Neat Averages Can Hide Urban Fire Hotspots

When a city looks at its fire statistics, it sees an average. That average tells a story—about progress, decline, and the effectiveness of public policies. Yet averages, as they often do, can smooth away the differences that matter most. They draw a single line across a landscape filled with peaks and valleys, turning a complicated reality into a neat narrative. A new study from the NYU Tandon School of Engineering suggests that in New York City, that neat narrative can conceal a much more varied picture from one neighborhood to the next.

The research, published in Fire Safety Journal, analyzed more than 411,000 structural fire incidents recorded by the FDNY between 2013 and 2023. Rather than examining the data only at the city or borough level, researchers divided the information across approximately 2,200 census tracts—small statistical subdivisions generally covering several thousand residents. Hannah Bonestroo, a doctoral student in Urban Systems who led the research with Assistant Professor Augustin Guibaud, explained that when cities examine inequality or allocate resources at the neighborhood level, different patterns can emerge that would otherwise be overlooked.

At the citywide level, the story appeared relatively straightforward. Structural fires declined by 11.3 percent over the decade—a clear improvement. But that average concealed significant variation underneath it. When the researchers examined the data at the census-tract level, they found hotspots where fires actually increased over the ten-year period. Eastern Queens and Staten Island were among the areas moving against the city's overall trend.

Response times revealed a different pattern. Instead of being concentrated in particular pockets, changes in response times formed broad clusters of neighborhoods experiencing similar trends. The average response time increased by 30 seconds over the decade, from 4 minutes 6 seconds to 4 minutes 36 seconds—longer than the four-minute travel-time benchmark referenced by the National Fire Protection Association. Guibaud noted that in a fire, every second matters. Thirty seconds may not sound like much, but it can give flames enough time to intensify and smoke to fill a space, potentially leaving the first arriving crew facing conditions beyond what it can safely manage alone.

The findings led to a second insight: fire frequency and response time capture largely different dimensions of risk. Areas with the highest number of fires tended to have faster response times rather than slower ones. Bonestroo interpreted this as a reassuring indication that FDNY resources were reaching places where fires occurred most frequently. But because the two measurements moved independently, she and Guibaud developed a Dual Risk Index designed to identify areas that might not stand out based on fire frequency or response time alone, but that scored highly when both factors were considered together.

The index combines fire frequency—which reflects the likelihood of a fire occurring—with response time, which influences how severe a fire can become after ignition. When mapped across the city, areas with higher combined risk clustered in Manhattan, the Bronx, and parts of the outer boroughs, while central Brooklyn and Queens received lower scores. When tested against locations of multi-alarm fires—the most serious incidents—the index performed somewhat better than either individual measure at identifying higher-priority areas.

The researchers emphasized that the study was not designed to tell the FDNY where it should reallocate resources. Instead, it was intended to identify neighborhoods that may not follow the city's overall trend and therefore warrant further investigation into why their patterns differ. Bonestroo and Guibaud plan to test the index further, while also noting the need for better publicly available data on fire outcomes. For now, the study serves as a reminder that in a city made up of thousands of blocks, a single number can never tell the whole story.

AI Image Disclaimer: The images in this report were generated using artificial intelligence tools and do not depict real locations or individuals.

Sources: NYU Tandon School of Engineering, Fire Safety Journal, Phys.org