Project Details
Title:
Limitl8ss (Public Service)
Details:
Accra floods every year. In 2015, over 150 people died. In 2024, the Accra-Tema motorway was submerged, thousands were displaced, and deaths were recorded — more than a decade after the city pledged reform. The flooding has not stopped. The response has not improved enough.
Tools exist in fragments. Ghana's Meteorological Agency issues rainfall warnings. A flood risk app exists showing static maps of what could flood in a once-in-ten-year event. A government early warning system monitors water levels at sensor stations. But none of these answer the question an Accra resident actually needs answered at 5am before leaving for work: is my neighbourhood going to flood today?
No publicly accessible tool uses AI to combine real-time rainfall forecasts with terrain, drainage, and land cover data to predict neighbourhood-level flood probability on a daily basis in Accra. What exists is either static, institution-facing, or too general to be actionable.
Project ARK addresses this gap — delivering daily, AI-driven, neighbourhood-level flood risk predictions directly to residents, drivers, and NADMO officers through a mobile app and open platform, powered by data that is updated every 24 hours.
We aim to develop a machine learning model that predicts flood probability at 100m resolution across Greater Accra using terrain, land cover, drainage infrastructure, and antecedent rainfall features, Deliver flood risk intelligence through three purpose-built interfaces: a mobile application for residents and drivers, a desktop analytical dashboard for city planners and NGOs, and an operations dashboard for emergency responders, Publish an open REST API and JavaScript SDK enabling Ghanaian developers to build flood-aware applications on top of the platform,
Flooding is not an abstract risk in Ghana — it is an annual reality that kills people, destroys homes, and disrupts millions of livelihoods across Accra. Yet the people most affected — residents of Nima, Alajo, Agbogbloshie, Lapaz — have no reliable way to know, before they leave their homes in the morning, whether their street will be underwater by evening. Project ARK is built specifically for this reality. It uses Ghana's own rainfall data collected from 190 weather stations across the country, combined with satellite observations of where Accra has historically flooded, to build a prediction system designed around how Accra actually floods — not how European or American cities flood. The app is designed to work on the phones Ghanaians actually have, on the connections they actually have, in plain language they can act on. The operations dashboard is designed around how NADMO actually works. The open API is designed so Ghanaian developers can build on top of it — extending its reach without requiring the core team to do everything.