Annually, floods cause tens of thousands of fatalities worldwide, disrupt the lives of millions and cause financial damages in the billions. Accurate, actionable and reliable early warning systems must be developed and deployed in order to save both lives and livelihoods.

Since 2018, the Flood Forecasting team at Google Research has been working to apply advanced machine learning methods on a broad range of data sets, such as satellite imagery, river data and weather data. These data sets help to generate high quality flood forecasts that predict when rivers will overflow and the exact locations where flooding will take place. The team's long term vision is to empower governments, relief organizations and citizens to leverage these forecasts to take relevant actions and save lives. The flood alerts can be made accessible through Google Search, Google Maps and via notifications on Android mobile phones.

The system currently covers an area populated by more than 370 million people across several countries in South East Asia and Latin America. In 2021 the Flood Forecasting system reached 23 million people with more than 115 million notifications based on user location and flood maps. Flood Forecasting and alerting plans to expand globally in the near future to ensure everyone around the world can benefit from these early warnings.