Google is Expanding its AI-based Flood Warning System in India and Bangladesh

While we now live in an age where Google can also now predict the possibility of floods for us, all with the help of machine learning that pinpoints areas which are prone to flooding or alerts the users living there about the upcoming trouble, this service now covers all of India and is even working in extended parts of Bangladesh as well - right where it indeed is required.

Google first came up with the idea of flood alert service for the Patna region of India exclusively in 2018, but as the monsoon season in the sub-continent causes a great amount of distress in the form of floods every year, the company has worked on increasing the coverage by taking help from the local government.

This year in June, Google was able to reach the milestone of covering all the worst-flood hit areas of India and extended Bangladesh. And when counted in numbers this also means that nearly 200 million people in India and 40 million people in Bangladesh will now be able to receive alerts from Google’s flood forecasting system.

Furthermore, Google is also working on improving the accuracy of its forecasts and has even changed the way such alerts appear on Android devices owned by people from the similar region. Apparently, Google says that it has sent over 30 million notifications to users who are living in the red zone so far.

Lately, Google has been taking a lot of interest in creating a system that should alert users about any potential natural disasters or emergencies like floods, wildfires, and earthquakes. The company is making its aim possible with the help of the Public Alerts program and just last month they have also introduced a new feature for the Android devices that can make the mobile phone turn into a seismometer only to detect the vibrations of earthquakes and alert users about it right on time.

However, when it comes to flood forecasting, Google isn’t taking any help from the devices that consumers own. Instead, its machine learning model makes predictions based on a mixture of historical and contemporary data associated of rainfall, river levels, and flood simulations.

But with that being tried and tested, Google also says that it is working on new models to improve the forecasts in times to come. As a result, the company’s latest forecast model has double lead time as compared to its previous system and even provides information about the depth of the flood - with a margin of error of 15 centimeters only.

There was also a study conducted by scientists from Yale in Ganges-Brahmaputra river basin on the accuracy of Google’s forecast system and the results were really positive with 70 percent of people receiving the flood alert even before the floodwater arrived 65% managed to take some action on time. The researchers were also glad to see that even in areas where the literacy rate is low and poverty is high, a lot of them still managed to act on the information they received.

But with all the positives, there is one big problem that is still left to be solved and that is the lack of access to smartphones in these areas and lack of trust in technology as well. There were also some respondents in the survey who said that they would prefer trusting such a warning more if it was to be announced by a local leader.

As Google has taken this response seriously, therefore, the company is already in talks with the International Federation of Red Cross and Red Crescent Societies to find a solution to this problem. They are hoping to share more accurate forecasts with such organizations that can then develop a network to spread such information in areas where it matters.



Read next: Google is extending Secure DNS to Chrome 85 for Android soon
Previous Post Next Post