OpenCastKit
Open-source solutions of global data-driven high-resolution weather forecasting.
https://github.com/hfailab/opencastkit
Category: Atmosphere
Sub Category: Meteorological Observation and Forecast
Keywords
foucastnet weather-forecast
Last synced: about 1 hour ago
JSON representation
Repository metadata
The open-source solutions of FourCastNet and GraphCast
- Host: GitHub
- URL: https://github.com/hfailab/opencastkit
- Owner: HFAiLab
- License: mit
- Created: 2022-06-09T11:21:34.000Z (almost 3 years ago)
- Default Branch: master
- Last Pushed: 2023-04-18T02:57:00.000Z (about 2 years ago)
- Last Synced: 2025-04-17T22:07:16.209Z (10 days ago)
- Topics: foucastnet, weather-forecast
- Language: Python
- Homepage:
- Size: 3.38 MB
- Stars: 339
- Watchers: 6
- Forks: 89
- Open Issues: 24
- Releases: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
README.md
OpenCastKit: an open-source solutions of global data-driven high-resolution weather forecasting
简体中文 | English
本项目是由幻方AI团队复现优化，并开源的全球AI气象预报模型工具库。基于 FourCastNet 和 GraphCast 的论文，我们构建了一个新的全球AI气象预报项目——OpenCastKit，它能够与欧洲中期天气预报中心（ECMWF）的传统物理模型——高分辨率综合预测系统（IFS），进行直接比较。
我们将基于1979年1月到2022年12月的ERA5数据训练出来的模型参数开源到 Hugging Face 仓库中，并上线了一个每日更新的 HF-Earth，展示模型的预测效果。
下面是一些预测案例：
依赖
- hfai >= 7.9.5
- torch >=1.8
训练
原始数据来自欧洲中期天气预报中心（ECMWF）提供的一个公开可用的综合数据集 ERA5 ，需要通过 data_factory/convert_ear5_hourly.py 脚本提出数据特征，转化为高性能训练样本格式 FFRecord 下的样本数据。
训练 FourCastNet
本地运行：
python train_fourcastnet.py --pretrain-epochs 100 --fintune-epochs 40 --batch-size 4
也可以提交任务至幻方萤火集群，使用96张A100进行数据并行训练
hfai python train_fourcastnet.py --pretrain-epochs 100 --fintune-epochs 40 --batch-size 4 -- -n 12 --name train_fourcastnet
训练 GraphCast
本地运行：
python train_graphcast.py --epochs 200 --batch-size 2
也可以提交任务至幻方萤火集群，使用256张A100进行流水线并行训练
hfai python train_graphcast.py --epochs 200 --batch-size 2 -- -n 32 --name train_graphcast
引用
@article{pathak2022fourcastnet,
title={Fourcastnet: A global data-driven high-resolution weather model using adaptive fourier neural operators},
author={Pathak, Jaideep and Subramanian, Shashank and Harrington, Peter and Raja, Sanjeev and Chattopadhyay, Ashesh and Mardani, Morteza and Kurth, Thorsten and Hall, David and Li, Zongyi and Azizzadenesheli, Kamyar and others},
journal={arXiv preprint arXiv:2202.11214},
year={2022}
}
@article{remi2022graphcast,
title={GraphCast: Learning skillful medium-range global weather forecasting},
author={Remi Lam, Alvaro Sanchez-Gonzalez, Matthew Willson, Peter Wirnsberger, Meire Fortunato, Alexander Pritzel, Suman Ravuri, Timo Ewalds, Ferran Alet, Zach Eaton-Rosen, Weihua Hu, Alexander Merose, Stephan Hoyer, George Holland, Jacklynn Stott, Oriol Vinyals, Shakir Mohamed, Peter Battaglia},
journal={arXiv preprint arXiv:2212.12794},
year={2022}
}
Owner metadata
- Name: High-Flyer AI
- Login: HFAiLab
- Email:
- Kind: organization
- Description: 幻方AI
- Website: https://www.high-flyer.cn/
- Location: China
- Twitter:
- Company:
- Icon url: https://avatars.githubusercontent.com/u/90186689?v=4
- Repositories: 23
- Last ynced at: 2024-04-15T04:57:52.410Z
- Profile URL: https://github.com/HFAiLab
GitHub Events
Total
- Watch event: 36
- Fork event: 13
Last Year
- Watch event: 36
- Fork event: 13
Committers metadata
Last synced: 6 days ago
Total Commits: 15
Total Committers: 1
Avg Commits per committer: 15.0
Development Distribution Score (DDS): 0.0
Commits in past year: 7
Committers in past year: 1
Avg Commits per committer in past year: 7.0
Development Distribution Score (DDS) in past year: 0.0
| Name | Commits | |
|---|---|---|
| VachelHU | v****u@q****m | 15 |
Committer domains:
- qq.com: 1
Issue and Pull Request metadata
Last synced: 1 day ago
Total issues: 47
Total pull requests: 2
Average time to close issues: about 1 hour
Average time to close pull requests: less than a minute
Total issue authors: 20
Total pull request authors: 1
Average comments per issue: 1.7
Average comments per pull request: 0.0
Merged pull request: 2
Bot issues: 0
Bot pull requests: 0
Past year issues: 0
Past year pull requests: 0
Past year average time to close issues: N/A
Past year average time to close pull requests: N/A
Past year issue authors: 0
Past year pull request authors: 0
Past year average comments per issue: 0
Past year average comments per pull request: 0
Past year merged pull request: 0
Past year bot issues: 0
Past year bot pull requests: 0
Top Issue Authors
- lk1983823 (4)
- zhrli (4)
- liuweishuo (4)
- morestart (4)
- oubahe (3)
- Afshinshafei (2)
- ChongjianGE (2)
- HaiXiang-Zhao (2)
- zongyi-li (2)
- yingtaoluo (2)
- Jeffrey-JDong (2)
- CXL-edu (2)
- chunniunai220ml (2)
- AlexJJJChen (2)
- vrcsds (2)
Top Pull Request Authors
- VachelHU (2)
Top Issue Labels
Top Pull Request Labels
Score: 5.8944028342648505