A collection of articles on Large Weather Models (LWMs), to make it easier to find and learn. 👏 Contributions to this hub are welcome!
- 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];
| LWM name | From | Date(1st) | Publication | Links | Model Licence | Weights Licence |
|---|---|---|---|---|---|---|
| MetNet | 2020.03 | - | [paper] [github] | [MIT] | N/A | |
| FourCastNet | NVIDIA | 2022.02 | PASC 23 | [paper] [github] | [BSD-3] | [BSD-3] |
| MetNet-2 | 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 | 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 | 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 name | From | Date(1st) | Publication | Links |
|---|---|---|---|---|
| WeatherBench | 2020.02 | JAMES 2020 | [paper] [github] | |
| ERA5 | ECMWF | 2020.05 | - | [paper] [link] |
| SEVIR | MIT | 2020.06 | NeurIPS 2020 | [paper] [github] [link] |
| WeatherBench2 | 2023.08 | - | [paper] [github] | |
| CRA5 | Shanghai AI Lab | 2024.05 | - | [paper] [github] |
| WEATHER-5K | Beijing PRESKY Technology | 2024.08 | - | [paper] |
- 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: 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: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators (FourCastNet) [pdf]
- Accurate medium-range global weather forecasting with 3D neural networks (Pangu-Weather) [pdf]
- Learning skillful medium-range global weather forecasting (GraphCast) [pdf]
- ClimaX: A foundation model for weather and climate (ClimaX) [pdf]
- 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: 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: Large AI-Driven Global Ocean Modeling System (AI-GOMS) [pdf]
- XiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting [pdf]
- Fourier Neural Operator with Learned Deformations for PDEs on General Geometries [pdf]
- SFNO: Spherical Fourier Neural Operators: Learning Stable Dynamics on the Sphere [pdf]
- 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]
- 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]
- 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]
- 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]
- 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.