calc.potential_temperature does not recognize temperature units

[ThunderRoad75]

Using v 0.12.2 on Windows 10 with Anaconda, Python 3.8

Get a ValueError
ValueError: potential_temperature given arguments with incorrect units: temperature requires "[temperature]" but given "none".
Any variable x can be assigned a unit as follows:
from metpy.units import units
x = x * units.meter / units.second

As you'll see in the sample code below, I clearly assign tmpk units exactly as shown in the docs, yet it still says it does not have units.

import numpy as np
import sys
import os
from datetime import datetime, timedelta
from siphon.catalog import TDSCatalog
#from netCDF4 import Dataset
from metpy.calc import potential_temperature
from metpy.units import units

model_url = "https://www.ncei.noaa.gov/thredds/catalog/model-rap130/202009/20200909/catalog.html?dataset=rap130/202009/20200909/rap_130_20200909_1800_000.grb2"
vtime = datetime.strptime('2020090918','%Y%m%d%H')

# Get the data
model = TDSCatalog(model_url)
ds = model.datasets[0]
ncss = ds.subset()
query = ncss.query()
query.accept('netcdf')
query.time(vtime) # Set to the analysis hour only
query.add_lonlat()
query.variables('Temperature_isobaric')
data = ncss.get_data(query)

tmpk = data.variables['Temperature_isobaric'][0,20,:,:] # 600 hPa is 20th element

thta = potential_temperature(600 * units.mbar, tmpk * units.kelvin)

[jthielen]

@ThunderRoad75 It looks like this issue is due to numpy/numpy#15200, since in your example tmpk appears to be a masked array (what netcdf4-python uses), and you are attempting to construct the quantity by multiplying by units on the right, which is currently broken by NumPy.

To resolve this issue, you could instead multiply by units on the left, wrap your masked array by np.asarray to convert to an ndarray (example), or bypass all this entirely by using xarray instead (example).

[ThunderRoad75]

@ThunderRoad75 It looks like this issue is due to numpy/numpy#15200, since in your example tmpk appears to be a masked array (what netcdf4-python uses), and you are attempting to construct the quantity by multiplying by units on the right, which is currently broken by NumPy.

To resolve this issue, you could instead multiply by units on the left, wrap your masked array by np.asarray to convert to an ndarray (example), or bypass all this entirely by using xarray instead (example).

Changing the order of the units assignment worked. Thanks!