big bugs on ORCA025 and ORCA12 grid on the north pole

[smasson]

Following ticket:613 Ian Culverwell checked coordinates files of other resolutions and found several error for ORCA025.
A careful review of those files are therefore needed.

ORCA2:
I checked again coordinates_ORCA2_v3.2.nc. It looks OK. NO lbc errors and no errors visible around the geographical north pole (see attached pdf)

ORCA05
I checked a file called coordinates_ORCA_R05.nc.
There is 1 error for the norh pole folding of the arrays gphif, e1f and e2f. However this error is located on i = 360 that correspond to a pole of ORCA grid located on land => no impact.
Visual check of the geographical north pole looks OK (see attached pdf)

ORCA025
I checked a file called coordinates_ORCA_R025.nc.
There is big problems on the north pole folding of gphi[uvf], e1[uvf] and e2[uvf]. gphi[vf] are completly wrong! See detailed report bellow. In addition geographical north pole shows several bugs (see attached pdf)

gphiu : north pole periodicity (1) : 4 errors (located on : 718, 719, 720, 721)
e1u : north pole periodicity (1) : 4 errors (located on : 718, 719, 720, 721)
e1u : north pole periodicity (2) : 8 errors (located on : 1, 2, 3, 4, 656, 657, 658, 659)
e2u : north pole periodicity (1) : 4 errors (located on : 718, 719, 720, 721)
e2u : north pole periodicity (2) : 8 errors (located on : 1, 2, 3, 4, 656, 657, 658, 659)
gphiv : north pole periodicity (1) : 1441 errors
e1v : north pole periodicity (1) : 5 errors (located on : 3, 658, 720, 782, 1437)
e1v : north pole periodicity (2) : 4 errors (located on : 3, 658, 782, 1437)
e2v : north pole periodicity (1) : 5 errors (located on : 3, 658, 720, 782, 1437)
e2v : north pole periodicity (2) : 4 errors (located on : 3, 658, 782, 1437)
gphif : north pole periodicity (1) : 1441 errors
e1f : north pole periodicity (1) : 20 errors
e1f : north pole periodicity (2) : 16 errors
e2f : north pole periodicity (1) : 20 errors
e2f : north pole periodicity (2) : 16 errors

ORCA12
I checked a file called coordinates_ORCA_R12_lbclnk.nc
As its name is suggesting, north pole folding has no error in this file (but there is a minor error on the east-west periodicity for e1v and e2v locate dat j=3058)
However, plot of TUVF points position around the geographical north pole show very big problems! See attached pdf

Conclusion: big bugs on ORCA025 and ORCA12 around the geographical north pole. We must understand why...

seb

[idcatukmo]

Just to check: coordinates_ORCA2_v3.2.nc means the original (pre NEMO vn2) ORCA2 coords file, as ticket:613 suggests, right? Otherwise I suspect it's still wrong.

As for ORCA025: I have a hand-made fix for these coords, but I don't know why the original coords are wrong.

[smasson]

ORCA2 grids points position around the geographical north pole

[smasson]

ORCA05 grids points position around the geographical north pole

[smasson]

ORCA025 grids points position around the geographical north pole

[smasson]

ORCA12 grids points position around the geographical north pole

[smasson]

I updated the pdf files that where not really visibles!

coordinates_ORCA2_v3.2.nc is the one coming with version 3.2 that corresponds pre NEMO vn2. By the way, we also found a minor problem in this file: in the southern hemisphere, glamt is not exactly equal to glamv (and same problem with gkamu and glamf...).

Regarding ORCA025, how do you fix the problem? It is not only a problem of LBC. Some points are not properly positioned (gphi is wrong). Scale factors have some LBC errors too but I didn't checked if the values itself have errors...

seb

[idcatukmo]

Old ORCA025 coords near N pole

[idcatukmo]

New ORCA025 coords near N pole

[idcatukmo]

Thanks for ORCA2 clarification, Sebastien.  Didn't realise that glamt ne glamv etc in S Hem - bit worrying that there's a problem even where grid is regular!

As for ORCA025 - I made polar stereo projection of (glamx,gphix), noted (by eye) which points didn't lie on the regular lattice, calculated their correct posns in polar stereo space from those of the T-points (which look OK), and then derived the corresponding (glmax,gphix). Not as good as correcting the grid generation code, but then I couldn't get that to work at all.m Before and after pics attached.

Scale factors looked OK to me - certainly I didn't correct them.

Regards,

Ian.

[cbricaud]

ORCA025 and ORCA12 grid have been corrected to the north pole.