Wrapping issue with certain nc files

[danlipsa]

A curviliniear grid shows wrap-around patterns. This is a picture with the grid saved from uvcdat and the nc file as shown in paraview. lon is between 0 and 360 so we do not do any wrapping. Data is not passed through a geographic projection either.

It is either a problem with how cdms reads the data or how we convert data into vtk.

[danlipsa]

See #1728 for more files and a script that shows them all.

[danlipsa]

@durack1 Do you have an image that shows how this should look like? Please attach that to this bug report. I think I have seen something like this but I cannot find the issue. I'll close #1728 as this was for an unrelated bug.

[danlipsa]

@durack1 OK, I found the issue: #1833

[durack1]

@danlipsa sure - which file is plotted above? I will plot using ferret which doesn't seem to show the same issue (I say this without being 100% certain).

[danlipsa]

@durack1 cmip5.CMCC-CESM.historical.r0i0p0.fx.ocn.fx.basin.ver-v20130416.latestX.WARN2.nc

[durack1]

@danlipsa good pick - I think this input file is going to cause grief for most software packages. The output of the basin variable plotted using ferret is below

So from your plot above, UV-CDAT is doing a better job of resolving the grid.

[danlipsa]

@durack1 It wasn't by design. :-) I just picked the first one in the ls order. You can try all the files in the zip file you attached to #1728 or please attach areacello_fx_CCSM4_1pctCO2_r0i0p0.nc from #1833 which seems to exhibit a similar problem.

[durack1]

Ok so this one is resolved better (areacello_fx_CCSM4_1pctCO2_r0i0p0.nc):

[danlipsa]

@durack1 I don't have the nc file. Can you attach it? Thanks.

[durack1]

No problem - attached below.
areacello_fx_CCSM4_1pctCO2_r0i0p0.nc.gz