ENH: xarray grid output

[syedhamidali]

• Implementation of Does PyART have the slice option for the grid objects #1476 (comment)

To streamline the process of extracting a subset from pyart.core.grid or the grid object, I propose to add a few more variables to the xarray dataset that mirrors the grid object. When this xarray dataset is written to a NetCDF file, it becomes equivalent to the original pyart.core.grid object.

How can this be utilized?

import pyart

# Reading a grid from a netcdf file
grid = pyart.io.read_grid("file.nc")

# Converting the grid to an xarray dataset
ds = grid.to_xarray()

# Slicing the dataset based on x and y coordinates
ds = ds.sel(x=slice(5000, 70000), y=slice(5000, 70000), drop=True)

# Saving the sliced dataset to a new netcdf file
ds.to_netcdf("outfile.nc")

Subsequently, it is possible to read the modified dataset using pyart as follows:

# Reading the modified dataset with pyart
grid_small = pyart.io.read_grid("outfile.nc")

This enhances the process of working with grid data, making it more convenient and efficient for users.

[mgrover1]

What is the reason for using a byte-encoded name here?

[syedhamidali]

Because some radar data use the following convention:

print(grid.radar_name)

{'long_name': 'Name of radar used to make the grid',
 'data': masked_array(data=[[b'D', b'O', b'W', b'8']],
              mask=False,
        fill_value=b'N/A',
             dtype='|S1')}

To make it look like b'DOW8, because if you read a pyart or radx generated grid using xarray, it shows up like array([b'DOW8'], dtype='|S4'). That's why, I am using byte-encoded name so the result looks like

ds["radar_name"] = xarray.DataArray(
                np.array([b"".join(grid.radar_name)])
print(ds["radar_name"])

array([b'DOW8'], dtype='|S4')

[mgrover1]

I think this variable, along with the other origin_latitude, etc. variables belong as coordinates - can you add those to the list of coordinates? I think that is what is causing the failure in CI here.

[syedhamidali]

@mgrover1 My apologies for the late response; I've been tied up with exams recently and have another busy week ahead. I'll get to it as soon as possible. Thanks for your suggestion.

[mgrover1]

No worries! Take your time - just let me know when you are ready for another review!