HOME

ESMF Ticket #3614101

Up to Transformations and interpolations

Posted by Anonymous at April 3 2018

Dear OASIS/ESMF communities, (ESMF Ticket #3614101)

For our coupling system MOSSCO (www.mossco.de), we need ESMF-OASIS interoperability.

Specifically, our idea is to

(1) have an ESMF component that can interface with other models through OASIS calls.

- it can initialise oasis (based on a communicator provided by ESMF)

- can write a namcouple based on the MOSSCO coupling and ESMF Field/Grid specifications

- it can (re-)read a namcouple and use this information to create ESMF Field and Grid representations

- can both send as well as receive from/to other OASIS coupled applications

(2) run an ESMF-coupled system after OASIS initialization,

- fields are not known at OASIS init time

- possibly iterative system with ESMF runtime field generation -> namcouple generation

- ESMF shutdown/OASIS init/ ESMF init

I would like to explore in the communities - whether you think this is a bad idea overall :=)

- if there are like attempts and experiences

- if other groups are interested in developing this

- if you know of (or are yourself) a developer that is interested in coding this

We will be able to support such work financially.

Kind regards, Carsten Lemmen, 

Posted by Anonymous at April 9 2018

Hi Carsten,

It was interesting to read your plans for MOSSCO.

The parts that particularly caught my eye were, in item (1), the namcouple writing and re-reading and in item (2) the part about (coupling) fields not being known at OASIS init time. My interest comes from the point of view of coupling the Met Office Unified Model atmosphere to NEMO ocean, CICE seaice and a chemistry model. In particular we need to couple various configuration-dependent 3D fields between the atmos and chemistry models. The flexibility required by this means it's not really practical to have static pre-generated namcouple files (because we would need so many to cover all the different possibilities), so we have to generate them at run-time. We also add special instrumentation to namcouples which enables our models to read namcouples, thereby treating them as the single definitive source of all information relating to coupling exchanges.

So while we're not immediately concerned with ESMF and the scenario you are developing, there may be a couple of areas where it sounds like we might be trying to do similar things and could share experiences. I can supply more info via email if you're interested.

Regards, Richard Hill

Reply to this