Welcome to EAGLE!

Welcome to the Github repository for Project EAGLE (Experimental AI Global and Limited-area Ensemble forecast system)! EAGLE is a joint effort between NOAA Research Laboratories and the Earth Prediction Innovation Center (EPIC) in the Office of Oceanic and Atmospheric Research (OAR), and the National Weather Service (NWS). The goal of Project EAGLE is to provide the community with tools to test and deploy their own AI models for weather prediction. As a part of Project EAGLE, this repository current contains various configurations to guide users through a complete scientific workflow to train an AI model. This pipeline includes environment creation, data preprocessing, model training, inference to create a forecast, verification of forecasts, and visualization tools. See the various pages below for more details.

For further information about Project EAGLE, please see the NOAA EPIC website

Table of Contents

User Guide

• 1. Quickstart Guide
  • 1.1. Building and Running EAGLE
• 2. Runtime Environment
  • 2.1. Build the Environment
  • 2.2. Available Make Targets
• 3. Configuration
  • 3.1. Config Creation
  • 3.2. Config Description
  • 3.3. Config Realization
  • 3.4. Config Validation
• 4. Drivers
  • 4.1. Inspect Available Tasks
  • 4.2. Provision a Runscript
  • 4.3. Editing a Provisioned Runscript
  • 4.4. Executing Without Batch Submission

ML Pipeline Overview

• 1. Create Training Data
  • 1.1. ufs2arco Overview
  • 1.2. ufs2arco Quick Tips
• 2. Train a Graph-Based Model
  • 2.1. anemoi-core Overview
  • 2.2. anemoi-core Quick Tips
• 3. Generate a Forecast
  • 3.1. anemoi-inference Overview
  • 3.2. anemoi-inference Quick Tips
• 4. Verify Model Performance
  • 4.1. eagle-tools Overview
  • 4.2. wxvx Overview
  • 4.3. wxvx Quick Tips

EAGLE Models

• nested-EAGLE
• global-EAGLE

Reference

• Glossary
• FAQ

Community

• Contributing
• Support