[MITgcm-support] Question about closing heat budget

[Jacob Wenegrat]

Hello All,

First off let me thank everyone who has contributed to the MITgcm and 
the mailing list. It's a great tool, and I've gotten quite a bit of help 
already digging through the mailing list archive.

I am currently running a simple reentrant channel model of a front, and 
finding that while the momentum budget closes nicely following previous 
advice found in the archive, closing the heat budget is proving more 
challenging.

To date I've followed the instructions posted here for closing the heat 
budget: 
http://mitgcm.org/download/daily_snapshot/MITgcm/doc/Heat_Salt_Budget_MITgcm.pdf.

Specifically, considering only subsurface grid cells (and with no 
shortwave fluxes):

TOTTTEND/86400 = ADV_Tend + KPP_Tend + DIFF_Tend

Where ADV_tend = d/dx ADVx_TH + d/dy ADVy_TH + d/dz ADVr_TH (accounting 
for cell volume in calculating this).
KPP_Tend = d/dz KPPg_TH (accounting for cell vol).
DIFF_Tend = d/dx DFxE_TH + d/dy DFyE_TH + d/dz DFrE_TH+ d/dz DFrI_TH 
(accounting for cell volume)

I find that this balance works well at the beginning of my simulation, 
works quite poorly in the middle (when the flow goes baroclinically 
unstable), and then works reasonably well again towards the end as the 
front spins down, although it is not completely closed.

Specific questions are:
1) I am using tempAdvScheme: 33, which I gather from the documentation 
has associated diffusion which can be difficult to diagnose. Does the 
associated diffusive flux from this end up within any of the above 
diagnostics? Might this be the cause of the large residuals after the 
formation of sharp frontal regions?

2) The tendency term appears to be a small residual between the 
across-channel (y) and vertical flux divergences. As the features 
sharpen to near grid-scale, might the finite differencing (and later 
interpolation back to the tracer grid points) be causing problems?

More generally any suggestions of avenues to pursue would be greatly 
appreciated. I'm not sure how the mailing list deals with attachments, 
so I haven't included my data file, but can easily do so if useful.

Thanks,
Jacob

-- 
Jacob Wenegrat
Postdoctoral Fellow
Dept. of Earth System Science
Stanford University

Email: jwenegrat at stanford.edu
Web: http://web.stanford.edu/~jacob13