Get VnExpress first in Google Search 
See more of the news you trust. Make VnExpress a preferred source to prioritise our updates in your Google search results
On the morning of Sept. 17, freshman Duy Phuc rode his motorbike from home in Ha Dong to the Hanoi National University of Education.
Not knowing where to find flood updates, he followed his usual route.
On To Huu Street, floodwaters rose above the vehicle's wheels and stalled the engine, forcing him to walk it back. A passing car sent up a wave of water that knocked him and his motorbike over.
"If I had known this section was flooded, I would have taken another route or asked to be excused from class," he says.
Vu Anh, 52, of Long Bien Ward, now checks social media groups whenever he needs to travel during heavy rain. But with message boards flooded even more than the roads, he struggles to determine which roads are usable.
Referring to the same route, one person might report deep flooding while another might claim traffic is moving normally.
"There is a lot of information, but you don't know which reports are accurate," Anh says.
Anh used to depend on HSDC Maps, a flood-warning map developed by the Hanoi Drainage Company, but found that reports of newly flooded areas did not appear quickly enough to be of use.
He recently tried the "Rain and Flood Map" feature on the iHanoi app. During the rain on Sept. 17, however, many users said some areas experiencing rain and flooding appeared dry on the map.
The mini app was removed from iHanoi within just 14 days.
For Xuan Ninh, 23, a ride-hailing driver, the most useful warnings come from VOV Traffic and a group chat with fellow drivers. When drivers encounter a flooded section, they share its location and photos so that those following them can avoid it.
After once becoming trapped at an unreported flooded spot and having to cancel a ride, Ninh began checking the group chat or looking for livestreams from flooded locations before accepting a booking.
"I just hope for something that tells us which roads are passable and which ones to avoid, instead of finding out only when we arrive," he says.
 |
| Vehicles travel along a flooded street in Hanoi's Tay Mo Ward as a car is towed after its engine stalled in the water, Sept. 17, 2026. Photo by VnExpress/Hoang Giang |
The lack of information prompted Nguyen Dinh Mau, an artificial intelligence engineer at the Japan Advanced Institute of Science and Technology, to seek a solution.
Mau returned to Hanoi in early September amid a bout of heavy rain. On his way from Thai Ha Street to 19-8 Hospital to visit a relative, he rolled up his pants and waded through knee-deep water past stalled vehicles.
He realized that popular traffic apps could warn users about congestion, but flood information remained scattered across news reports and social media.
In four days, he built Nuoc Len ("Rising Water"), a website that aggregates flood information.
The system collects data from news outlets, VOV Traffic, meteorological sources, social media, and user reports. AI identifies locations and flood levels in the reports and plots them on a map.
To filter out false reports and outdated images, the system cross-checks multiple reports from the same location, compares them with news reports and prioritizes data from government agencies, local authorities and specialized organizations, Mau explains.
He launched the website on Sept. 17 as Hanoi was experiencing heavy rain, and it attracted hundreds of thousands of visits.
He acknowledges that the current version only shows which locations are flooded and cannot yet automatically recommend alternative routes and that information from the public and social media may be inaccurate, and so the website can only act as a reference.
Associate Professor Luu Duc Hai, director of the Institute for Urban Research and Infrastructure Development, says the fact that people have to scramble for information whenever heavy rain hits reveals a lack of real-time data and coordination among authorities.
Current flood-monitoring systems do not have enough sensors, agencies do not synchronize their data, and community reports are not being used effectively, he says.
He wants the city Department of Construction to develop a shared data platform that integrates flood maps with the intelligent transportation system.
The Hanoi Drainage Company should provide data on rainfall, floodwater depth, and how long the water takes to recede, and the traffic police should combine flood data with the camera system to issue warnings and redirect traffic before motorists reach affected areas, he says.
An effective map should show water depth, the extent of flooding, the time of the latest update, whether the water is rising or receding, and which types of vehicles can pass, he says.
Eventually, the system should recommend alternative routes and accept verified reports from residents, he says.
"A static flood map provides information only when it is already too late. A dynamic map must help people decide whether to continue, stop, or take another route."
Severe flooding in Hanoi’s Thai Ha Street disrupts traffic on Sept. 17, 2026. Video by Huy Manh
Mau says addressing the flood-information problem requires three elements working together: technology capable of issuing early warnings and automatically finding alternative routes; flood data connected to transportation and urban infrastructure systems; and cooperation between government agencies, technology companies, and community platforms.
A personal project cannot easily deliver such a system without open data, mapping APIs and server infrastructure, he adds.
