Tech giant Google has launched Groundsource, an AI system designed to predict flash floods in urban areas up to 24 hours in advance.
According to Google, Groundsource uses Gemini to analyze decades of public reports, identifying more than 2.6 million historical flood events across over 150 countries.
The system then relies on Google Maps to determine precise geographic boundaries for each event, helping create a dataset focused specifically on flash floods.
“For years, as part of Google’s Crisis Resilience efforts, we’ve provided early warnings about natural hazards to help communities stay safe,” Google CEO Sundar Pichai said.
“However, high-fidelity data for certain disasters like flash floods simply did not exist. This data gap has long prevented our ability to train AI models to predict flash floods before they happen — until now.”
Google Unveils Groundsource, AI methodology To Forecast Flash Floods
The AI-powered methodology transforms public information into a high-quality record of historical disaster data.
It curates flood details by analyzing available news reports, and transforms public information into a structured, localized event archive covering more than 150 countries and spanning from the year 2000 to the present.
The core innovation of Groundsource lies in its ability to leverage advanced AI to extract signals from global news media.
Google is tapping artificial intelligence to make sense of the vast amounts of unstructured data on historical floods, including news articles, government reports, and local bulletins.
The methodology first isolates flood-related content from reports in 80 languages, converting everything into English using Google’s Cloud Translation API.
The critical extraction is then handled by the Gemini Large Language Model (LLM), guided by a sophisticated prompt that ensures rigorous verification.
Gemini classifies reports to distinguish actual, past, or ongoing floods from articles discussing future warnings or policy meetings.
It applies temporal reasoning to pinpoint event dates, translating references like “last Tuesday” into precise times, and ensures spatial accuracy via a map.
Google said the expansion aims to enhance community preparedness by providing timely warnings before floods strike, particularly in urban areas that previously lacked historical flash-flood data.
Also Read: Google Launches Gemini Pro Plan Giving Free Access to Students in Kenya
Turning Public Reports into Actionable Flood Data
The urban flash flood forecasts are available in Google’s Flood Hub — along with Google’s existing riverine flood forecasts, which cover 2 billion people in more than 150 countries for the most significant riverine floods
The model and dataset are now part of Google Earth AI’s suite of geospatial tools.
Using the same AI-driven approach as Groundsource, the model converts verified public reports into actionable datasets, helping communities, partners, and scientists scale their impact.
Google says this approach has the potential to extend beyond floods, with applications for other natural disasters such as landslides and heat waves.
For communities around the world, the tech giant noted that the move means better preparedness before disasters strike.
For its partners and scientists, Groundsource provides a massive, open-source benchmark to scale their impact — particularly in urban regions that have lacked historical flash flooding data.
By transforming public information into predictive data, the company aims to build a more resilient future where no one is caught off guard by extreme weather events.
Also Read: Step-by-Step Guide on How to Recover Your Google Account
Flash Floods Remain Deadly, as Gaps in Global Data Exist
According to the World Meteorological Organization (WMO), flash floods cause approximately 85% of flood-related fatalities, resulting in over 5,000 deaths annually.
Limitations of Existing Data: Satellite-based databases, such as the Global Flood Database (GFD) and the Dartmouth Flood Observatory (DFO), are limited by cloud cover, satellite revisit times, and a bias toward long-lasting events.
Scale of the Deficit: The Global Disaster Alert and Coordination System (GDACS) provides an inventory of roughly 10,000 high-impact events. This volume is insufficient for training global-scale predictive models.
