blog/introducing-xwrf

[giscus[bot]]

blog/introducing-xwrf

https://xarray.dev/blog/introducing-xwrf

[rajkumar8581]

This is so cool. Would request you to provide a sample code for writing the post-processed output into a CF compliant netcdf file please.

[andersy005]

Ccing @jthielen / @lpilz

[lpilz]

Hey @rajkumar8581, I just responded to your post in our dicussions here. TL;DR: We are not currently guaranteeing full CF-compliance of the dataset, but expanding CF-compliance would be something to look into in the future.

[rajkumar8581]

Hi Lukas, thank your for your response. Unfortunately, i do not possess the talent to build the functionality. I get by with open source codes shared by the community for my experiments. I have 40 years' hi-res wrf run files with me - 140 TB, that i need to analyze. So I had been looking for alternatives to ARWpost to undertake this job. Thats why xWRF is "THE" solution for me. My immediate goal is to convert the wrfout files to netcdf and free up the scratch space in the server. So any help in this regard would be greatly appreciated. Also, I am not looking at full CF compliance. Just a working code would be fine.

[jthielen]

By default, xarray.Dataset.drop() will raise an error if a label is specified that is not present in the dataset. To get around this, you can only specify the relevant variables that are present that you wish to drop, or add the errors='ignore' option:

import xarray as xr
import xwrf

ds = xr.open_dataset("wrfout_d01.nc").xwrf.postprocess()
# sample or make changes to dataset
ds.drop(["metpy_crs","wrf_projection"], errors='ignore').to_netcdf("modified_wrfout.nc") # drop the CRS/proj objects as they are not serializable

[rajkumar8581]

Thank you Jon, that worked.

[jthielen]

2 - I found that the post process x and y are in UTM easting and northing and not Decimal lat and lon. Was it designed to be so?

Note that these are not UTM easting and northing, they are projection dimension coordinates (meters in x and y directions from the projection origin in the grid mapping space). We designed it so that decimal lat and lon are left available in the XLAT and XLONG auxiliary coordinates, since both forms of coordinates are useful for different use cases (and sometimes you need both)! If you were expecting lat/lon dimension coordinates (i.e., 1D lat and lon), then you will need to regrid your data using a tool like xesmf.

[rajkumar8581]

Sounds good. Let me try my luck with xesmf. :)

[rajkumar8581]

Thank you Jon, that worked.
Hey @jthielen Looks like to_netcdf is returning a blank file.

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
Cell In [48], line 1
----> 1 u.to_netcdf("modified_wrfout2.nc")

File ~/miniconda3/lib/python3.9/site-packages/xarray/core/dataset.py:1903, in Dataset.to_netcdf(self, path, mode, format, group, engine, encoding, unlimited_dims, compute, invalid_netcdf)
   1900     encoding = {}
   1901 from ..backends.api import to_netcdf
-> 1903 return to_netcdf(  # type: ignore  # mypy cannot resolve the overloads:(

[josephko91]

It's noted that one of the key features is "generating projection coordinates and providing a pyproj.CRS object." How does one extract the pyproj CRS object from the xarray dataset?

I tried doing: wrf_projection = pyproj.CRS(ds.wrf_projection.values)

But ended up getting the following CRSError:

---------------------------------------------------------------------------
CRSError                                  Traceback (most recent call last)
Cell In [96], line 1
----> 1 wrf_projection = pyproj.CRS(ds.wrf_projection.values)

File ~/.conda/envs/geo_env_clone/lib/python3.9/site-packages/pyproj/crs/crs.py:348, in CRS.__init__(self, projparams, **kwargs)
    346     self._local.crs = projparams
    347 else:
--> 348     self._local.crs = _CRS(self.srs)

File pyproj/_crs.pyx:2352, in pyproj._crs._CRS.__init__()

CRSError: Invalid projection: epsg:+proj=lcc +x_0=0 +y_0=0 +a=6370000 +b=6370000 +lat_1=33.900001525878906 +lat_2=33.900001525878906 +lat_0=33.90000915527344 +lon_0=-98.0 +type=crs: (Internal Proj Error: proj_create: no database context specified)

[josephko91]

My apologies, didn't realize there was documentation that answered my question :)

For anyone else who sees this and was wondering the same though, you can use ds_new['wrf_projection'].item() (as per the documentation) to extract the pyproj object.

[stephentjemkes]

Your introduction you explain why you developed xwrf, and demonstrate the value of the code on a large dataset.
I do see the power of xarray and xwrf, as we are dealing with a large diverse set of wrf output.
But i fear the uphill battle to get going.
What would be good is to to take the three plots of the python-wrf tutorial and start your demo with that.
Then a naive user like me, is up and running in 10 min or less.

[singledoggy]

Create Cartopy projection from pyproj.Proj seem to be unsolved question. I would like to know if xwrf can directly provide cartopy's Coordinate Reference System (CRS), or if there are plans for further development. Currently, I am still using salem for simple CRS reading.

[lpilz]

As of now, generating cartopy CRS objects is not supported in xwrf itself, however it is possible to do this using e.g. metpy. We don't have any active plans to develop this functionality but might revisit this if there is enough demand.

import xwrf
from metpy.plots.mapping import CFProjection
ds = xwrf.tutorial.open_dataset("lambert_conformal").xwrf.postprocess()
CFProjection(ds['wrf_projection'].item().to_cf()).to_cartopy()

[singledoggy]

Thanks! I never use the metpy before. Maybe this can be added to the examples in doc if others have the same confuse.