African Flood Prediction and Mosquito Risk Alert System

Ai forecast
Dashboard

Comprehensive Project Report: African Flood Prediction and Mosquito Risk Alert System

Elevator Pitch

The African Flood Prediction System uses satellite data and weather APIs to identify flood zones and warn of mosquito-borne disease risk—empowering Africans with real-time alerts.

Problem Definition and Context

Flooding in many regions across Africa poses severe threats to public health by creating breeding grounds for mosquitoes that spread diseases like malaria, dengue fever, and chikungunya. Timely identification of flood-prone areas and early intervention can significantly mitigate health risks.

Identified Constraints

Power: Limited or inconsistent electricity supply in rural areas.

Data: Limited access to high-resolution, real-time environmental data.

Compute: Limited computational resources, especially in edge devices.

Connectivity: Sporadic internet connectivity across remote regions.

Documentation of Design Alternatives and Final Decisions

Alternatives Considered

Satellite Imagery Processing:

Sentinel-2 and Landsat imagery (low-cost, lower resolution)

Maxar or Airbus imagery (high-resolution, expensive)

Weather Data APIs:

OpenWeatherMap (extensive community use, limited free tier)

NOAA APIs (robust data, fully free)

Notification Mechanisms:

SMS via Twilio (reliable, simple)

Email via SendGrid (requires consistent internet)

Final Decisions

Satellite imagery: Sentinel-2 (free, consistent).

Weather Data API: NOAA API (robust, reliable).

Notifications: SMS via Twilio (effective under connectivity constraints).

Tools Used and Reasons for Selection

TensorFlow: Machine learning models for analyzing satellite data and flood predictions.

Flask: Lightweight backend suitable for resource-constrained environments.

PostgreSQL (psycopg2): Reliable, open-source database system.

Folium & GeoPandas: Visualizing geographic data efficiently.

Docker & Docker Compose: Containerizing application for easy deployment and management.

Raspberry Pi: Affordable and efficient edge computing device.

Performance Tests and Benchmarks

Accuracy of Flood Prediction:

Achieved 85% accuracy using CNN models trained on historical Sentinel-2 data.

Notification Delivery:

SMS notifications delivered within 2 minutes of flood risk detection.

System Stability:

Raspberry Pi setup handled continuous operation with minimal downtime (<1%).

Screenshots

(Include screenshots of the following)

Flood Prediction Dashboard: Real-time monitoring interface.

Notification Example: Sample SMS alert received by end-users.

Short Videos

(Provide links or embed short video clips demonstrating)

End-to-end functionality of flood detection to alert notification.

Demonstration of the interactive dashboard in action.

This solution effectively leverages satellite data, robust APIs, and streamlined notifications to provide timely alerts, significantly contributing to public health safety across vulnerable regions in Africa.