Momahmoses/ng-flood-evacuation-routing

||Real-time AI flood evacuation routing for Nigeria using LightGBM road risk scoring, NetworkX A* routing and population-priority dispatch for NEMA field coordinators. # Flood Evacuation Routing, Nigeria > Real-time AI-powered evacuation route optimizer for Nigerian flood emergencies, LightGBM road-segment risk scoring + NetworkX dynamic graph routing + population-weighted prioritization. Built for NEMA, state SEMAs, and Red Cross Nigeria field coordinators. --- ## The Problem Nigeria loses an average of **$4B/year** to flood damage, with 2022's Anambra/Kogi/Bayelsa floods displacing **1.4 million people** in a single season. The core evacuation challenge is not the water, it's the routing: - Roads become impassable at different flood depths (laterite roads fail at 10cm; paved highways at 60cm) - Emergency managers typically rely on paper maps and local intuition - A single wrong route choice can trap convoys in rising water - No system existed to dynamically re-route around cut-off bridges and submerged roads in real time --- ## Solution: Dynamic Flood-Aware Evacuation Router | Component | Technology | Function | |---|---|---| | **Road Risk Scorer** | LightGBM | Predicts passability of each road segment from flood depth, road type, elevation | | **Evacuation Router** | NetworkX + A\* | Finds optimal route on a dynamically penalized road graph | | **Safe Zone Finder** | Haversine + OSM | Identifies nearest elevated safe areas (schools, churches, LGA HQs) | | **Priority Ranker** | Population heatmap | Scores communities by vulnerability × isolation to dispatch convoys first | --- ## Coverage High-priority flood-prone states: **Kogi, Anambra, Bayelsa, Delta, Rivers, Cross River, Kebbi, Niger, Benue, Taraba** Road network sourced from OpenStreetMap Nigeria extracts (~280K road segments across 36 states + FCT). --- ## Evacuation Decision Pipeline ``` [Flood Alert Triggered] │ ▼ [SRTM Elevation + CHIRPS Rainfall → Flood Depth Estimation per segment] │ ▼ [LightGBM: Passability Score (0-1) per road segment] │ ▼ [NetworkX Graph: edge weights = travel_time / passability_score] │ ▼ [A* Search: origin community → nearest …