After Hurricane Harvey, some Texas residents, politicians and scientists are wondering whether the whole U.S. system for predicting floods is any good.
The storm's deluge flooded parts of southeast Texas that had rarely, or never, been underwater before. Some areas got more than 50 inches of rain in a few days. "When the numbers started coming in it was a little scary," says Matt Zeve, the director of operations for the Harris County Flood Control District, which includes Houston.
Standing at a bridge over White Oak Bayou, one of the many streams that crisscross Houston, he points to pink hash mark about 20 feet above the stream bed. "This is how high the water got during Harvey at this location," he says.
Events like these are supposed to be rare, the kind that happen only once each century, or even once every 500 years. Maps developed by the Federal Emergency Management Agency and states are supposed to help predict where floods are likely to occur. If you own property inside the mapped flood zone, you usually have to buy flood insurance from FEMA.
But, in Texas, areas way outside of those mapped flood-plains have flooded multiple times in recent years. Before Harvey, there were major floods in the Houston region in both 2015 and 2016.
"You know, we've had three 500 year events in two years," County Judge Ed Emmett, who helps lead recovery efforts as Harris County's chief executive, said at a press conference after Harvey. "Does that mean our definition of a 500 year event is wrong?" Clearly, he added, "we've got to go back and look at what our flood plains are."
Some scientists already have. Flood expert Samuel Brody of Texas A&M University studied more than 30 years of floods in Texas. He found that about half of the insurance claims made after floods in Houston were for properties outside the mapped flood plain. In some parts of Texas, the rate was as high as 80 percent.
And Harvey looks just as bad, says Brody. "We're finding neighborhoods that are miles away from any FEMA-defined flood-plain, and every house is flooding," he says, and, "it's not just flooding once in these epic events. These are chronic, repetitive events."
Scientists in other parts of the country are finding the same thing. In Baton Rouge and Chicago huge numbers of properties outside FEMA flood zones are flooding. "A 100-year boundary is not going to capture the amount of risk and impacts from flooding now, and certainly not moving forward," says Brody.
In Houston, Brody chiefly blames development for the change in flood patterns. Pavement, rail lines and other structures trap water in a city that's relentlessly flat, he says.
Another problem with Houston's flood maps is that they're out of date. They're based on rainfall data compiled up to 1994, which also inform engineering standards for construction of buildings, roads and bridges — what engineers call "design standards."
Scientists at the National Oceanic and Atmospheric Administration point out that Houston has seen numerous huge storms since the 1990s. Sanja Perica at NOAA's Office of Water Prediction oversees an effort to update the nation's rainfall data using better methods and a longer rainfall history. She has factored in rainfall numbers from those recent storms and found that, for some parts of Texas, rainfall rates in big storms are underestimated.
The updated rainfall estimates, and the new flood maps they will enable, aren't ready for publication yet. But, Perica says, "we looked at preliminary estimates in the Houston area, [and] they will change significantly" when NOAA's new numbers are published.
How significantly? Her team calculates that big storms in Houston actually drop 30 to 40 percent more rainfall than they used to. "A 30 to 40 percent increase in precipitation estimates will probably affect flood estimates as well. A lot." Perica says.
NOAA says it will deliver the new rainfall data to Texas authorities in the next few months.
The other problem with flood maps is that they don't consider the effects of climate change. A warmer planet means rainfall patterns are changing, and that makes a flood plain a moving target. "We've always figured out the probabilities of flood events by looking back at the past," says Gerald Galloway, a flood scientist at the University of Maryland. "That doesn't work any more."
"When somebody says, 'This is a hundred year flood,' it's a guess, it's a concept. We know now that we're dealing with a new world," he says.
Galloway and Brody are part of a team of scientists advising Texas authorities on how to prepare Houston for more big storms. The new information could lead the city to build more reservoirs and floodwater diversions. New highways and houses may need to be built higher. More people will have to buy flood insurance.
And it will all cost a lot of money.