Plot has NaN values

[Chrismarsh]

I have a ugrid netcdf. Paraview can load this file and correctly displays the fields. However, when I load the same netcdf into uxarray and plot, the plot has a bunch of NaN values that don't exist in the original data. What am I doing wrong?

>>> ds.Elevation.values[np.isnan(ds.Elevation.values)]
array([], dtype=float64)

[philipc2]

Hi @Chrismarsh

Would you mind sharing the data file?

[Chrismarsh]

@philipc2 no problem. This is a simple non time varying parameter, but I can reproduce this when there is a time dim.
elevation.zip

version uxarray 2025.5.1

[philipc2]

Thanks for sharing! Do you know what coordinate system this data is in? For Mesh2_node_x, it has the following attributes but these don't seem to be correct considering the data is in the range (-1305003, -1254221)

standard_name :
longitude
long_name :
Longitude of 2D mesh nodes.
units :
degrees_east

[philipc2]

Doing the following works for me after converting the coordinates:

from pyproj import CRS, Transformer
import uxarray as ux
import xarray as xr

ds = xr.open_dataset(r"C:\Users\chmie\PycharmProjects\uxarray\data\evel\elevation.nc")

# NAD83 to WGS84
src_crs = CRS.from_epsg(5070)            
dst_crs = CRS.from_epsg(4326)            

transformer = Transformer.from_crs(src_crs, dst_crs, always_xy=True)
lon, lat = transformer.transform(node_x, node_y)

# Create a custom grid 
uxgrid = ux.Grid.from_topology(node_lon=lon, node_lat=lat, face_node_connectivity=ds.Mesh2_face_nodes)

# Create a custom UxDataArray
elevation = ux.UxDataArray(ds.Elevation.rename({"nMesh2_face": "n_face"}), uxgrid=uxgrid)

Seems like this would be a good opportunity to allow for different coordinate systems to be read in, as opposed to just typical Cartesian or Spherical ones.

[Chrismarsh]

Oh no! That's totally it! I forgot to switch my debugging code to back to outputting wgs84 instead of a projected coordinate system (albers as you identified). I will double check this tomorrow

[Chrismarsh]

I'm trying this

ds = xr.open_dataset("elevation.nc")
src_crs = CRS.from_epsg(5070)            
dst_crs = CRS.from_epsg(4326)   
transformer = Transformer.from_crs(src_crs, dst_crs, always_xy=True)
lon, lat = transformer.transform(ds.Mesh2_node_x, ds.Mesh2_node_y)
# Create a custom grid 
uxgrid = ux.Grid.from_topology(node_lon=lon, node_lat=lat, face_node_connectivity=ds.Mesh2_face_nodes)

I get

ValueError: coordinate Mesh2_face_x has dimensions ('nMesh2_face',), but these are not a subset of the DataArray dimensions ('n_face', 'n_max_face_nodes')

on this line

# Create a custom grid 
uxgrid = ux.Grid.from_topology(node_lon=lon, node_lat=lat, face_node_connectivity=ds.Mesh2_face_nodes)

[philipc2]

ds = xr.open_dataset("elevation.nc")

# NAD83 to WGS84
src_crs = CRS.from_epsg(5070)
dst_crs = CRS.from_epsg(4326)

transformer = Transformer.from_crs(src_crs, dst_crs, always_xy=True)
lon, lat = transformer.transform(ds.Mesh2_node_x, ds.Mesh2_node_y)

# Create a custom grid
uxgrid = ux.Grid.from_topology(node_lon=lon, node_lat=lat, face_node_connectivity=ds.Mesh2_face_nodes)

# Create a custom UxDataArray
elevation = ux.UxDataArray(ds.Elevation.rename({"nMesh2_face": "n_face"}), uxgrid=uxgrid)

Interesting, I'm not able to reproduce that error. The code above works fine for me (I am on the June release though)

You could try setting face_node_connectivity=ds.Mesh2_face_nodes.values and see if that changes anything?

[Chrismarsh]

Must have been a PEBKAC within jupyter and accidentally reused an incorrect variable as that all works correctly with a fresh kernel start. My apologies! This is all working as expected now, thanks very much.