CREDIT-data

About

This repository ties to the main repository of the the NSF NCAR MILES Community Runnable Earth Digital Intelligence Twin (CREDIT) and provides additional support in terms of data pre-processing, post-processing, and verification. It also hosts the result of CREDIT papers from the NSF NCAR MILES group.

Python environment

Same as the main CREDIT repository.

As of right now, it is only accessable to the NSF NCAR MILES group, but will be publicly available soon.

Navigaition

• libs: A collection of functions/scripts used by this repository.

  • verif_utils.py: functions used for forecasts post-processing and verification.
  • score_utils.py: function sued for computing certain verification scores.
  • graph_utils.py: data visualization functions.

• data_preprocessing: this folder contains data pre-processing step for CREDIT model runs.

  • This folder is under construction; it contains ERA5 pressure level and model level preprocessing, static field preprocessing, zscore computation, and residual norm computation.

• verification: this folder contains verification steps for CREDIT model runs. It can be implemented as follows:

  • Copy verif_config_template.yml to verif_config.yml and modify based on your file directories
  • Go through Jupyter notebooks from STEP00 to STEP02
  • Access scripts folder for large-scale verification runs
  • Note: the verification setup has only been tested on the data analysis server of NSF NCAR: casper.ucar.edu

• visualization: this folder hosts results of the CREDIT papers.