AI flood prediction could save lives as increasingly dangerous flooding threatens a fifth of the world

• More than one fifth of the world is at risk of significant flooding as waters rise.
• But artificial intelligence models could help save lives as conditions worsen in the coming years.
• AI flood models study not just the weather, but nearby terrain and history as well. 

Nearly two billion people are facing significant flooding risks amid the climate crisis — and experts only expect the problem to worsen with coastal flooding on track to skyrocket by 2100.

But as waters rise across the globe, scientists and researchers are increasingly turning to artificial intelligence to combat the growing problem, according to reports.

Flood monitoring firms from Iowa to Norway are using AI to combat damage from flooding by creating digital prediction models that could help save lives with early evacuation warnings and more accurate mapping data.

Cities around the world could see more than six feet of flooding by the turn of the century, and at least eight major American cities are at risk of disappearing almost entirely by 2100 including New Orleans, Miami, and New York City, Business Insider previously reported.

Low-lying coastal cities are especially at risk of flood damage brought on by rising seawaters, leaving much of Florida in a path of potential destruction.

A model flood

Physics-based flood models that rely on standard computing power were once the norm in flood prediction, according to the BBC.

But those models are prone to underestimating the real risk of incoming flooding, Northeastern Global News reported earlier this year, citing an incident in Waverly, Tennessee, in August 2021 when 20 people died after meteorologists predicted two to three inches of rain but wound up with more than 20.

There were at least 77 US flood fatalities in 2023, according to the National Weather Service. 

Artificial intelligence models, however, study not just the weather forecast but the nearby land and rivers, drainage information about the area, and regional layout to create more comprehensive predictive models.

According to the BBC, AI also has superior computing power, meaning models can do the work much quicker and with less human oversight.

One firm employing AI is 7Analytics, a Norway-based sustainability data platform focusing on climate risk management. The company already uses its AI model to provide real-time flooding predictions for local businesses and authorities, pinpointing at-risk areas, according to the BBC.

The company's founder Jonas Borland told the outlet that the model can predict up to seven days before a catastrophic weather event.

"All of this data takes into account where the water will flow, and where it will create problems," he said. "We can tell you that in five days you will have 50 cm of water at your entrance, and we can tell you when it will subside."

Torland told the outlet that donations and funding for the model have increased recently as awareness of flood risks has become more commonplace.

Big names

Neara, a London-based infrastructure modeling platform, uses AI to make digital simulations of floods that help electricity networks prepare for and respond to water damage, The BBC reported. Meanwhile, the University of Iowa also has an AI model called Flood AI.

Even Google has gotten in on the AI game, providing river flood warnings across more than 80 countries with its Flood Hub platform, which launched in the UK and the US earlier this year.

"Floods are one of the most devastating natural disasters, and impact hundreds of millions people per year," Yossi Matias, vice president of engineering and research and crisis response lead at Google, told the BBC "We really want to let people know before floods actually hit them."

The system uses satellite images to make pictures of rivers and nearby terrain, which AI then uses to predict how rivers could flood following rainfall.

Flood Hub researchers told the BBC that the system could send a warning anywhere from two to seven days before devastating weather hits.

In 2021 alone, Google sent 115 million flood alert messages to 23 million people, the company wrote in a sponsored Bloomberg article.

According to the BBC, AI models are especially helpful in places with limited historical weather data.

"The nature of these global models is they can actually learn from experiences happening in other rivers where you have more data and more historical experience, Matias told the BBC.

But AI models aren't a perfect solution, Amy McGovern, a computer scientist at the University of Oklahoma School of Meteorology, told the outlet. They are only as good as the data they collect; places that don't have a history of severe flooding, remain at risk as artificial intelligence races to gather enough information to predict the outcome of rare heavy rains there.