This example notebook accompanies this post I made on medium (link not yet active), explaining how to use kerchunk to speed up GOES-16/17 NetCDF4 access on AWS S3.
Click on the Binder link below to run this notebook in a cloud environment. The notebook also works if executed locally.
First, if you're not running this on Pangeo or Binder, you'll want to set up the environment. I've provided the environment.yml file that includes all packages needed to run the notebook.
conda env create -f environment.yml
Then, activate the new kerchunk-tutorial
environment (if you want to change the name, you will have to change the "name: kerchunk-tutorial" line at the very top of environment.yml before creating the environment)
conda activate kerchunk-tutorial
Download and install kerchunk
from github. You will also need to make sure you have numcodecs>=0.8.0
pip install kerchunk