barronh
commented
4 years ago There are three separate issues, and I will explain each. None of these problems are show stoppers, depending on what you want to do. I'll show a working example and you can decide if you want to pursue this in PseudoNetCDF or wrf-python (See link below).
First, The support for WRF meta-data is somewhat limited. In my experience, the time coordinate as text is sensitive to versions of numpy and the netcdf reader. So, reading is easy but writing it is harder. Further, WRF just hasn't been extensively used with PseudoNetCDF, so several things are harder than they need to be. Depending on what you want to do, this may be a trivial barrier. It is also an opportunity for me to make some updates.
Second, the uamiv example you're using is trying to read a wrf netcdf file as if it was a CAMx unformatted fortran file. That won't work. The -f option from the command line tells PseudoNetCDF what kind of file it is opening. The uamiv is the Urban Air Shed Model version IV, which CAMx adopted with adaptations. So, PseudoNetCDF is reading the data as if it was uamiv and it fails because it is not.
Third, the addcf option really only works for uamiv and ioapi formats. I fails silently, which I should update.
Now, depending on what you want to do, this may not be a problem. Below is an example that plots T2 in degrees celcius.
import PseudoNetCDF as pnc
wrfpath = 'wrfout_d03_2020-07-23_00_00_00'
rawf = pnc.pncopen(wrfpath, format='netcdf')
cff = rawf.sliceDimensions(bottom_top=0)
# Basically addcf asa a function.
pnc.conventions.ioapi.add_cf_from_wrfioapi(cff)
# Adds meta variables: time, longitude, latitude, x, y,
# and lambert_conformal_conic
# x, y are projected units, but no not include the false_easting/northing
# lambert_conformal_conic is CF projection definition including false_easting/northing
# in the future, x/y west_east, south_north should be added to match the lambert_conformal
# the time variable is not correct, so be careful.
# for now, I add these variables manually
x = cff.variables['x']
y = cff.variables['y']
cff.copyVariable(x - x.min(), key='west_east')
cff.copyVariable(y - y.min(), key='south_north')
# Using eval to convert temperature to celcius. This is a trivial example, but useful
# for computing WRF derived variables.
cff.eval('T2C = T2 - 273.15; T2C.units = "C"', inplace=True)
# Now make a plot using the coordinates added above
ax = cff.plot('T2C', plottype='west_east-south_north')
# Adding a map is usually part of the plot function, but needs to
# be added manually via the getMap function.
bmap = cff.getMap(
maptype='basemap',
llcrnrx=x.min(), urcrnrx=x.max(),
llcrnry=y.min(), urcrnry=y.max(),
)
bmap.drawcoastlines(ax=ax)
bmap.drawcountries(ax=ax)
# Selected some parallels and meridians based on yoru data
bmap.drawparallels([5, 10], labels=['5N', '10N'])
bmap.drawmeridians([71, 72], labels=['71E', '72E']);
# Save as a figure.
ax.figure.savefig('WRF_T2C.png')
del cff.variables['Times']
cff.save('test.nc')Like I said, I don't know what you want to do so this is just a toy example that may or may not help. See the link below for an alternative system.
Link to WRF example with xarray: https://scott-hosking.github.io/notebooks/wrf_with_xarray/
nishadhka
Many thanks for library.
I am trying to convert WRF output in NetCDF into NetCDF-CF format to get projected (x.y) array for certain variables.
The sample WRF output is attached here with https://drive.google.com/file/d/1rg9IG_dJG_Iz5kKOFBXSf6K3fSx7p4n1/view?usp=sharing
1.using Python import
Which ends up in error of
During handling of the above exception, another exception occurred:
Traceback (most recent call last): File "/home/Nishadh/miniconda3/lib/python3.7/site-packages/PseudoNetCDF/pncparse.py", line 901, in getfiles allreaders[file_format](ipath, **format_options) File "/home/Nishadh/miniconda3/lib/python3.7/site-packages/PseudoNetCDF/camxfiles/uamiv/Memmap.py", line 131, in init self.readheader() File "/home/Nishadh/miniconda3/lib/python3.7/site-packages/PseudoNetCDF/camxfiles/uamiv/Memmap.py", line 263, in readheader GDTYPE = self.GDTYP = {0: 1, 1: 5, 2: 2, 3: 6}[cproj] KeyError: 1413808128
During handling of the above exception, another exception occurred:
Traceback (most recent call last): File "/home/Nishadh/miniconda3/bin/pncgen.py", line 3, in
main()
File "/home/Nishadh/miniconda3/lib/python3.7/site-packages/PseudoNetCDF/pncgen.py", line 269, in main
ifiles, options = pncparse(has_ofile=True, interactive=False)
File "/home/Nishadh/miniconda3/lib/python3.7/site-packages/PseudoNetCDF/pncparse.py", line 611, in pncparse
ifiles.extend(pncprep(subarg)[0])
File "/home/Nishadh/miniconda3/lib/python3.7/site-packages/PseudoNetCDF/pncparse.py", line 839, in pncprep
fs = getfiles(ipaths, args)
File "/home/Nishadh/miniconda3/lib/python3.7/site-packages/PseudoNetCDF/pncparse.py", line 911, in getfiles
raise oute # from e
OSError: Unable to open path with uamiv(path, **{})
path="wrfout_d03_2020-07-23_00:00:00"
error="1413808128"