Raw and processed CAMSELS-based data for testing discharge assimilation strategies in short-range AI-driven discharge forecasts
Authors/Creators
-
1.
Université Gustave Eiffel
Description
These data are from the CAMELS-US (Newman et al., 2015) and the CAMELS-FR (Delaigue et al., 2025) datasets. Only a subsample of the original datasets has been reformatted and dropped here for the sake of the experiments we performed in this preprint. It is an extended version of this Version v1. See the README.txt for a short idea on this data_paper's structure. The codes associated with it are indicated in the preprint.
Technical info
For the use of these data, you are invited to get the LSTM_vUGE and/or the SACSMA_vUGE. For any git cloning, make sure you check the branch hydro_uge . For the MLP-based model (MLP_vUGE), you should checkout the branch "main". The MLP-based model will be made available upon the publication of the associated paper. However, for any questions, we kindly invite readers to contact us. In all cases, a README.md is provided within each models for running instructions.
Files
data_paper.zip
Files
(19.4 GB)
| Name | Size | Download all |
|---|---|---|
|
md5:21bfaa17ab1f00d831b8a5c0a084579b
|
19.4 GB | Preview Download |
|
md5:3cc1af9fb1c92f3bbfa1faadc88a07ae
|
2.8 kB | Preview Download |
Additional details
Related works
- Is new version of
- Dataset: 10.5281/zenodo.16944643 (DOI)
- Is supplement to
- Preprint: 10.5194/egusphere-2025-4244 (DOI)
Funding
- Agence Nationale de la Recherche
Dates
- Updated
-
2026-04This is an extended version of the V1 published in August 2025, it is completed with CAMELS-FR datasets, hindcasts and forecast archives from ECMWF. The structure from V1 is changed from data/** to data_paper/data/camels[fr,us].
Software
- Repository URL
- https://gitlab.univ-eiffel.fr/bob.saint-fleur/ai_operational_hydroforecast#
- Programming language
- Python , Jupyter Notebook
- Development Status
- Active