This PR provides updates on the run_graphcast.py script to provide the flexibility of running graphcast for the desired forecast lead time. In addition, it provides scripts to schedule a cronjob on both cloud and hera machines. As the hera machine does not have access to the internet, we needed to break down the job to three sections including, generating inputs to graphcast, running graphcast, and uploading the forecast to s3 bucket. The first and and last sections will be run on the control node while the second one (running graphcast) will run on the compute node. Therefore a script is provided to call the upload_to_s3 function to upload the forecasts to s3 bucket. In addition, some bugs with 'gdas_utility.py` got fixed.
Description
This PR provides updates on the
run_graphcast.py
script to provide the flexibility of running graphcast for the desired forecast lead time. In addition, it provides scripts to schedule a cronjob on both cloud and hera machines. As the hera machine does not have access to the internet, we needed to break down the job to three sections including, generating inputs to graphcast, running graphcast, and uploading the forecast to s3 bucket. The first and and last sections will be run on the control node while the second one (running graphcast) will run on the compute node. Therefore a script is provided to call the upload_to_s3 function to upload the forecasts to s3 bucket. In addition, some bugs with 'gdas_utility.py` got fixed.Linked Issues
Blocking Dependencies
Anticipated Changes
Input data
Needed libraries