Dataset Floods South Sudan
- Source
- DataCite
- Published
- 2026
- Country
- South Sudan
- License
- Creative Commons Attribution 4.0 International
- Themes
- Natural Disasters, Remote Sensing & Satellites, Machine Learning & AI
About this dataset
This dataset contains the static and dynamic covariates used to train a flood detection model for South Sudan. Static covariates include terrain-derived variables (Digital Elevation Model, slope, curvature, HAND, and Topographic Wetness Index), while dynamic covariates consist of Sentinel-1 SAR and Sentinel-2 optical imagery temporally aligned with flood events reported by UNOSAT/OCHA HDX. Flood extent polygons were refined using SAR-based water segmentation through principal component analysis (PCA) and thresholding, generating high-resolution flood point patterns used as reference data for model training. The dataset integrates topographic, hydrological, and multi-sensor remote sensing information to support Bayesian flood susceptibility and flood occurrence modeling.
How the data was collected
Static covariates include terrain-derived variables (Digital Elevation Model, slope, curvature, HAND, and Topographic Wetness Index), while dynamic covariates consist of Sentinel-1 SAR and Sentinel-2 optical imagery temporally aligned with flood events reported by UNOSAT/OCHA HDX. The dataset integrates topographic, hydrological, and multi-sensor remote sensing information to support Bayesian flood susceptibility and flood occurrence modeling.