Optionally add bolus velocity to MOC

[xylar]

By default, the bolus velocity is added as long as GM is turned on.

[xylar]

For the MPAS-O paper, we need the MOC with the bolus velocity included. It was decided at today's MPAS-O meeting that the default should probably be to plot the MOC including bolus, since this is what is typically included in publications where ocean models are compared with one another.

[xylar]

Testing

I ran the analysis on the QU240 test case only so far. Here is the MOC from a baseline run (without bolus):

And here is from this run with bolus:

[xylar]

Tomorrow, I'll run the analysis on an EC60to30 run to make sure things are working.

[xylar]

@milenaveneziani, once I've tested, I'd like your help to make sure this is implemented in a way you're happy with.

@maltrud, @jonbob and @mark-petersen, I've included you as reviewers just so you can take a look at the results once they're posted tomorrow and tell me if you think they're sensible or nutty.

[milenaveneziani]

nice

[xylar]

This PR should address #310

[milenaveneziani]

looks good to me @xylar.

[xylar]

Great, let's see what the results look like tomorrow. We're kind of hoping that the bolus increases the MOC strength some because MPAS-O has a pretty weak MOC in our latest EC60to30 G-case simulations. Either way, it's the version of the MOC with bolus that we want to be comparing with others.

[xylar]

Here are the results of the MOC with bolus:
http://portal.nersc.gov/project/m2833/mpas_analysis_output/20180209.GMPAS-IAF.T62_oEC60to30v3wLI.cori-knl.afterSalinityFix_moc_bolus/

Here are a series of before and after images of each MOC plot:
No bolus:

With bolus:

No bolus:

With bolus:

No bolus:

With bolus:

These are only years 8 through 13 of the spin-up with ice shelves. It would be more useful to redo these for the longer spin-up without ice shelves. But is seems clear that the bolus velocity is actually making the MOC weaker! That seems surprising but maybe it shouldn't -- eddy velocities are often counter to mean velocities. Still, we should make sure nothing is wrong...

[vanroekel]

@xylar I've checked out this branch and am running with and without bolus on 40 years of the current low res g-case. I'll post results when complete.

[vanroekel]

@xylar and @milenaveneziani is there an official naming convention for the MOC transect files? In the mpas-analysis code it looks for

gridRes_SingleRegionAtlanticWTransportTransects_masks.nc

but on anvil, all files follow

gridResTag_Atlantic_region_and_southern_transect.nc

Either name seems fine, I just want to make sure that we all follow the same convention.

[xylar]

@vanroekel

is there an official naming convention for the MOC transect files? In the mpas-analysis code it looks for gridRes_SingleRegionAtlanticWTransportTransects_masks.nc

Yes, that's the required naming convention now. We used to supply the file name because there was only one but it became clear (after adding a mask for ice shelves) that we really needed a naming convention so all the needed mask files could be found automatically. Anvil must have data from a rather old copy of the files at NERSC.

We're working to get this stuff put on a common SVN repo where it will be checked out and updated on all supported machines automatically. Bear with us until that happens...

[mark-petersen]

By visual inspection, this all looks correct. Thank you Xylar.

I was also expecting Bolus to increase the MOC, but you are right that the eddy transport often counters the mean.

[vanroekel]

@xylar I'll rename anvil files. Not a problem at all, just wanted to make sure I was on the same page for anvil.

[vanroekel]

@xylar one small issue with this PR is a directory path issue in the default config file. I have made the change and pushed here. Everything else worked great. Thank you!

Atlantic MOC With Bolus

without

[xylar]

@vanroekel, thanks for the fix.

That's a mighty weak MOC...

[maltrud]

i've been trying to confirm that the bolus amoc is counter to the eulerian mean, but haven't been able to find anything yet. i may just ping Gokhan. it just seems a bit weird that the global NH appears to be a bit stronger with bolus, but the NA is weaker. this implies that the NPac is also stronger with bolus. i just haven't looked at this kind of thing enough.

[vanroekel]

This PR works great. But it would be nice to hear from Gokhan about the bolus and MOC before making this the default.

[xylar]

@maltrud, any update on this one from your discussion with Gokhan?

[xylar]

@maltrud, thanks for the email update. Everyone (@milenaveneziani and @maltrud in particular), is there a consensus on whether something is wrong here or if the results are just unexpected (but correct)? Any reason not to merge this PR?

[milenaveneziani]

I think so.

[xylar]

Thanks for the reviews @vanroekel, @mark-petersen and @milenaveneziani. I decided to go ahead and merge because I'm not sure if @maltrud is around. We can always do some fixing up later on if we discover an issue on the analysis side.