Google is making significant strides in improving flood forecasting capabilities by expanding its AI-powered riverine flood forecasting model to over 100 countries, now covering 700 million people. This expansion provides critical flood information to more communities, enhancing preparedness and response efforts globally. Utilizing advanced labeled data, additional training locations, and new model architectures, the system integrates weather forecasting inputs to deliver more precise riverine flood predictions. Google’s Flood Hub platform now incorporates 250,000 forecast points via “virtual gauges,” expanding accessibility to regions lacking reliable physical sensors. Researchers and partners can leverage the forecasts through the upcoming API and the Google Runoff Reanalysis & Reforecast (GRRR) dataset, with historical data available for analyses dating back to 1981.
This expanded flood forecasting system has proven invaluable in aiding disaster preparedness and response. During the 2023 Rio Grande do Sul floods in Brazil, Google collaborated with local authorities and NGOs to provide actionable data, facilitating rapid relief efforts. Similarly, in Africa and India, humanitarian organizations have utilized Google’s technology to mitigate flood impacts. As part of the UN’s Early Warnings for All initiative, Google continues to refine its AI solutions to address climate crises, aiming to provide critical, life-saving information to vulnerable communities worldwide.
