Google Research has unveiled groundbreaking advancements in flood forecasting through the utilization of artificial intelligence (AI), as detailed in a recent Nature publication.
This innovative approach enables the accurate prediction of riverine flooding up to seven days in advance, spanning over 80 countries globally, including regions with limited data availability.
In a blog post, Yossi Matias, Google’s Vice President of Engineering & Research and Crisis Response Lead, emphasized the significance of providing reliable information during crises, particularly concerning floods, which pose a substantial risk to nearly 1.5 billion people worldwide.
Historically, the complexity of flood forecasting, coupled with data scarcity, hindered accurate predictions, exacerbating the vulnerability of communities, especially in developing nations.
“Through AI-based global hydrologic technologies, Google Research has revolutionized flood forecasting, significantly surpassing existing methods. By leveraging machine learning algorithms, Google’s models extend the reliability of global forecasts from zero to five days on average, with particularly notable improvements in regions of Africa and Asia,” the VP noted.
The journey toward achieving this milestone involved pivotal moments, including initial pilots in flood-prone regions like Bihar, India, where Google collaborated with local authorities to integrate flood forecasts into Google Public Alerts. Subsequent collaborations with academic researchers and the implementation of Long Short-Term Memory networks (LSTMs) further enhanced predictive capabilities.
“Recognizing the limitations of relying solely on local data, Google pivoted towards a global AI-based model, culminating in the launch of the Flood Hub platform in 2022.”
This platform initially provided forecasts in 20 countries and expanded to an additional 60 countries by 2023, covering over 460 million people worldwide.
Notably, access to flood forecasting in Africa has become comparable to that in Europe, marking a significant milestone in global disaster preparedness.
“Continued partnerships with international aid organizations, governmental agencies, and scientific communities underscore Google’s commitment to advancing flood forecasting and addressing climate-related challenges,” he added.
