On Tuesday, while Hurricane Milton barreled toward Florida’s coastline, officials were urging evacuations across the storm’s predicted path.
“There is concern that if you want to be out before Wednesday, you should be well underway now,” said Andrew Schroeder, Direct Relief’s vice president of research and analysis. Milton’s fierce winds, small eye, and rapid intensification alarmed meteorologists and emergency managers this week, dropping down to a Category 4 hurricane early Tuesday but gaining strength as the day wore on.
But while recent data were showing new evacuations underway, Schroeder was concerned that people were leaving later and less urgently than expected, given the severity of the storm.
“Delayed evacuation is one of the most significant causes of morbidity and mortality in the event of a storm,” Schroeder said.
Using anonymized and aggregated data from Meta’s Data for Good, CrisisReady, a research-response initiative at Harvard and Direct Relief, produces population dynamics data to track how numbers of people change over time in response to a disaster.
While these numbers don’t track individual movements, they offer useful insight into how evacuation and displacement work across populations over time.
Communities along Florida’s Gulf Coast, including St. Pete’s Beach and Madeira Beach in Pinellas County, had been heavily impacted by storm surge during Hurricane Helene. By late Monday, population levels in a number of coastal towns were as much as 60% lower.
“The fastest and most extensive evacuation really took place among coastal communities,” Schroeder said.
St. Petersburg in Pinellas County had been slower to see movement despite evacuation warnings, but he noted a 6% decrease in population late Monday — a good, if late, beginning.
In the Tampa Bay area, coastal communities like Holmes Beach and Bradenton Beach were showing as much as an 87% population reduction, but population centers like Bradenton, heavily impacted by Helene, and Sarasota were seeing increases up to 9%.
“It’s an interesting and somewhat concerning pattern,” Schroeder said.
The increased risk isn’t just to people who stay in the area: “There’s a chance for some real bottlenecking to happen,” according to Schroeder, as last-minute changes and emergency needs strain resources.
CrisisReady will continue to track population dynamics as the storm approaches.
Hurricane Helene
Across the southeastern U.S., displacement from Hurricane Helene remained widespread. Asheville, a North Carolina city devastated by the storm, had seen a 40% drop in population. While that number was slowly increasing, “we usually see a more rapid rate of return” to a city like Asheville, according to Schroeder. People usually come back quickly to search for missing family and friends and check for damage to their housing.
A quick rebound in population isn’t in itself meaningful. “A return to baseline does not mean a return to normal,” Schroeder said. But slower and later population increases, despite increasingly open roads in the Asheville area, were a noteworthy indicator of Helene’s severe impacts.
Helene’s impact spanned an unusually vast area — communities across a 600-mile stretch in the southeastern U.S. were badly impacted — and Schroeder was concerned about continuing rural displacements across much of those 600 miles.
While most rural communities in the region’s interior hadn’t experienced impacts on the same scale as Asheville’s or coastal Florida’s, deaths, injuries, and damage were widespread. Damaged infrastructure and displaced residents have resounding consequences in rural economies, where people tend to be older and less mobile, economies are more fragile, and fewer resources (like hospitals, health centers, and pharmacies) serve a larger area, Schroeder explained.
Widespread health impacts in rural communities are often delayed until “weeks or sometimes months after the storm,” Schroeder said. Public awareness frequently moves on after the most acute stage of a disaster, which makes long-term health crises less visible.
Power and mobile network outages have been widespread since Helene, complicating the numbers, but Schroeder said the impacts on social media data were likely smaller than, for example, phone communications. People affected by the hurricanes were likely to check in with family and look for community updates online.
And while misinformation about Helene and Milton has caused serious problems for affected communities and responders, Schroeder said the population metadata are unaffected. Regardless of the information social media users encounter, they’re providing reliable data about population dynamics through their metadata.