An artificial intelligence startup now says it can provide that warning. The company, One Concern, has announced that it can predict whether your block will flood — and if so, by how much — five days in advance of an incoming storm.
Founded by Stanford University graduates, the startup has launched a flood forecasting product called Flood Concern meant to give leaders hyperlocal predictions of where flooding will occur, allowing them to swiftly prepare and respond. High on its roster of potential clients is the Houston area, which lost over a hundred lives and suffered billions in damage last year during Hurricane Harvey.
The startup has begun approaching city officials and leaders in Houston’s private sector about bringing the technology to the region.
“They’re interested in multiple use cases, all the way from planning to responding,” One Concern CEO Ahmed Wani said of the discussions. Texas A&M University has already partnered with One Concern in anticipation of the potential benefits for the region.
“The use of artificial intelligence is potentially a game changer,” said Tony Knap, associate director of A&M’s Superfund Research Center. “It’s a different way of looking at things.”
Artificial intelligence allows computers to look for patterns from past events to predict what will happen in the future. Predictions become more accurate as the system collects more data — the Superfund Research Center is contributing data about hazardous chemicals so that a flood analysis can also understand potential health concerns.
“The aim is to get the prediction correct,” Knap said. “And artificial intelligence is something that we don’t use and they do. So if that can inform the model … it’s good for Houston.”
[…]
Eric Berger, a meteorologist whose forecasts on the Space City Weather website drew 1 million page views a day during Hurricane Harvey, said he could imagine artificial intelligence providing realistic worst-case scenarios for incoming storm systems. But he is skeptical of One Concern’s claim that it can predict flooding on a block-by-block basis.
To illustrate his point, he described a storm he was tracking that Tuesday afternoon that would hit Southeast Texas Friday night. Most of the region would likely see 2 to 4 inches of rain, but certain pockets could receive up to 8 — and those pockets would have a chance of flooding.
But where would they be?
“Three days before this heavy rainfall event, we can say this area is ripe for rain,” Berger said. “We could say that Harris County is at a greater risk than Galveston County. But to specify it even on a city-by-city basis is not possible. … There’s not the underlying meteorological data to support it.”
Here’s One Concern’s press release. As the story notes, Google is working in this space as well, though their claims aren’t as bold. I tend to agree with Berger that the data isn’t there for predictions this granular, but I like the direction they’re going, and I hope they can provide some value now, even if it’s not quite what they hope to achieve.