Weather Forecasting Papers: From Classical NWP to Modern NeuralGCMs

A curated bibliography tracing 120 years of weather prediction science

Overview

This repository documents the evolution from Vilhelm Bjerknes' 1904 primitive equations through Richardson's numerical methods, Charney's filtered dynamics, ensemble data assimilation, machine learning breakthroughs (GraphCast, Pangu-Weather), and hybrid physics-ML models (NeuralGCM).

Paper Evolution Categories

Each folder has its individual readme of papers it contains with a brief description of how it contributes to the weather prediction science research.

• Theoretical Foundations (1900-1940)
• Filtered Equations & First Success Forecasts (1947-1960)
• Instability & Predictability Limits (1956-1963)
• Data Assimilation Methods (1960-2016)
• Machine Learning Emergence (2019-2024)
• Hybrid Physics-ML Models (2024-)
• Important Related Works

Contributing

Contributions are welcome! This bibliography aims to be comprehensive yet focused on seminal works that shaped weather prediction science.

What to Contribute

• Missing seminal papers - Foundational works that influenced the field's trajectory
• Key review papers - Comprehensive surveys that synthesize major developments
• Historical corrections - Fixes to citations, dates, or descriptions
• Context improvements - Better explanations of papers' contributions
• New categories - Suggest additional thematic organization (e.g., "Parameterization Development", "Operational NWP History")

What NOT to Contribute

• Every paper on weather forecasting (this is a curated list, not exhaustive)
• Textbooks or books (journal articles only)
• Recent papers without demonstrated impact (wait for citations to accumulate)
• Tangentially related work (e.g., general climate modeling unless directly relevant to forecasting)

How to Contribute

1. Fork this repository

2. Add your paper(s) to the appropriate folder's README.md following the existing format:

   ### Author(s), Year
   
   **"Paper Title"**  
   _Journal Name_, Volume, Pages.
   
   **Contribution:** [1-2 sentences explaining impact on weather forecasting science]
   
   **Key concepts:** [comma-separated keywords]

3. Update the folder's README if adding a new paper

4. Submit a Pull Request with:

   • Clear description of what you're adding
   • Why the paper is significant
   • Which category it belongs to

Suggesting New Categories

If you believe a major theme is missing (e.g., "1974-1990 Parameterization Revolution" or "Ensemble Forecasting Methods"), open an Issue with:

• Proposed category name and date range
• 3-5 papers that would fit this category
• Justification for why it deserves separate organization

Quality Standards

• Papers must be peer-reviewed journal articles (no arXiv-only unless historically significant)
• Descriptions must be neutral and factual (not promotional)
• Focus on methodological/conceptual contributions over incremental improvements
• Maintain chronological and thematic coherence

Citation Format

When citing this repository:

@misc{weather-forecasting-papers,
  author = {[Your Name/Organization]},
  title = {Weather Forecasting Papers: From Classical NWP to Modern NeuralGCMs},
  year = {2026},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/ieadoboe/weather-forecasting-papers}}
}

License

This bibliography is released under CC BY 4.0.

Note: Individual papers retain their original copyrights. This repository provides bibliographic information and context only—not the papers themselves.

Acknowledgments

This collection was compiled to support researchers tracing the intellectual lineage from classical numerical weather prediction to modern hybrid physics-ML systems. Special thanks to the meteorology and machine learning communities whose work this bibliography documents.

Contact

Questions, suggestions, or major contributions? Open an Issue or reach out via [your preferred contact method].