Google Rolls Out AI-Powered Flood Forecasting In India, Says Head Of Tel Aviv R&D Center

Google is harnessing artificial intelligence tech to create forecasting models that can better predict when and where floods will occur, and it has partnered with India’s Central Water Commission (CWC) to roll out early warnings in Google Search in the subcontinent, Google VP of engineering and the managing director of Google’s R&D Center in Tel Aviv, Yossi Matias, announced this week.

Deadly floods are common in some parts of India, especially during monsoon season, which runs from July to September every year. In August, India’s southern state of Kerala experienced the worst flood in the region in nearly 100 years, with over 400 killed and more than a million people displaced. A number of other areas in India have seen more devastating floods over the past decade, with death tolls running into the thousands. The 2004 tsunami is still the worst water-related natural disaster to have occurred in the country, with over 10,000 lives claimed in India alone.

“Floods are devastating natural disasters worldwide — it’s estimated that every year, 250 million people around the world are affected by floods, also costing billions of dollars in damages,” Matias wrote in a blog post. Existing warning systems can be inaccurate and uninformative while being wholly unavailable in some areas, “resulting in far too many people being underprepared and unaware before a flood happens,” he added.

SEE ALSO: Google’s Data-Light Search ‘Go’ Can Now Read Out The Web, Powered By AI Tech Developed In Tel Aviv

Google is now “using AI and significant computational power to create better forecasting models that predict when and where floods will occur, and incorporating that information into Google Public Alerts,” to help improve preparedness for impending floods, he wrote.

The tech giant feeds a number of elements – past events, river readings, elevation calculations – into its models to generate maps and “run up to hundreds of thousands of simulations in each location,” Matias explained.

“With this information, we’ve created river flood forecasting models that can more accurately predict not only when and where a flood might occur, but the severity of the event as well,” he said.

The partnership with India’s CWC was first announced in June by the agency. Under the terms of the agreement, the CWC would use “state-of-the-art advances made by Google in the field of Artificial Intelligence, Machine Learning and geospatial mapping for effective management of water resources particularly in the field of flood forecasting and dissemination of flood-related information to the masses widely using the dissemination platforms developed by Google.”

The CWC said in a statement that until recently, it was disseminating flood levels with maximum lead time of one day, but the cooperation with Google would allow for a lead time of up to three days.

The collaborative arrangement, the CWC said, is likely to save millions of rupees “which otherwise would have to be spent by the government on acquiring high-resolution DEM [digital elevation models], high-end computational resources and developing dissemination platforms widely used by the masses.”

A flood alert is shown to users in Patna. Photo via Google

“This would enable the [Indian] government as well as disaster management organizations to identify well in advance the locations and population, which are at risk from floods and require warnings and information,” the CWC added, calling the initiative a “milestone in flood management and in mitigating the flood losses.”

Matias said the new project was rolled out in India first, “where 20 percent of global flood-related fatalities occur.” The first alert went out earlier this month in the Patna region in northeast India which experienced heavy rainfall.

Google is looking to expand this service to other countries to help save lives, Matias wrote.

SEE ALSO: Need A Dinner Reservation? Israeli Tech Powers Google’s New Human-Sounding Assistant To Book It

Google had another major project emanating from India. Late last month, it announced that its Go app, a lightweight, faster version of its search function for devices with little space and unreliable internet connections, can now read articles and web content out loud in 28 supported languages. The tech was partly developed in Israel and was based on user research in India. It launched this year in India and Indonesia.

The technology is powered by natural language processing and speech synthesis AI. It can read aloud billions of web pages in a natural sounding voice, using minimal cellular data, Matias explained in a post last month, co-written by Simon Tokumine, a Google senior product manager.

Google said in a statement that the new feature was “inspired by user research in India, where we heard from people how important it is to understand information effortlessly. Especially for people coming online for the first time, consuming long-form text on a small device can be difficult and time-consuming. With this new feature, you can just press play and follow along.”

With this technology, “consuming long-form text becomes as easy as watching TV or listening to the radio,” and can be helpful when multi-tasking (cooking, exercising, driving), Matias and Tokumine explained.

The feature can also be useful for people with visual impairments and those wanting to learn new languages, as the text on a given webpage is highlighted when read aloud, allowing users to follow along. Image descriptions in articles are being considered for future Go versions.