When Hurricane Sandy slammed into New York City in October 2012, floodwaters swallowed subway tunnels, destroyed thousands of homes, and killed 43 people across the city. More than a decade later, a peer-reviewed study published in Science Advances concludes that roughly half of New York City’s population, approximately 4.4 million people, now lives in areas exposed to extreme flood damage. The finding, drawn from an analysis of coastal cities stretching from Texas to Maine, places New York alongside New Orleans as the most at-risk major city on the U.S. Gulf and Atlantic coastline.
What the research found
Researchers evaluated 16 distinct risk factors, including building age, elevation, proximity to the coast, and prior flood losses, and ran the data through three separate artificial intelligence models to estimate exposure across dozens of U.S. coastal cities. Rather than relying on theoretical projections alone, the team anchored its analysis in FEMA’s historical damage database, which catalogs documented losses from past flood events.
New York City’s combination of dense population, aging infrastructure, and low-lying geography pushed it to the top of the risk rankings. According to the Associated Press, the study singled out New York and New Orleans as facing the most alarming profiles among all cities examined. Where a smaller coastal town might lose scattered properties, New York stands to lose entire neighborhoods’ worth of housing, transit lines, and commercial corridors in a single extreme event.
The underlying dataset, hosted on the Dryad research repository, is open-access, meaning independent researchers and city planners can verify the inputs and test their own models against the same baseline. That transparency matters. Flood risk assessments often depend on proprietary insurance models that local governments cannot interrogate, limiting their ability to contest premiums or design targeted protections.
Why coastal flooding keeps getting worse
A separate study published in Nature, titled “Sinking coastal cities” and led by Virginia Tech geoscientist Leonard Ohenhen and colleagues (2024), helps explain the mechanics behind the worsening trend. Land subsidence, the gradual sinking of ground surfaces, is compounding sea-level rise along the Eastern Seaboard. The study found that cities such as New York, Baltimore, and Norfolk are subsiding at rates of roughly 1 to 2 millimeters per year in some areas, with localized spots sinking faster, a pace that can effectively double the relative impact of ocean rise on the ground. Even modest increases in absolute sea level translate into significantly larger flood zones when the land beneath buildings and roads is simultaneously dropping.
That sinking is not a distant forecast. It is an active process already reshaping risk calculations for infrastructure that was considered safely elevated when it was built decades ago. Subway tunnels, electrical substations, wastewater treatment plants, and major roadways in New York City now sit within expanding flood boundaries that their designers never anticipated.
The danger extends well beyond waterfront properties. Storm surges and heavy rainfall push floodwaters inland through drainage systems that were never engineered for the volumes they now receive. Low-lying sections of Brooklyn, Queens, and Staten Island have experienced repeated flooding in recent years, and the Science Advances research suggests those patterns will intensify as physical drivers compound.
Compounding hazards make the problem harder to manage. High tides, intense rainstorms, and storm surge can overlap, overwhelming sewers and backup pumps even when no single factor looks extreme on its own. As climate change increases the frequency of heavy downpours, these collisions become more likely, stretching emergency response systems and complicating evacuation planning.
What the city has done, and what it has not
New York City has not stood still since Sandy. The city committed billions of dollars to coastal resilience projects, including the East Side Coastal Resiliency project along Manhattan’s Lower East Side, which is raising parkland and installing flood walls to protect against storm surge. The Department of Environmental Protection has also expanded green infrastructure, adding rain gardens and permeable surfaces in an effort to absorb stormwater before it overwhelms the sewer system.
But the scale of completed protections remains small relative to the scope of the threat. Large stretches of the Brooklyn and Queens waterfronts, as well as much of Staten Island’s eastern shore, lack the kind of engineered barriers that the Manhattan project provides. Federal funding for coastal resilience has been inconsistent, and the city’s own capital budget faces competing demands from housing, transit, and schools.
No official response to the Science Advances study from New York City’s emergency management agencies or elected leadership has been publicly documented as of May 2026. That silence leaves open the question of whether current adaptation spending matches the risk the research describes.
Gaps that still need answers
Several uncertainties prevent a complete picture of what the 4.4 million figure means block by block. The study does not break down exposure by neighborhood or demographic group, leaving unresolved whether flood risk falls disproportionately on lower-income communities or communities of color. Historically, flood damage in American cities has hit hardest in areas with older housing stock, less green space to absorb runoff, and fewer political connections to secure recovery funding.
The reliance on FEMA’s historical damage database introduces a known blind spot. FEMA records capture insured and reported losses but miss damage to uninsured properties and informal housing. The Government Accountability Office has repeatedly flagged the National Flood Insurance Program’s data gaps, noting that the program’s loss records undercount damage in communities with low insurance uptake. In a city where significant portions of the housing stock are older, rent-regulated, or occupied by tenants without renter’s insurance, the true historical damage baseline is almost certainly higher than federal records reflect. That gap means the AI models trained on FEMA data may actually understate the risk.
The population math also deserves scrutiny. New York City’s population fluctuates with migration, housing costs, and economic cycles. Whether the researchers used decennial census counts, American Community Survey estimates, or another demographic source is not specified in available descriptions of the study. That ambiguity does not undermine the broad conclusion that millions face serious exposure, but it complicates efforts to translate the percentage into specific planning targets for housing, hospitals, or transit systems.
There is also limited public detail on how the 16 risk factors were weighted for different urban contexts. Elements like building age, elevation, and prior damage interact in ways that are highly local. Without a neighborhood-level breakdown, residents and officials are left with a powerful but coarse warning rather than a map they can use to prioritize street-by-street investment.
What the research demands of New York
Taken together, the Science Advances study, the Nature subsidence research by Ohenhen et al., and FEMA’s own loss records point in the same direction: millions of New Yorkers already live in areas where extreme flooding could cause catastrophic damage, and the physical forces driving that risk are strengthening. The precise count of people exposed and the neighborhoods that will bear the greatest losses remain open questions requiring more granular analysis.
But the absence of block-level precision does not erase the core signal. For city planners, the study is less a distant forecast than a stress test of current policy. For the 4.4 million residents inside the risk zone, it raises a more immediate question: how much damage is the city willing to absorb before the next major storm makes the decision for it?
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
