Transforming (rotating+transposing) llc-data, requires to know the mate variable when transforming data variables associaed with a vector field. For example, in a rotated facet, the local U-variable is actually -V since the local I-index (or X in oceanspy) is parallel to latitude, and increasing i (or 'X') yields a lower (southern) latitude.
The way oceanspy works with this, is through a function name mates (line 619 in llc_rearrange.py), which iterates through all variables of a dataset and associates its mate as long as said variable is defined in a list of mated-vars vars_mates. The function in hand is:
def mates(ds):
vars_mates = [
"ADVx_SLT",
"ADVy_SLT",
"ADVx_TH",
"ADVy_TH",
"DFxE_TH",
"DFyE_TH",
"DFxE_SLT",
"DFyE_SLT",
"maskW",
"maskS",
"TAUX",
"TAUY",
"U",
"V",
"UVELMASS",
"VVELMASS",
"dxC",
"dyC",
"dxG",
"dyG",
"HFacW",
"HFacS",
"rAw",
"rAs",
"CS",
"SN",
]
for k in range(int(len(vars_mates) / 2)):
nk = 2 * k
if vars_mates[nk] in ds.variables:
ds[vars_mates[nk]].attrs["mate"] = vars_mates[nk + 1]
ds[vars_mates[nk + 1]].attrs["mate"] = vars_mates[nk]
return ds
Not the most elegant way to do this, but the easiest I could think of. This has worked so far with LLC4320 since these are all the data variables and grid variables that we had available.
However, there is a pair of forcing variables (from the newly available forcing variables) that need to be included in the list above, so that these can be properly transformed when necessary. These are:
['SIuice', 'SIvice']
In addition, the list above has TAUX and TAUY, but their default name is oceTAUX and oceTAUY (that is also how they appear when reading theEGshelfIIseas2km_ASR_full dataset -- see Kogur.ipynb) .
Before, those variables weren't available so it never caused an issue. This is a super quick fix.
Mikejmnez
commented
1 year ago
Description
Transforming (rotating+transposing) llc-data, requires to know the
matevariable when transforming data variables associaed with a vector field. For example, in a rotated facet, the local U-variable is actually -V since the local I-index (orXinoceanspy) is parallel to latitude, and increasing i (or 'X') yields a lower (southern) latitude.The way oceanspy works with this, is through a function name
mates(line 619 inllc_rearrange.py), which iterates through all variables of a dataset and associates its mate as long as said variable is defined in a list of mated-varsvars_mates. The function in hand is:Not the most elegant way to do this, but the easiest I could think of. This has worked so far with LLC4320 since these are all the data variables and grid variables that we had available.
However, there is a pair of forcing variables (from the newly available forcing variables) that need to be included in the list above, so that these can be properly transformed when necessary. These are:
In addition, the list above has
TAUXandTAUY, but their default name isoceTAUXandoceTAUY(that is also how they appear when reading theEGshelfIIseas2km_ASR_fulldataset -- seeKogur.ipynb) .Before, those variables weren't available so it never caused an issue. This is a super quick fix.