Meeting on NEMO assimilation component strategy

Thursday March 26 2009 01:22:34 pm

Objectives of meeting

During the last NEMO Steering Committee (20 January 2009), there was a discussion on the strategy for the development of the next generation of assimilation methods with NEMO. It was agreed that:
- the sustainable development of the NEMO Assimilation Component is of high importance and priority for all consortium members
- the development of NEMOTAM must be ensured
- the “NEMOVAR” consortium (now existing as a Memorandum of Understanding between ECMWF, Met-Office and CERFACS) identifies the resources to develop a data assimilation code for use with NEMO. This is an important step towards the sustainable development of the NEMO Assimilation Component
- it is necessary to begin the process of integrating an assimilation component in NEMO
- a strategy, taking in account all the major developers of this Assimilation Component, has to be elaborated as soon as possible.
- a meeting between these NEMO Assimilation Component developers should be held to discuss the best way forward and clarify the strategy.
C. Lévy and M. Bell were asked to propose the detailed scope and to organize the meeting.

Expected outcome
The expected outcome of this meeting is the definition of a strategy, taking in account all the major developers of this Assimilation Component. The aim of this strategy should be to establish rapidly the new Assimilation component of the NEMO reference.

Expected attendees, chair and secretary
The expected attendees are technical experts representing each of the groups willing to contribute to the sustainable development of the “NEMO Assimilation Component”, i.e. not all the teams involved in the use of data assimilation within NEMO, but only those who intend to bring some expertise and manpower in the choices and development of this NEMO Assimilation component.
Expected attendees include:
- One or more technical expert representatives from each of the members of the NEMO Consortium (CNRS, Mercator Océan, Met-Office, NERC-NOCS):
- One or more technical expert representatives from each of the members of the NEMOVAR consortium (CERFACS, ECMWF, Met-Office)
- Technical expert representatives from other groups developing substantial assimilation capabilities for NEMO (INGV- Bologna, LPO-Brest, LEGI –Grenoble… list to be completed)
It is proposed that the roles of chair and secretary are taken by Eric Blayo and either Mike Bell or Adrian Hines (exact distribution of roles to be decided).

Organisation, likely date / venue
After gathering all the inputs from the community using this forum, the chair and secretary will organize the meeting, inviting participants who will contribute most to the meeting and represent the various interests of the community
The meeting will take place in Paris over two days, starting at lunchtime on the first day, and finishing at lunchtime on the second day.

The first day will be devoted to short presentations (10-15mn max) from selected speakers (selection by chair and secretary and Gurvan Madec as NEMO Scientific Leader) on the present assimilation systems using NEMO, ongoing developments and workplans, and proposed contributions to the development of a common NEMOASSIM component. This first day will end with an introduction to the questions that will be addressed on the second day and agreement on how to organise the discussions.

The second day will probably start with a few groups working in parallel to answer the specific questions. A preliminary list of questions is:
- What is the detailed description of the NEMOASSIM component (i.e. the new component to be implemented in the NEMO reference)? What is the best strategy to efficiently assemble this puzzle (routines already existing, or to be developed, etc) ?
- What is the best way to complete and include a first version of the NEMOTAM component at this point? Do we need all the sub-components (dynamics, sea-ice, tracers) to be coded ? Which priorities ? What is the technical option in order to keep NEMOTAM up-to-date with the direct components?
- What could be the reference configurations to be used to validate the NEMOASSIM components (including the validation tests)?

The conclusions from the meeting will be drafted by the chair and secretary. They will be submitted to the meeting participants, the NEMO Steering Committee, and the NEMOVar Consortium.

The meeting will take place in Paris on 22 and 23 June
This forum is now open for all comments suggestions and expressions of intention to participate to this meeting.

Mike Bell and Claire Levy

Monday May 25 2009 06:40:49 pm

Meeting on NEMO assimilation component (Mike Bell, Eric Blayo and Claire Levy)

Objectives of the meeting
The overall objective is to make assimilation tools for NEMO available to the users community. This would require the development of the so-called NEMO-ASSIM component.
This meeting should be the kick-off meeting for this development.

The questions that will be addressed are:
1. What should the NEMO-ASSIM component contain exactly, in the short term and in the longer term ?
2. Which technical solutions ?
3. What is already available? What is to be developed ?
4. Who does what ? Which workplan ? With which resources ? Which coordination with the NEMO System Team ?
5. How can we ensure the sustainability? (In particular with regard to the future evolutions of NEMO)
6. Which policy for the distribution?

Tentative Program
The talks on Monday afternoon will typically provide inputs to questions 1 and 3, while the discussions on Tuesday morning will address all questions.

Monday afternoon (June 22): 15-20mn talks – list to be completed and organized
- The NEMO system and its constraints (someone from the NEMO team)
- NEMOTAM. Arthur Vidard
- NEMOVAR computation of analysis increments (inner loop). Anthony Weaver.
- NEMO Assimilation Tools at Mercator-Ocean C.-H. Testut
- SEEK with NEMO at LEGI J.-M. Brankart
- NEMO Assimilation Tools at INGV ??
- Ensemble Kalman filtering with NEMO ??
- Other suggestions ?

Eventually short (5-10mn) talks on:
- NEMOVAR comparison of model with observations (outer loop). Kristian Mogensen
- Observation quality control in NEMOVAR. Matt Martin
- NEMOVAR project management. Kristian Mogensen.
- A talk on diagnostics ?
At the end of the afternoon: synthesis and definition of the program (detailed list of questions) and of the organization (one plenary discussion or several working groups in parallel ?) on Tuesday morning.

Tuesday morning (June 23): plenary discussion or working groups in parallel will address the list of questions.

Practical details:
Meeting will take place in Paris at Instititut d'astrophysique de Paris (IAP) : 98bis, bd Arago 75014 Paris. Subway station: Denfert-Rochereau (Direct links with both ORLY and CDG airports).
How to get there:http://www.iap.fr/english/Discove...racticalInformation/howtogetIAP.html
If you intend to come, please register (see below)

Expected attendees, chair and secretary
The expected attendees are technical experts representing each of the groups willing to contribute to the sustainable development of the “NEMO Assimilation Component”, i.e. not all the teams involved in the use of data assimilation within NEMO, but only those who intend to bring some expertise and manpower in the choices and development of this NEMO Assimilation component.

Expected attendees include:
- One or more technical expert representatives from each of the members of the NEMO Consortium (CNRS, Mercator Océan, Met-Office, NERC-NOCS):
- One or more technical expert representatives from each of the members of the NEMOVAR consortium (CERFACS, ECMWF, Met-Office)
- Technical expert representatives from other groups developing substantial assimilation capabilities for NEMO (INGV-Bologna, LJK-Grenoble, LEGI –Grenoble… list to be completed)

It is proposed that the roles of chair and secretary are taken by Claire Levy and Mike Bell (exact distribution of roles to be decided).
The conclusions from the meeting will be drafted by the chair and secretary. They will be
submitted to the meeting participants, the NEMO Steering Committee, and the NEMOVar Consortium.
======================================================================================
Comments about these objectives and below a list of questions:
In order to have an efficient meeting, we would like to gather some inputs on the preceding points before the meeting. Following are some tentative comments and questions.
Everything can be discussed.
We would like you to send us some feedback by June 10 on these different points, in order to help us to prepare the meeting.

1. What should the NEMO-ASSIM component contain exactly, in the short term and in the longer term ?
As a starting point one could consider the simulator (e.g. the model plus some interfaces) and the assimilation engine. In variational idiom this would correspond to the outer and inner loops respectively and to predictor and corrector (or forecast/analysis) in the stochastic approach.
Therefore the following aspects can be proposed as components of NEMO-ASSIM:
- Common to all methods (and written in a standardized way):
- Management of the observations
- Management of the analysis increment
- A catalogue of assimilation methods (variational, sequential…)
Question: do you have some methods that you want to be included in NEMO-ASSIM? Is the code ready or not? If not, is it a heavy task to make it available? Can you do it?
Note that we will probably not be able to support more than a small number of assimilation frameworks.
- Quality control
- Tangent and adjoint models (TAM)
- Diagnostics (can be useful also for NEMO, without assimilation)
- Sensitivity / stability analysis (singular vectors… for NEMO with or without assimilation)

2. Which technical solutions?
- Definition of a common interface for the computation of the model equivalent to the observations (i.e. format of the output of H(x)) : which information must be transmitted ?
- What could be the reference configurations to be used to validate the NEMOASSIM components (including the validation tests)?
- How will we manage all the inputs to NEMO-ASSIM?

3. What is already available? What is to be developed?
we need your input

4. Who does what ? Which workplan ? With which resources ? Which coordination with the NEMO System Team ?
we need your input

5. How can we ensure the sustainability? (In particular with regard to the future evolutions of NEMO)
we need your input

6. Which policy for the distribution?
The NEMO-ASSIM component should be integrated in NEMO, and thus be disseminated with the same rules than NEMO.

PLEASE ADD YOUR ANSWER BEFORE June 10th in a "reply" below. Thank you for your contribution
======================================================================================
TO REGISTER, please send a mail to NEMO System Team (nemo_st @ locean-ipsl.upmc.fr): indicating, your First and Family name and your institution, with the following Subject: registration for NEMO-assim meeting