Japan's AMAGOI Project Aims to Trigger Rain at Sea to Cut Flood Damage on Land

Japan is developing an experimental project to artificially trigger heavy rainfall over the ocean in an effort to reduce the torrential downpours that increasingly batter the country's coastline, under a Cabinet Office Moonshot Research and Development Program goal targeting weather control by 2050. The project, called AMAGOI, is led by Professor Shunji Kotsuki of Chiba University's Institute for Advanced Academic Research and Center for Environmental Remote Sensing.

Japan's torrential rains, driven by typhoons and stationary linear rain bands, have caused flood damage exceeding 1 trillion yen in severe years and can result in loss of life, while conventional countermeasures such as flood-control dams and levees have clear limits, Kotsuki said. Because the most damaging rainstorms are driven primarily by large inflows of water vapor from the ocean, the project's approach is to artificially induce cumulonimbus cloud formation over the sea, triggering heavy rainfall in advance to reduce the amount of water vapor that reaches land.

Numerical simulations conducted so far suggest it may be possible to reduce torrential rainfall by 10% to 20%, according to Kotsuki. Based on those results, the team is planning an experimental artificial rainfall trial over the ocean in which dry ice would be dispersed from an aircraft to serve as a nucleus for water vapor and induce precipitation, aiming to reduce the water vapor that ultimately flows onto land. The team is also exploring the feasibility of an offshore dome structure roughly 600 meters wide and 300 meters high, designed to redirect wind flow and induce rainfall over the ocean, working with a manufacturing partner to assess the engineering requirements.

One of the project's biggest technical challenges is accurately reproducing current weather conditions within numerical models, which requires assimilating real-time observational data into simulations of the evolving atmosphere — a process Kotsuki said will make extensive use of AI. Today's typhoon track forecasts, issued every three hours up to 24 hours ahead and every six hours up to 120 hours ahead, are calculated by running 21 different scenarios and statistically processing the results, but increasing the number of scenarios sharply raises computational cost and processing time.

Kotsuki said an AI-based approach could replace much of that heavy computation, potentially calculating 1,000 scenarios in a short time and updating forecasts every 30 minutes, and could significantly improve predictions of linear precipitation bands, which are currently very difficult to forecast accurately. He said only about 5% to 6% of available meteorological data is currently used in operational weather prediction, and that incorporating more of the unused observational information could make forecasts more accurate.