Flood Vulnerability Assessment API

Global, real-time flood risk analysis powered by:

• Google Earth Engine (terrain)
• OpenStreetMap (water proximity)
• SHAP (explanations)
• Multi-hazard modeling (fluvial, coastal, pluvial)

Features

• Single location and batch CSV upload assessment
• 95% confidence intervals + uncertainty analysis
• Multi-Hazard detection (Fluvial, Coastal Surge, and Pluvial)
• Building height prediction using DL models
• Integration with Global Building Atlas

Tech Stack

• Backend: FastAPI + Python
• Frontend: Vanilla HTML/CSS/JavaScript
• Data Sources: GEE + OSM + Natural Earth
• ML: ONNX models for predictions
• Performance: @lru_cache for 100x batch speed

Project Structure

flood-vulnerability-api-copy/
├── api/                  # (Future: API module organization)
├── static/              
│   ├── css/
│   │   └── styles.css   # All frontend styles
│   └── js/
│       └── app.js       # Client-side application logic
├── templates/
│   └── index.html       # Main HTML template
├── height_predictor/    # Building height prediction modules
├── main.py              # FastAPI entry point
├── vulnerability.py     # Vulnerability calculation logic
├── spatial_queries.py   # GEE and OSM spatial queries
├── gee_auth.py          # Google Earth Engine authentication
└── explainability.py    # SHAP-based explanations

Quick Start

Local Development

# Install dependencies
pip install -r requirements.txt

# Run the development server
uvicorn main:app --reload

# Access the application
open http://localhost:8000

Try It Online

Live on Hugging Face

Example Coordinates

• 29.17, -95.31 → MODERATE risk
• 27.7, 86.7 → LOW risk

### Citation
Olagunju, S. O., Sharipova, A., Serikkyzy, A., Satybaldiyeva, D., Varol, H. A., & Karaca, F. (2026). Global flood vulnerability model: Building-level assessment using multi-source remote sensing. Remote Sensing, 18(9), 1425. https://doi.org/10.3390/rs18091425