Google’s AI correctly predicts floods up to seven days early

You may have heard plenty about Google’s various generative AI products like Circle to Search, its AI wallpaper tool, Search Generative Experience (SGE), and more. But you may not have heard nearly as much about the other ways it is using AI, such as predicting floods. Recently, Google was able to accurately predict flooding up to seven days in advance thanks to machine learning (ML).

Today, Google announced it was able to significantly improve global-scale flood forecasting with the help of machine-learning technologies. According to the firm, it was able to improve the reliability of “global nowcasts, on average, from zero to five days.” In some cases, it was even able to predict floods a full week out before they happened.

One of the reasons why it can be difficult to predict floods ahead of time is the fact that most rivers don’t have what’s called a streamflow gauge. These gauges help provide relevant data, like precipitation and physical watershed information. However, the tech giant was able to get around this problem by feeding its ML technology on all available river data and applying the ML model to basins where no data was available.

Lending further credence to how impressive of an accomplishment this is, the company’s findings were published in Nature. For context, Nature is a leading multidisciplinary science journal that publishes peer-reviewed research.

The ultimate goal of the technology is to scale the accuracy of flood forecasting to a global level, even in areas where local data is not available. Google has been able to provide forecasts to over 80 countries through its Flood Hub. It also delivers alerts on Google Search, Maps, and Android notifications.