Getting started with NetCDF4 and Python.

This isn't authoritative and it probably won't be the perfect method for how you use python, but it's a good first step. I also don't know much about using python on Windows outside of the WSL (Ubuntu but linux). If anyone has any suggestions that I should add please do let me know.

Getting miniconda:

Follow the instructions on this site if you don't already have anaconda.

The packages:

NetCDF4

This is a low-level interface for reading .nc files efficiently. It's a pain to work with in most cases, but you need to install this library so you can open Netcdf4 files (scipy only can read netcdf3 and will throw weird errors).

Run conda install -c anaconda netcdf4

xarray

Documentation This is the nice interface that will allow you to manipulate data less painfully. It conforms to the conventions of python libraries: DataArrays behave like numpy arrays with a few extra moving parts, while DataSets behave somewhat like Pandas dataframes.

Run conda install xarray

Cartopy

Documentation This is a plotting library that's really good for geospatial data. It's sometimes a bit idiosyncratic but it's usually worth trying if you want to make good global plots (it interfaces well with matplotlib).

Run conda install cartopy.

Optional:

wrf-python

Documentation This is from the Weather Research and Forecasting model and contains some useful interpolation and geophysical routines. Use with caution, it uses variable conventions that don't always agree with the models that generated the data you're working with.

Run conda install -c conda-forge wrf-python

pynio

Documentation This is a project that is slowly porting NCL's functionality to python, it's the one that Joe mentioned. It might have what you're looking for, it might not. It's worth a look.

Run conda install -c conda-forge pynio