rectilinearGridExample.py

This script applies the CF Conventions to a basic NetCDF file so Vapor can read it. It performs the the following:

• Creates “Coordinate Variables” for the spatial and time dimensions

• Adds the “axis” attributes to these new Coordinate Variables

• Adds the “units” attribute to these new Coordinate Variables

The sample data for this script can be downloaded here:

https://github.com/NCAR/VAPOR-Data/blob/main/netCDF/simple.nc

Import packages:

import xarray as xr
import numpy as np
from pathlib import Path
# sphinx_gallery_thumbnail_path = '_images/rectilinear.png'

Open sample data as an xarray dataset. Modify the ‘home’ variable to point to the directory containing sample data

home = str(Path.home())
simpleNC = home + "/simple.nc"
ds = xr.open_dataset(simpleNC)

ds.info()

Create a Coordinate Variable for our ‘time’ dimension and assign a value to it

ds['time'] = np.linspace(start=0, stop=0, num=1);

Generate Coordinate Variables for our x, y, z dimensions. The coordinates will monotonically increase in spacing logarithmically, and are generated with numpy’s geomspace function. https://numpy.org/doc/stable/reference/generated/numpy.geomspace.html

ds['y'] = np.geomspace(start=1, stop=100, num=48)
ds['x'] = np.geomspace(start=1, stop=100, num=48)
ds['z'] = np.geomspace(start=1, stop=50, num=24)

ds.info()

Give our Coordinate Variables ‘axis’ attributes so Vapor knows which axes they apply to

ds.time.attrs['axis']      = 'T'
ds.x.attrs['axis']         = 'X'
ds.y.attrs['axis']         = 'Y'
ds.z.attrs['axis']         = 'Z'

Give our Coordinate Variables ‘units’ attributes

ds.time.attrs['units']     = 'seconds since 2000-0101'
ds.x.attrs['units']        = 'm'
ds.y.attrs['units']        = 'm'
ds.z.attrs['units']        = 'm'

It’s optional but advisable to give our scalar variables a ‘units’ attribute

ds.temperature.attrs['units'] = 'K'

ds.info()

Save our file for reading into Vapor

ds.to_netcdf( home + "/regularCompliant.nc")

Plot a 2D cross section of temperature

ds.isel(time=0, z=0).temperature.plot(size=6, robust=True);