Google's AI model can predict floods 7 days before the disaster strikes

The forecasting system provides alerts through multiple channels such as Google Search, Maps, Android notifications and via the Flood Hub.

Can we predict floods a few days before it actually hits? Google said that it’s artificial intelligence (AI) model can predict the disaster seven days in advance.

In a paper published in science journal Nature, the tech major said it combines two AI models, which process publicly available data to make the prediction. The “hydrologic model forecasts the amount of water flowing in a river, and the inundation model predicts what areas will be affected and how high the water level will be”.

Floods are one of the most common natural disasters and causes financial damages worth over $50 billion annually. Climate change has further accelerated the incidents of floods since 2000, which affects around 19 percent of the world population.

ALSO READ: Bringing Gemini to iPhones would fix Apple’s and Google’s AI problems

A reliable flood forecasting system can save millions of dollars and thousands of lives every year. Google, which started its research to predict floods in 2017, trained its machine learning models through historical data, terrain data, river level readings, elevation, running simulations and generating localized maps for each location.

The forecasting system provides alerts through multiple channels such as Google Search, Maps, Android notifications and via the Flood Hub.

Though the AI-powered approach has improved the reliability of global flood forecasts and extended the average prediction time from zero to five days, the US-based firm said it needs more data to scale. “In order to scale globally, especially in places where accurate local data is not available, more research advances were required.”

The technology has been particularly helpful in African and Asian countries, where the data is limited. The company used AI and ML models to use the available data and predict floods.

“In 2018, we launched a pilot early warning system in the Ganges-Brahmaputra river basin in India, with the hypothesis that ML could help address the challenging problem of reliable flood forecasting at scale. The pilot was further expanded the following year via the combination of an inundation model, real-time water level measurements, the creation of an elevation map and hydrologic modeling,” it said in its blog.