add an option to post-process minimal nemo variables for ECMean

[etiennesky]

current hiresclim2 post-nemo.sh does many time-consuming calculations which are not use by ECMean , even with nemo_extra=0, such as calls to cdfheatc cdfmoc and cdficediags

It would be nice to identify which variables are needed and to only calculate them if a flag is set (tentatively called nemo_basic, maybe could be nemo_ecmean).

Any thoughts?

[plesager]

Why not? I noticed that the NEMO part was slower, but not much of an issue for me. Do you have timings?

[etiennesky]

Hi I did in the past but lost them. I would say about 2 minutes vs 5 minutes for 1 year lowres, so not insignificant if doing long runs, especially at HR.

I have a simple change which works, moving the nino34 and heatc computations into the nemo_extra part.

This will have to be revised when the timeseries works.

[etiennesky]

In fact, my solution does not involve having a new config variable, simply computes these variables if nemo_extra=1

[oloapinivad]

I agree on this: second order NEMO results are provided by barakuda directly from the output, so we probably don't need them for hiresclim. I guess the main choiche will depend mainly if we want to maintain or not the oceanic timeseries

[etiennesky]

closing this issue as my changes have been merged by Philippe