LLC4320 SSU/SSV data at Pangeo cloud mismatched

Cloud
1

Since access to LLC4320 from ECCO portal is too slow, I’m using LLC4320 data at pangeo cloud for my research.

However, it seems that the SSU/SSV variable of LLC4320 is wrongly mismatched at Pacific Ocean.

U/V from Pangeo cloud

pangeo
pangeo1920×1200 167 KB

2

And here is the right one from ECCO portal:

U/V from ECCO portal

llcreader
llcreader1920×1200 170 KB

3

In order to create the first set of figures, you have to assemble the 13 faces in the dataset into a world map. How are you doing that? Are you using llcreader? That is what I recommend. If yes, which version of llcreader are you using (and please supply a few lines of code)?

By design, U points in the i direction and V points in the j direction on each of the individual faces. Because i and j do not line up with latitude and longitude on some faces, you need llcreader to do a transformation in order to get a map that makes sense. We have had errors with this in the past, but my impression is that they have been fixed.

1 Like
4

I use xmitgcm.llcreader for process. The xmitgcm version is 0.5.2.
Here is the code to create the first figure:

import numpy as np
import xarray as xr
from xmitgcm import llcreader
import matplotlib.pyplot as  plt

from intake import open_catalog
cat = open_catalog("https://raw.githubusercontent.com/pangeo-data/pangeo-datastore/master/intake-catalogs/ocean/llc4320.yaml")
display(list(cat))

# Get variables from catalog
grid = cat.LLC4320_grid.to_dask()
sst = cat.LLC4320_SST.to_dask()
u = cat.LLC4320_SSU.to_dask()
v = cat.LLC4320_SSV.to_dask()

ds = xr.merge([grid,sst, u, v])
ds = llcreader.llcmodel.faces_dataset_to_latlon(ds, metric_vector_pairs=[('dxC', 'dyC'), ('dyG', 'dxG')])
display(ds)

fig = plt.figure(figsize=[16,10])
ax = fig.add_subplot(211)
ds.isel(time=0).U.plot.imshow(ax=ax,vmin=-1,vmax=1,cmap="RdBu")
ax = fig.add_subplot(212)
ds.isel(time=0).V.plot.imshow(ax=ax,vmin=-1,vmax=1,cmap="RdBu")
# fig.savefig("pangeo.png",dpi=600)
5

I looked into this further, and it seems you are describing a known issue that happens because some of the metadata is missing. Please read through this thread: to_latlon does not rotate velocities · Issue #216 · MITgcm/xmitgcm · GitHub and use the hack provided by @dhruvbalwada .

6

OK, it did work. Thank you so much!