HRRR data from NOMADS not showing up in xarray.DataArray

[MattLwx7]

When getting data from the HRRR Nomads opendap data source, it shows no values when opening the data as an array. I can plot the data but I can't do any calculations with it as a result. Does anyone know why this is occurring and how I can fix it?

[DanielAdriaansen]

Hi @MattLwx7, thanks for sharing your code snippet. Can you describe what calculations you're trying to perform on the data? When I use your code, I get the data shown in the attached image. Note the white shading in this image are missing data values, or "nan" as shown when you do print(temp.values) in your code. I will note that this looks like the HRRR data interpolated to a lat/lon grid? I am unfamiliar with this particular source of HRRR data, but it seems this grid is larger than the HRRR grid itself and thus the missing data values inserted around the actual grid cells where there are HRRR values.

[MattLwx7]

Hi @MattLwx7, thanks for sharing your code snippet. Can you describe what calculations you're trying to perform on the data? When I use your code, I get the data shown in the attached image. Note the white shading in this image are missing data values, or "nan" as shown when you do print(temp.values) in your code. I will note that this looks like the HRRR data interpolated to a lat/lon grid? I am unfamiliar with this particular source of HRRR data, but it seems this grid is larger than the HRRR grid itself and thus the missing data values inserted around the actual grid cells where there are HRRR values.

Thanks for the reply. I am trying to calculate LCL height from dewpoint and temperature, but with the nan values I don't get any calculated values.

[DanielAdriaansen]

Can you try this function to inspect the min value and see what you get?
https://numpy.org/doc/stable/reference/generated/numpy.nanmin.html

[MattLwx7]

I get a value doing this.

[DanielAdriaansen]

OK that's great- that means you probably have LCL values everywhere there's model data (see image above). Can you also try:
np.nanmin(lclheight) just to make sure? I think everything is OK but these nan values around the area with valid data are confusing.

For reasons I don't understand, this HRRR data has been put onto a different grid which is larger than the HRRR grid itself. Thus, there are missing data values around the edges of the extent of the HRRR data. Because when you print a NumPy N-D array it serves only a preview of data (i.e. the ellipses ... in the output implying there's more data between), and this preview is only at the beginning and end of the array object, it will only ever show the nan values. And, regular numpy.min and numpy.max do not ignore nan so those will always return nan as the min/max, so you'll need to use numpy.nanmin and numpy.nanmax.

Are you familiar with why this HRRR data is on a lat/lon grid as opposed to the Lambert Conformal grid it is usually on? I think as long as you're using this source of data, you will always have regions of nan values on the grid edges and will have to take care to ignore them or exclude them if necessary.

[MattLwx7]

Thanks! I think the lat/lon grid is associated with the data source: https://nomads.ncep.noaa.gov/
When using the unidata Thredds server it is on a Lambert Conformal grid, and it shows data instead of nan. I use this data occasionally but it doesn't have the amount of variables that the nomads data has. Thanks again for the help. I'll let you know if any other issues come up.