The UFS-Analysis package provides a suite of Python tools for evaluating NOAA UFS forecasts against verification datasets like ERA5.
This package has 4 major modules.
-
datareader: Download NOAA UFS forecast data from S3 buckets and ERA5 verification datasets from Google Cloud Platform (GCP), in Zarr format. Includes tools for describing data and retrieving subsets.
-
regridder: Resample (temporal), Regrid (spatial), and Align (temporal) forecast and verification datasets to get perfectly matching data structures.
-
util: Helper routines, along with an array of highly specialized analytical functions used for generating diagnostic notebooks.
-
notebooks: A collection of interactive Jupyter notebooks showing key forecast system diagnostics. As of May 2026, this covers ENSO Index and Teleconnections analysis for the Seasonal Forecast System (SFS), with NAO and PNA analysis notebooks to come shortly. More notebook sets are planned for UFS's Sub-seasonal to Seasonal (S2S) and Medium Range Weather (MRW) applications as well.
This package is designed to run on Linux operating systems only. Package managers such as Conda, Miniforge, and Mamba are
strongly recommended for building the prerequisite Python environment. Make use of the environment.yml file in this
repository for doing so.
Alternatively, all analysis notebooks can be launched on free cloud computing platforms like Binder and Google Colab. The UFS-Analysis package is accompanied by a public-facing Github IO page which contains url hyperlinks for running analysis notebooks right out of the box, no extra work needed. Explore the UFS-Analysis dashboard here: https://ufs-community.github.io/ufs-analysis/
cd tests
pytest