A physics-informed hydrological forecasting system for the Upper Niger Basin, integrating Wflow SBM with Machine Learning for flood prediction.
Wflow-ML-Flood-Forecasting/
├── data/
│ ├── static/ # DEM, flow direction, slope (from SRTM)
│ └── processed/ # ERA5 daily forcing
├── notebooks/
│ ├── 00_static_data_acquisition.ipynb # Download SRTM, derive LDD
│ ├── 01_data_acquisition.ipynb # ERA5-Land forcing
│ └── 02_wflow_model_build.ipynb # Model inspection
├── models/wflow_sbm/wflow_niger/
│ ├── staticmaps.nc # Static model parameters
│ ├── inmaps.nc # ERA5 forcing (precip, temp, PET)
│ ├── cyclic_lai.nc # Monthly LAI climatology
│ └── wflow_sbm.toml # Wflow configuration
├── scripts/
│ ├── create_model_files.py
│ └── verify_model.py
└── requirements.txt
- Region: Upper Niger Basin (Guinea/Mali)
- Bbox: [-10.5°, 9.5°, -6.5°, 13.0°]
- Resolution: 0.1° (~10 km)
- Period: 2019-01-01 to 2020-02-29
| Dataset | Source | Resolution |
|---|---|---|
| DEM | CGIAR SRTM 90m | 90m |
| Flow Direction | Derived (pyflwdir) | 0.1° |
| Forcing | ERA5-Land | 0.1° daily |
| LAI | Synthetic climatology | Monthly |
pip install -r requirements.txtRun notebooks/00_static_data_acquisition.ipynb or:
python scripts/create_model_files.pyRequires Julia + Wflow.jl:
cd models/wflow_sbm/wflow_niger
julia -e "using Wflow; Wflow.run(\"wflow_sbm.toml\")"| # | Notebook | Description |
|---|---|---|
| 00 | 00_static_data_acquisition |
Download SRTM, derive flow direction |
| 01 | 01_data_acquisition |
ERA5-Land forcing download |
| 02 | 02_wflow_model_build |
Model inspection and visualization |
- Type: Wflow SBM (Soil-Bucket Model)
- Routing: Kinematic wave
- Timestep: Daily (86400 s)
- Snow: Disabled (tropical region)
q_river: River discharge (m³/s)satwaterdepth: Saturated zone depthustorelayerdepth: Unsaturated storage