Lokoja ANN Flood Shield

Inspiration

In 2022, floods at the Niger-Benue confluence in Lokoja displaced over 600,000 people, destroying homes, schools, clinics, and livelihoods. This is not a one-time disaster—it happens almost every year, yet existing warning systems remain slow and unreliable.

Seeing communities repeatedly suffer preventable losses inspired the idea behind Lokoja ANN Flood Shield: a system that doesn’t just respond to floods, but predicts and prevents them using data and AI.

What it does

Lokoja ANN Flood Shield is an AI-powered early warning system that predicts flood risks 24–48 hours in advance.

It:

• Uses an Artificial Neural Network trained on 36 years of historical flood data
  
• Processes real-time inputs such as rainfall, river stage, and discharge
  
• Sends automated alerts via SMS and WhatsApp to communities and authorities
  
• Includes a “Resilience Branch” model that reinvests saved costs into flood-resistant infrastructure
  

The system transforms flood management from reactive disaster response to proactive prevention.

How we built it

The system was built using a combination of data science, AI, and automation tools:

• Historical flood data (1990–2025) used to train a feed-forward ANN model
  
• Integration of real-time environmental data streams
  
• A prediction engine for flood risk classification
  
• Automation tools for instant alert delivery
  
• A dashboard for monitoring and visualization
  

Workflow: Data → ANN Model → Prediction → Alerts → Community Action

Challenges we ran into

• Accessing and cleaning long-term historical flood data
  
• Integrating reliable real-time environmental data
  
• Designing alerts that are simple and actionable
  
• Ensuring the system works in low-connectivity environments
  
• Modeling the complex, nonlinear behavior of river systems
  

Accomplishments that we're proud of

• Achieved a projected ≥92% prediction accuracy
  
• Built a system capable of 24–48 hour early warnings
  
• Developed a fully automated alert pipeline
  
• Introduced a self-sustaining resilience funding model
  
• Combined AI with local knowledge for real-world impact
  

What we learned

• Flood systems are highly nonlinear and require advanced models
  
• Local data significantly improves prediction accuracy
  
• Fast communication is critical for saving lives
  
• Technology must be combined with community trust and engagement
  
• Long-term impact requires both prediction and resilience planning
  

What's next for Lokoja ANN Flood Shield

• Pilot deployment in high-risk communities in Lokoja
  
• Expansion to other flood-prone regions in Nigeria
  
• Integration of satellite data and IoT-based sensors
  
• Continuous improvement of model accuracy
  
• Scaling into a national AI-powered flood monitoring system
  

Built With

• artificial-neural-networks-(ann)
• climate-data-modeling
• data-analytics
• flood
• power-automate
• power-bi
• prediction
• python
• risk
• sms-apis
• tensorflow
• whatsapp-api