README
A Global, AI-Ready Dataset for 3D Cloud Reconstruction
Cesar Aybar¹ • Shirin Ermis² • Lilli Freischem² • Stella Girtsou³⁴ • Kyriaki-Margarita Bintsi⁵
Emiliano Diaz Salas-Porras¹ • Michael Eisinger⁶ • William Jones² • Anna Jungbluth⁶ • Benoit Tremblay⁷
¹Universitat de València, ²University of Oxford, ³National Observatory of Athens, ⁴National Technical University of Athens,
⁵Harvard Medical School & Massachusetts General Hospital, ⁶European Space Agency, ⁷Environment and Climate Change Canada
About This Dataset
This repository contains the AI-ready datasets presented in the paper "Global 3D Reconstruction of Clouds & Tropical Cyclones". The dataset provides paired 2D multispectral geostationary satellite imagery with co-located 3D vertical cloud property profiles from CloudSat radar, organized into three components:
- Pre-training: Self-supervised learning on geostationary imagery (no labels)
- Fine-tuning: Supervised training with CloudSat vertical profiles as ground truth
- Tropical Cyclones: Benchmark dataset for model evaluation on intense storms
Dataset Structure
The dataset is organized by satellite and purpose:
How to Access
Pretraining (no CloudSat labels, 512×512):
Finetuning (with CloudSat labels, 256×256):
Benchmark:
Acknowledgements
This work was enabled by the Frontier Development Lab (FDL) Earth Systems Lab, a public-private partnership between the European Space Agency (ESA), Trillium Technologies, and the University of Oxford. We thank the teams behind the GOES, Himawari, and Meteosat satellites for their invaluable data. This research was supported by computational resources from Google Cloud.