[MITgcm-support] balancing the heat equation
Martin.Losch at awi.de
Wed Sep 5 03:12:11 EDT 2007
without checking details in the code I would suggest that you use the
diagnostics package for diagnosing the different contributions. As
far as I can see, the diagnostics are all available:
diagName = 'TOTTTEND'
diagTitle = 'Tendency of Potential Temperature'
uTx/vTy/wTz: 'ADVx/y/r_TH '
KTxx/yy/zz: 'DFx/y/rE_TH ' (and 'DFrI_TH ' for the implicit part
which is relevant for KPP)
Then there is the non-local term in KPP, which can be diagnosed with
'KPPghat '; however that is not the contribution to diffusive flux,
and you'll have to alter the code a little to get that contribution
with the diagnostics package. Here's a hack that abuses DFrE_TH for
this purpose: In gad_calc_rhs.F find the calls to KPP_TRANSPORT_T
etc. They are within a #ifdef ALLOW_KPP/#endif block. just before
this there is a diagnostics call (#ifdef ALLOW_DIAGNOSTICS/#endif).
Just exchange the order of these two blocks and DFrE_TH will give you
the flux-contribution of the non-local term (but the explicit
diffusive flux will be overwritten and not diagnosed, which is only
OK, if you are using KPP without GMREDI!).
A better solution is to add a new diagnostic and insert it after the
calls to KPP_TRANSPORT_T etc).
On 4 Sep 2007, at 21:19, Allison Byrd wrote:
> Hello mitcgm-ers!!
> I am trying to solve the heat equation
> with the goal being the left and right hand side equaling each
> other to machine precision. We have a large imbalance between the
> two sides and are lost to the small detail we are missing.
> for the left hand side we use gTnm1 for dT/dt, and then take the
> forward difference of T to solve for uTx,vTy, and wTz. For the
> right hand side we use Khorizontal=500 m2/sec. and Kvertical =
> KPPdiffKzT+2*10^-4m2/sec. Perhaps we are misusing the KPP
> schemes? Has anyone ever done this before?? Any help would be
> greatly appreciated.
> Allison Byrd
> FSU Dept of Oceanography
> Physical Oceanography
> Office Room 403
> byrd at ocean.fsu.edu
> This message was sent using IMP, the Internet Messaging Program.
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
More information about the MITgcm-support