Skip to content

Latest commit

 

History

History
184 lines (174 loc) · 21.4 KB

README.md

File metadata and controls

184 lines (174 loc) · 21.4 KB

🌍 Awesome Large Weather Models (LWMs) | AI for Earth (AI4Earth) | AI for Science (AI4Science)

🧭 Guideline

A collection of articles on Large Weather Models (LWMs), to make it easier to find and learn. 👏 Contributions to this hub are welcome!

🆕 LWMs News

  • 2024/09/20: IBM and Nasa Prithvi-WxC Foundation model [link]
  • 2024/08/15: MetMamba, a DLWP model built on a state-of-the-art state-space model, Mamba, offers notable performance gains [link];
  • 2024/07/30: FuXi-S2S published in Nature Communications [link];
  • 2024/06/20: WEATHER-5K: A Large-scale Global Station Weather Dataset Towards Comprehensive Time-series Forecasting Benchmark [link];
  • 2024/05/24: ORCA: A Global Ocean Emulator for Multi-year to Decadal Predictions [link];
  • 2024/05/22: Generalizing Weather Forecast to Fine-grained Temporal Scales via Physics-AI Hybrid Modeling [link];
  • 2024/05/20: Aurora: A Foundation Model of the Atmosphere [link];
  • 2024/05/09: FuXi-ENS: A machine learning model for medium-range ensemble weather forecasting [link];
  • 2024/05/06: CRA5: Extreme Compression of ERA5 for Portable Global Climate and Weather Research via an Efficient Variational Transformer [link];
  • 2024/04/15: ClimODE: Climate and Weather Forecasting with Physics-informed Neural ODEs [link];
  • 2024/04/12: FuXi-DA: A Generalized Deep Learning Data Assimilation Framework for Assimilating Satellite Observations [link];
  • 2024/03/29: SEEDS: Generative emulation of weather forecast ensembles with diffusion models [link];
  • 2024/03/13: KARINA: An Efficient Deep Learning Model for Global Weather Forecast [link];
  • 2024/02/06: CasCast: Skillful High-resolution Precipitation Nowcasting via Cascaded Modelling [link];
  • 2024/02/04: XiHe, the first data-driven 1/12° resolution global ocean eddy-resolving forecasting model [link];
  • 2024/02/02: ExtremeCast: Boosting Extreme Value Prediction for Global Weather Forecast [link];
Expand to see more LWMs news
  • 2024/01/28: FengWu-GHR, the first data-driven global weather forecasting model running at the 0.09∘ horizontal resolution [link];
  • 2023/12/27: GenCast, a ML-based generative model for ensemble weather forecasting [link];
  • 2023/12/16: Four-Dimensional Variational (4DVar) assimilation, and develop an AI-based cyclic weather forecasting system, FengWu-4DVar [link];
  • 2023/12/15: FuXi-S2S: An accurate machine learning model for global subseasonal forecasts [link];
  • 2023/12/11: A unified and flexible framework that can equip any type of spatio-temporal models is proposed based on residual diffusion DiffCast [link];
  • 2023/11/13: GCMs are physics-based simulators which combine a numerical solver for large-scale dynamics with tuned representations for small-scale processes such as cloud formation. [link];
  • 2023/12/13: FuXi is open source [link];
  • 2023/11/14: GraphCast published in Science [link];
  • 2023/10/25: IBM and Nasa Prithvi-100M Model [link];
  • 2023/09/14: Pangu-Weather published in Nature [link];
  • 2023/08/25: ClimaX published in ICML 2023 [link];

🗂️ LWMs Lists

LWM name From Date(1st) Publication Links Model Licence Weights Licence
MetNet Google 2020.03 - [paper] [github] [MIT] N/A
FourCastNet NVIDIA 2022.02 PASC 23 [paper] [github] [BSD-3] [BSD-3]
MetNet-2 Google 2022.09 Nature Communications [paper] [github] [MIT] N/A
Pangu-Weather Huaiwei 2022.11 Nature [paper] [github] Not Specified [CC-BY-NC-SA 4.0]
GraphCast DeepMind 2022.12 Science [paper] [github] [Apache 2.0] [CC-BY-NC-SA 4.0]
ClimaX Microsoft 2023.01 ICML 2023 [paper] [github] [MIT] Not specificied ([MIT]?)
Fengwu Shanghai AI Lab 2023.04 - [paper] [github] Not Specified Not Specified
MetNet-3 Google 2023.06 - [paper] - -
FuXi Fudan 2023.06 npj 2023 [paper] [github] [Apache 2.0] [CC-BY-NC-SA 4.0]
NowcastNet Tsinghua 2023.07 Nature [paper] - -
AI-GOMS Tsinghua 2023.08 - [paper] - -
Prithvi-100M IBM / Nasa 2023.08 [paper] [Hugging Face] [Apache 2.0] [Apache 2.0]
FuXi-Extreme Fudan 2023.10 - [paper] - -
NeuralGCM DeepMind 2023.11 - [paper] - -
FengWu-4DVar Tsinghua 2023.12 ICML 2024 [paper] - -
FengWu-Adas Shanghai AI Lab 2023.12 - [paper] - -
FuXi-S2S Fudan 2023.12 Nature Communications [arXiv paper] [NC paper] - -
GenCast DeepMind 2023.12 - [paper] - -
DiffCast HITsz 2023.12 CVPR 2024 [paper] - -
FengWu-GHR Shanghai AI Lab 2024.01 - [paper] - -
ExtremeCast Shanghai AI Lab 2024.02 - [paper] [github] Not Specified Not Specified
XiHe NUDT 2024.02 - [paper] [github] Not Specified Not Specified
CasCast Shanghai AI Lab 2024.02 ICML 2024 [paper] [github] Not Specified Not Specified
KARINA KIST 2024.03 - [paper] - -
SEEDS Google 2024.03 Science Advances [paper] - -
FuXi-DA Fudan 2024.04 - [paper] - -
ClimODE Aalto University 2024.04 ICLR 2024 (Oral) [paper] [github] [MIT] Not specified or non applicable
FuXi-ENS Fudan 2024.05 - [paper] - -
Aurora Microsoft 2024.05 - [paper] [github] [MIT] [CC-BY-NC-SA 4.0]
WeatherGFT Shanghai AI Lab 2024.05 NeurIPS 2024 [paper] [github] Not Specified Not Specified
ORCA Shanghai AI Lab 2024.05 - [paper] [github] - -
MetMamba Beijing PRESKY Technology 2024.08 - [paper] Not Specified Not Specified
Prithvi-WxC IBM / Nasa 2024.09 [paper] [Hugging Face] [CDLA Permissive 2.0] [CDLA Permissive 2.0]

🗃️ Dataset Lists

Dataset name From Date(1st) Publication Links
WeatherBench Google 2020.02 JAMES 2020 [paper] [github]
ERA5 ECMWF 2020.05 - [paper] [link]
SEVIR MIT 2020.06 NeurIPS 2020 [paper] [github] [link]
WeatherBench2 Google 2023.08 - [paper] [github]
CRA5 Shanghai AI Lab 2024.05 - [paper] [github]
WEATHER-5K Beijing PRESKY Technology 2024.08 - [paper]

📖 Papers

WeatherBench

  • WeatherBench: A benchmark dataset for data-driven weather forecasting [pdf]
  • WeatherBench 2: A benchmark for the next generation of data-driven global weather models [pdf]

MetNet

  • MetNet: A Neural Weather Model for Precipitation Forecasting (MetNet) [pdf]
  • Deep learning for twelve hour precipitation forecasts (MetNet-2) [pdf]
  • Deep Learning for Day Forecasts from Sparse Observations (MetNet-3) [pdf]

FourCastNet

  • FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators (FourCastNet) [pdf]

Pangu-Weather

  • Accurate medium-range global weather forecasting with 3D neural networks (Pangu-Weather) [pdf]

GraphCast

  • Learning skillful medium-range global weather forecasting (GraphCast) [pdf]

ClimaX

  • ClimaX: A foundation model for weather and climate (ClimaX) [pdf]

FengWu

  • FengWu: Pushing the Skillful Global Medium-range Weather Forecast beyond 10 Days Lead (FengWu) [pdf]
  • FengWu-4DVar: Coupling the Data-driven Weather Forecasting Model with 4D Variational Assimilation [pdf]
  • Towards an end-to-end artificial intelligence driven global weather forecasting system [pdf]
  • FengWu-GHR: Learning the Kilometer-scale Medium-range Global Weather Forecasting [pdf]
  • ExtremeCast: Boosting Extreme Value Prediction for Global Weather Forecast [pdf]

FuXi

  • FuXi: A cascade machine learning forecasting system for 15-day global weather forecast (FuXi) [pdf]
  • FuXi-Extreme: Improving extreme rainfall and wind forecasts with diffusion model (FuXi-Extreme) [pdf]
  • FuXi-S2S: An accurate machine learning model for global subseasonal forecasts [pdf]
  • Fuxi-DA: A Generalized Deep Learning Data Assimilation Framework for Assimilating Satellite Observations [pdf]
  • FuXi-ENS: A machine learning model for medium-range ensemble weather forecasting [pdf]

AI-GOMS

  • AI-GOMS: Large AI-Driven Global Ocean Modeling System (AI-GOMS) [pdf]

XiHe

  • XiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting [pdf]

FNO

  • Fourier Neural Operator with Learned Deformations for PDEs on General Geometries [pdf]
  • SFNO: Spherical Fourier Neural Operators: Learning Stable Dynamics on the Sphere [pdf]

Nowcast

  • Earthformer: Exploring Space-Time Transformers for Earth System Forecasting [pdf]
  • PreDiff: Precipitation Nowcasting with Latent Diffusion Models [pdf]
  • DGMR: Skilful precipitation nowcasting using deep generative models of radar [odf]
  • Skilful nowcasting of extreme precipitation with NowcastNet (NowcastNet) [pdf]
  • DiffCast: A Unified Framework via Residual Diffusion for Precipitation Nowcasting [pdf]
  • CasCast: Skillful High-resolution Precipitation Nowcasting via Cascaded Modelling [pdf]
  • Generalizing Weather Forecast to Fine-grained Temporal Scales via Physics-AI Hybrid Modeling [pdf]

Physics-AI

  • Neural General Circulation Models for Weather and Climate [pdf]
  • ClimODE: Climate and Weather Forecasting with Physics-informed Neural ODEs [pdf]
  • Generalizing Weather Forecast to Fine-grained Temporal Scales via Physics-AI Hybrid Modeling [pdf]

Datasets

  • WeatherBench: A benchmark dataset for data-driven weather forecasting [pdf]
  • The ERA5 global reanalysis [pdf]
  • SEVIR : A Storm Event Imagery Dataset for Deep Learning Applications in Radar and Satellite Meteorology [pdf]
  • WeatherBench 2: A benchmark for the next generation of data-driven global weather models [pdf]
  • CRA5: Extreme Compression of ERA5 for Portable Global Climate and Weather Research via an Efficient Variational Transformer [pdf]
  • WEATHER-5K: A Large-scale Global Station Weather Dataset Towards Comprehensive Time-series Forecasting Benchmark [pdf]

More

  • Can deep learning beat numerical weather prediction? [pdf]
  • AtmoRep: A stochastic model of atmosphere dynamics using large scale representation learning [pdf]
  • Anthropogenic fingerprints in daily precipitation revealed by deep learning [pdf]
  • GenCast: Diffusion-based ensemble forecasting for medium-range weather [pdf]
  • KARINA: An Efficient Deep Learning Model for Global Weather Forecast [pdf]
  • SEEDS: Generative emulation of weather forecast ensembles with diffusion models [pdf]
  • Aurora: A Foundation Model of the Atmosphere [pdf]
  • ORCA: A Global Ocean Emulator for Multi-year to Decadal Predictions [pdf]

🚀 Code

  • ECMWF AI Models: AI-based weather forecasting models.
  • Skyrim: AI weather models united.
  • NVIDIA Earth2Mip: Earth-2 Model Intercomparison Project (MIP) is a python framework that enables climate researchers and scientists to inter-compare AI models for weather and climate.
  • AI Models for All: Run AI NWP forecasts hassle-free, serverless in the cloud!
  • OpenEarthLab: OpenEarthLab, aiming at developing cutting-edge Spatiaotemporal Generation algorithms and promoting the development of Earth Science.

🎉 Contributors