.jussieu.fr/nemo

Andreas Sterl
KNMI
02.06.2009

Friday June 05 2009 03:10:52 pm

There is the schematic diagram which tells you where are the call concerning the coupling in NEMO.
(I couldn't find the way to keep the blank so I put _ )
opa
--call cpl_prism_init(localComm)
--step
----sbcmod
------call sbc_cpl_rcv
------call sbc_ice_lim_2
--------call sbc_cpl_ice_tau
--------call ice dyn
--------call sbc_cpl_ice_flx
--------call ice thermo
...
----call sbc_cpl_snd
...
--call cpl_prism_finalize

Tuesday June 22 2010 04:55:19 pm

another small bug in AR5 diagnostics: scalar values + bottom pressure (this doesn't change the simulation results). see https://forge.ipsl.jussieu.fr/nemo/ticket/682 and associated changeset: https://forge.ipsl.jussieu.fr/nemo/changeset/1948