Using AI to expand global access to reliable flood forecasts
Not a pilot. Google deployed ML-powered flood forecasts across 80+ countries—extending warning time from zero to five days.
Why it matters
Google's machine learning system is solving a massive real-world problem at scale: providing reliable flood forecasts to 1.5 billion people in data-scarce regions. This demonstrates how AI can be operationalized for humanitarian impact while advancing climate resilience globally.
The key facts
10 to know$50 billion in annual flood damages worldwide
1.5 billion people (19% of world population) exposed to severe flood risk
Flood-related disasters doubled since 2000
ML extends forecast reliability from 0 to 5 days average lead time
Coverage expanded to 80+ countries via Flood Hub
Up to 7-day advance river forecasts now available
5,680 streamflow gauges used for model training (1980-2023)
LSTM-based model matches GloFAS nowcast accuracy at 4-5 day lead times
Research published in Nature (peer-reviewed)
Collaboration with WMO, Red Cross, Yale, JKU Institute for Machine Learning
Go to the source
Google Research Blogblog.research.google
Publisher excerpt: Posted by Yossi Matias, VP Engineering & Research, and Grey Nearing, Research Scientist, Google Research Floods are the most common natural disaster, and are responsible for roughly $50 billion in annual financial damages worldwide. The rate of flood-related disasters has more than doubled since…
