FLOODCAST: interim results on labels, transfer and spatial skill for flash-flood hazard in small Slovenian catchments
Technical note recording interim, not peer-reviewed results of FLOODCAST, a machine-learning flash-flood first guess under development at the Slovenian Environment Agency (ARSO). It accompanies the poster "Beyond forecast precipitation? What ECMWF forecasts and EFAS reanalysis add to flash-flood warning in small Alpine catchments" (doi:10.5281/zenodo.24284170). It reports skill at ARSO's warning units, the effect of label defects, transfer to unseen basins, skill in time versus space, the effect of added labels versus model architecture, and emergency-service records as labels. A journal paper is in preparation, and its numbers supersede these.
Authors
- Aman Arora (ORCID: https://orcid.org/0000-0003-2547-4424)
- Sašo Petan (ORCID: https://orcid.org/0000-0003-2916-1097)
Institutions
- University of Nova Gorica (SI)
- Slovenian Environment Agency (SI)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-16
- DOI
- https://doi.org/10.5281/zenodo.24284668
- Primary Topic
- Flood Risk Assessment and Management
- Type
- article
- Field-Weighted Citation Impact
- 0.00