---
task_categories:
- time-series-forecasting
tags:
- climate
- time-series
- satellite-imagery
license: unknown
---

# ClimateBench-M: A Multi-modal Climate Data Benchmark 

ClimateBench-M is a multi-modal climate benchmark designed to support the development of artificial general intelligence (AGI) in climate applications. It aligns data across three critical modalities at a unified spatio-temporal resolution:

1.  Time-series climate variables from ERA5.
2.  Extreme weather event records from NOAA.
3.  Satellite imagery from NASA HLS.

*   [**ClimateBench-M-TS**](https://huggingface.co/datasets/Violet24K/ClimateBench-M-TS): Climate time series with extreme events aligned and labeled.
*   [**ClimateBench-M-IMG**](https://huggingface.co/datasets/Violet24K/ClimateBench-M-IMG): Satellite imagery data
*   [**📖 Paper**](https://arxiv.org/abs/2404.00225)

If you find this repository useful in your research, please consider citing the following paper:

```
@inproceedings{DBLP:conf/kdd/ZhengJLTH24,
  author       = {Lecheng Zheng and
                  Baoyu Jing and
                  Zihao Li and
                  Hanghang Tong and
                  Jingrui He},
  editor       = {Ricardo Baeza{-}Yates and
                  Francesco Bonchi},
  title        = {Heterogeneous Contrastive Learning for Foundation Models and Beyond},
  booktitle    = {Proceedings of the 30th {ACM} {SIGKDD} Conference on Knowledge Discovery
                  and Data Mining, {KDD} 2024, Barcelona, Spain, August 25-29, 2024},
  pages        = {6666--6676},
  publisher    = {{ACM}},
  year         = {2024},
  url          = {https://doi.org/10.1145/3637528.3671454},
  doi          = {10.1145/3637528.3671454},
  timestamp    = {Sun, 08 Sep 2024 16:05:58 +0200},
  biburl       = {https://dblp.org/rec/conf/kdd/ZhengJLTH24.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}
```