Duplicated north cells with jperio == 4

Configurations Global
, ,
1

Hi,

On our eORCA025 config, NEMO v4.0.1 with jperio == 4. The first two and last two columns (x dims) are the same due to east-west cyclicity.

Then, my understanding is that each wet cell on the j == jpjglo row has exactly one twin on the other side (here “other side” is with respect to grid index i) of the North American continent, on the j == jpjglo - 1 row. No other cells are duplicated, and no cell is there three times. At least that’s what comes out when searching for geographical duplicates in our mesh_mask files.

Is that correct with jperio == 4? Got confused because because if it is, then I don’t get this line of the CDFTOOLS sea-ice diagnostics:

  CASE (4) ! ORCA025 type boundary
     tmask(1:2,:)=0.
     tmask(:,npjglo)=0.
     tmask(npiglo/2+1:npiglo,npjglo-1)= 0.

It’s meant to compute integrated sea-ice metrics (volume, extent etc.). The code above masks the duplicated cells to avoid double accounting: first the east-west cyclicity duplicates, then the top row, and finally one half of the second-to-top row. I don’t get that last step - seems to me that it’s removing an extra half-row… but I want to be sure that it’s not due to some inconsistencies between jperio and our coordinate files.

2

Salut Charles :grinning:

About North Pole Folding: you may find these sketches useful (thanks to Seb), Migrating from 4.0.x to 4.2.x — NEMO release-4.2.0 documentation

You have a “T-point” folding, i.e. the symmetry at T-points concerns 3 lines along j ; the center (invariant) of the transform is located at the T-point located at (ip = npiglo/2 + 1 ; jp = npjglo-1). Hence, row at j=npjglo-1 is symmetric along i around this point. So it makes sense counting points only once for half of this line.
This said, I would have rather written in the block above:
tmask(npiglo/2:npiglo,npjglo-1)= 0.,
e.g. kept the pivot as part of “non-duplicated” points.

@smasson, aka “NEMO NPF master”, would easily confirm this or not.

Jérôme

NB: Everything above is based on Fortran indexing

1 Like
3

Many thanks Jérôme, and thanks to the person (presumably the NEMO NPF master) who made these new sketches. So if I got it right:

  • jperio == 4:
    • every cell on the j=npjglo-1 line has a symmetric on the same line;
    • every cell on the j=npjglo line has a symmetric on the j=npjglo-2 line.
  • jperio == 6: every cell on the j=npjglo has a symmetric on the j=npjglo-1 line, and that’s it.
4

You got it (there is aditionnal complexity for vector fields of course).

5

I wrote too fast. Of course, it should rather be;
tmask(npiglo/2+2:npiglo,npjglo-1)= 0.

1 Like