[MITgcm-devel] updates on seaice code consolidation

[Patrick Heimbach]

Hi there,

I am summarizing a brief telecon which we had today between
JPL (Dimitris, Ian, An) and MIT (Gael, Jean-Michel, myself).

Our goal is (has been for a while) to
* incorporate Ian's changes into "main" code
* clean up code
* be able to get rid of the _if versions in pkg/seaice (and MITgcm_contrib/ )

The status as of September was:
* the "legacy" code in pkg/seaice
* the "_if" code in pkg/seaice
* latest code from Ian in MITgcm_contrib (at my request) for the purpose of merging with code in pkg/seaice

At the time of Martin's and Dimitris' visit we had hoped to be able to 
run the MITgcm_contrib/ code, but attempts by several people to do so 
failed despite repeated iterations of fixes.
Moving forward we had formulated a strategy at MIT to 
get a code that 
1. tries to incorporate Ian's modifs,
2. do so in a controlled way (traceable via CVS),
3. cleans up many of the issues of legacy code (yneg, and many other issues)
5. a cleaned-up sequence that is more amenable to modularization, e.g. for coupling
   (e.g. move age calculations and similar to separate routines,
   deal with salinity code and salt_plume params),
5. solves non-conservation issues.

Main work on 1.-4. was taken on by Gael, and item 5. tackled by Martin.
As far as I understand, starting point for Gael's merge was the "legacy" 
code with main merges taken from _if routines.

I guess this has gone a long way, but
in order to bring it to successful completion we have the following issues:
1. extent to which code improvements in Ian's MITgcm_contrib are still missing
   (Ian voiced concerns that some important terms described in his thesis
   as adapted from McPhee are still missing)
2. make sure that major setups that we are aware of run "successfully"
   with the latest code, so that we can complete the clean-up

To deal with 1. Ian has agreed to document what pieces are still missing
(forthcoming on mitgcm-devel).

To deal with 2. some of the major setups need to re-run (and retuned?).
Especially, Matt who's probably the only one having used the _if code
(I lost track which one) should re-run SOSE and see if he's happy with "latest code".
Suggestion is that this be done with the next checkpoint (c62p).
"full list of setups are:
* Matt: SOSE
* Dimitris/Hong: ECCO2
* An/Gunnar: Arctic
* Gael/Patrick: ECCO-GODAE codes
* Holly: Atlantic setup
* Martin: his setup(s)
* Ian: Lab Sea setup, with special focus on adjoint behaviour

Hope I captured it all, not too confusing and not too inaccurate.
Cheers
-Patrick

---
Patrick Heimbach | heimbach at mit.edu | http://www.mit.edu/~heimbach
MIT | EAPS 54-1518 | 77 Massachusetts Ave | Cambridge MA 02139 USA
FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE patrick.heimbach