SNU’s Department of Computer Science has developed an AI system predicting catastrophic flooding in the Jubba Valley with 87% accuracy up to 72 hours ahead, in partnership with the IGAD Climate Prediction and Applications Centre.
n
Technical approach
n
The model uses an LSTM neural network trained on 30 years of hydrological and meteorological data. It was deployed on a pilot basis with the Somali Disaster Management Agency and successfully informed evacuation advisories during the November 2025 flooding events.
n
The research was published in Natural Hazards and Earth System Sciences. Lead researcher Mr. Warsame received the IGAD Young Scientist Award.