More than a million square miles of the continental United States face a high risk of suffering flood damage over a 14-year period, according to a new study by NC State University researchers.
In North Carolina, the risk of flood damage is largest in low-lying coastal counties, according to the study, which was published in the journal Environmental Research Letters. Dare, Hyde, Tyrrell and New Hanover counties were all among the 50 with the highest likelihood of suffering flood damage out of about 3,100 counties nationwide.
Nationally, researchers found, the total at-risk area is significantly higher than the roughly 210,000 square miles that the Federal Emergency Management Agency says has a 1% chance of flooding each year. That means much more of the country — and many more people — are at risk of being injured or suffering property damage than indicated by federal flood data.
“For us, it was critically important to break free from current hazard mitigation policies such as the 100-year floodplain, as they create a representation of risk as inside versus outside,” said Georgina Sanchez, a research associate at NC State University’s Center for Geospatial Analytics who was among the study’s authors.
The nationwide map creates what Sanchez calls “a wall-to-wall understanding of flood damage probability.”
Researchers divided the United States into 100-square-meter grids, then used a National Ocean and Atmoshperic Administration database of 71,434 flood events from December 2006 to May 2020 to evaluate which areas had suffered flooding and what made them vulnerable. They used machine learning to evaluate flooding caused by rainfall and swollen waterways, but not lakeshore, storm surge or tidal flooding.
Elyssa Collins, a doctoral candidate at the NC State Center for Geospatial Analytics who was the study’s lead author, said the top risk factors included proximity to streams, low-lying elevation and high levels of precipitation over a three-day period.
“The big thing with the machine learning approach is it doesn’t cost as much money, and we can also run these models pretty quickly, therefore we are more efficiently able to update these maps as changes occur,” Collins said.
An average flood model took about five hours to run through a computer, Collins said, potentially giving emergency planners the ability to quickly analyze how climate change scenarios or potential development could shift the reality of flooding.
Across the Triangle, the likelihood that an average 100-by-100 meter area would suffer flood damage like injuries, damage to infrastructure, damage to property or another kind of economic disruption over a 14-year period included:
- Chatham County: 51.1% (37 in NC; 1,199 nationally)
- Durham County: 57.9% (12 in NC; 545 nationally)
- Franklin County: 43.7% (82 in NC; 2,051 nationally)
- Johnston County: 48.9% (52 in NC; 1,471 nationally)
- Orange County: 48.7% (53 in NC; 1,492 nationally)
- Wake County: 53.4% (27 in NC; 952 nationally)
It makes sense that a national analysis like NC State’s would find expanded areas of flood risk, said Chad Berginnis, president of the Association of State Floodplain Managers. A 2020 report from the group found that FEMA has completed flood maps for about 46% of the nation’s coastlines and 33% of its streams.
“FEMA flood maps are a start but never an ending point when it comes to flood risk,” Berginnis said.
By using damage reports to capture the risk of a given area, Berginnis added, the NC State report likely shows risk of flooding from heavy rainstorms in developed urban areas, something floodplain professionals are grappling with in mapping.
Studies like the one from NC State can be used to show where risk exists but not necessarily the steps that should be taken to reduce it, said Berginnis, who worked as a floodplain manager in Ohio.
For instance, the study gives information about the likelihood of flood damage in a given area but not how high to build a home’s first floor to keep it dry.
“Big data, machine learning and artificial intelligence can be very helpful and I’m excited to see how it can be used in our field of flood risk management,” Berginnis said, “but we also need to make sure that we understand the practical uses of those data and information.”
Collins, the study’s lead author, said researchers will next use their models to evaluate how different greenhouse gas emissions scenarios could influence flooding across the country.
This story was produced with financial support from 1Earth Fund, in partnership with Journalism Funding Partners, as part of an independent journalism fellowship program. The N&O maintains full editorial control of the work.