International Journal of Engineering Science
Abstract: This study presents a tech-based flood alert system in River Tana integrating ultrasonic sensors, predictive AI modeling, and SMS alerts. Developed incrementally with input from KenGen, KURA, and KeRRA, it achieved 85% predictive accuracy and fast dissemination (within 5 seconds). The low-cost design and stakeholder engagement ensure scalability. Limitations like flash flood prediction and sensor resilience are addressed with future improvements. The system bridges gaps in traditional forecasting and enhances resilience through real-time community-level preparedness tools.