Github: https://github.com/0deve/FloodSentry
Climate change is no longer a distant threat; it is a current reality. Recent devastating floods, such as those in Galati, Romania, have exposed a critical flaw in current infrastructure: traditional warning systems are too slow and generic. When alerts trigger only as water levels rise, we are already too late, leading to avoidable loss of life, destroyed infrastructure, and significant economic damage.
FloodSentry is a pan-European, proactive flood risk monitoring system that transforms raw Copernicus satellite data and Galileo positioning into impact-based flood forecasts. We move beyond generic weather alerts to provide a 3D Hydrological Digital Twin that identifies exactly which hospitals, schools, and communities are at risk, predicting precisely when the hazard will arrive.
FloodSentry bridges the gap between raw environmental data and actionable intelligence:
3D Digital Twin Visualization: Using Deck.gl, we render NUTS-2/NUTS-3 regions as 3D extruded polygons color-coded by real-time risk scores and populated with critical infrastructure markers.
The Hybrid ML Engine:
XGBoost Classifier: Predicts immediate risk (fluvial, pluvial, or snowmelt) based on soil moisture, rainfall, NDWI, and temperature.
GNN-Lite Risk Propagator: A Graph Neural Network-inspired algorithm that models how flood waves travel downstream through the hydrographic network over a 7-day period.
Standardized Emergency Response: The system automatically generates Common Alerting Protocol (CAP) XML files ready for integration with national systems like RO-Alert or EU-Alert.
Executive PDF Reports: Automatic generation of comprehensive reports for emergency services summarizing risk scores and specific infrastructure at risk.
FloodSentry is built on the backbone of the Copernicus and Galileo programs:
Copernicus Sentinel-1 (SAR): Enables flood extent mapping and soil moisture detection through cloud cover for all-weather monitoring.
Copernicus Sentinel-2 and 3: Monitors the Normalized Difference Water Index (NDWI), Fractional Snow Cover, and Snow Water Equivalent to predict melt-induced flooding.
Copernicus Services: We utilize the Copernicus Digital Elevation Model (DEM) for the 3D terrain foundation and Copernicus EMS historical data for training ML models.
Galileo High-Accuracy Positioning: Powers the Locate Me feature, allowing field rescue teams to instantly identify their specific NUTS region and local risk levels.
Copernicus Climate Change Service (ERA5): Integrates historical and forecasted meteorological data to drive predictive risk assessments.
FloodSentry directly addresses the EU Space for Water challenge by providing a scalable tool for mitigating the destructive extremes of the water cycle. By moving beyond general river alerts to local, infrastructure-specific impacts, we empower authorities to pre-deploy resources—like sandbags and rescue teams—before the flood wave arrives.
We are a multidisciplinary team of three experts leveraging space tech for climate resilience:
Liontescu Stefan, Full-Stack and System Architect: Responsible for the FastAPI backend architecture and the integration of the 3D Digital Twin visualization.
Pavel Stefan, ML and Data Scientist: Lead developer of the XGBoost hazard classifier and the Graph Neural Network (GNN) propagation logic.
Hutanu Emanuel, Geospatial and Space Tech Lead: Specialist in Copernicus data ingestion, SAR processing, and Galileo-based positioning services.