AI-Augmented Early Warning System — Mozambique's INGD/INAM Pilot for Cyclone and Flood Forecasting

Following Cyclone Idai in March 2019, which caused 603 deaths and roughly US$3 billion in damage, Mozambique's National Institute for Disaster Risk Management and Reduction (INGD) and National Meteorological Institute (INAM) built a multi-hazard early-warning system with World Bank support, under the US$265 million Disaster Risk Management and Resilience Program. Satellite- and radar-based forecasts from INAM are disseminated through 70 community radio stations and trained community brigades that evacuate at-risk households ahead of a storm. Since approximately 2023-2025, INGD and INAM have layered a machine-learning and satellite-data-analytics forecasting pilot onto this system, in partnership with UNDP, the World Bank, the Norwegian Meteorological Institute and the Green Climate Fund, aiming to shorten the time between hazard detection and public alert for floods, cyclones and droughts.

Cyclone Freddy's 2023 death toll in Mozambique was 183, compared with 603 for Cyclone Idai in 2019 — a reduction attributed to the broader multi-hazard early-warning system that predates the AI/ML layer. The AI/ML forecasting component itself remains at pilot stage, with promising but not yet fully proven gains in flood-forecast accuracy; donor-funded forecasting improvements have not always translated into consistent warning dissemination, given institutional-coordination and financing-continuity constraints.

Mozambique has committed to a 2027 target for full 'Early Warnings for All' coverage, but the specific contribution of the AI/ML layer to outcomes has not yet been separately and rigorously evaluated.