Metro Flood

Inspiration

Metro Flood predicts flooding at individual subway entrances up to 6 hours in advance using weather forecasts, flood maps, and historical data. Risk Forecasting: Station-specific severity levels (mild → severe). Action Dashboard: Alerts commuters, officials, and crews. Interventions: Deploy barriers, clear drains, stage pumps. Crowdsourced Reporting: Riders submit real-time flood updates.

How It Works Data: NOAA radar, FEMA flood zones, NYC DEP stormwater maps, MTA entrances, elevation + drainage data. Model: Machine learning classifier predicts flood probability at each entrance. Interface: Interactive map with live predictions, history, and action steps. Action Engine: Matches short-notice infrastructure (e.g., AquaDam) to vulnerabilities.

Key Challenges Integrating fragmented datasets. Achieving reliable prediction confidence. Ensuring proposed barriers/pumps are deployable in under 2 hours.

Achievements Built a working demo that connects predictions to real-world actions. Created a scalable framework for integration with MTA, DEP, and DOT. Launched a citizen reporting tool to bridge public + official data.

Lessons Learned Prediction is only useful if paired with rapid response. Hyperlocal, entrance-level analysis is critical. Collaboration across AI, civic tech, and emergency management drives resilience.