[MITgcm-support] AMOC estimation in cs grid

[Jean-Michel Campin]

Hi Estanislao,

To compute the AMOC (or, in general, the Meridonal Overturning Circulation) it is recommended
to NOT interpolate the horizontal velocity before integrating zonally and vertically the meridional
transport. The main reason is that the interpolation will not preserve the small divergence of
the flow field, resulting in large artificial error.

The recommended method is to direcly integrated the transport on the original model grid,
following the edges of grid-cell that are close to a latitude target, along a "broken line".

The matlab script to generate these "broken lines" and to use these to compute the MOC
are in the process of beeing updated, see PR #729, https://github.com/MITgcm/MITgcm/pull/729
I would advise to check this updated version (rather than the original scripts in
 MITgcm/utils/matlab/cs_grid/bk_line) because the updated method is simpler and therefore
easier to convert into python:
>From this branch: https://github.com/jm-c/MITgcm/tree/update_matlab_cs_grid
 in: MITgcm/utils/matlab/cs_grid/
 from the "README" file, lines 65-70:
> mk_isoLat_bkl.m :: define "broken-lines" as close to latitude-target vector
>         "yLat", for computing Meridional Transport stream-function.
>         => write matlab file: isoLat_${ydim}_cs32.mat (with ydim = length(yLat))
> use_bk_line.m :: an example that uses "isoLat_59_cs32.mat" to compute & plot
>         Meridional Transport Stream-function.
> 
 and the 2 matlab scripts are there.

Cheers,
Jean-Michel

On Thu, Aug 31, 2023 at 12:57:26PM +0800, ESTANISLAO-GAVILAN-PASCUAL-AHU wrote:
> Dear Community,
> 
> 
> I want to evaluate the amoc in my cs grid. I can linearly interpolate the u and v transport to the grid's centre, then compute the meridional component using the anglesn and anglecs. That should give me the meridional transport in the curvilinear grid. The last step is, to sum up or plot cross sections the points at the same latitude. Is there a method that does not require interpolating from curvilinear to a regular latlon grid? Thank in advance for your help in this matter.
> 
> Kind regards,
> 
> 
> Estanislao
> 
> 
> 
> 
> 

> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support