A project of the NYU Furman Center, FloodzoneData.us describes the people and housing located in the U.S. floodplain.
We estimated characteristics of the housing and population in the floodplain by overlaying U.S. Census data with maps of the 100-year and 500-year floodplain created by the Federal Emergency Management Agency.
First, we calculated a floodplain weight for each Census tract — that is the share of the housing units in the Census tract that fell within the 100- and 500-year floodplain (we created a weight for the 100-year floodplain and a weight for the combined 100- and 500-year floodplain). Second, we gathered Census tract housing and population indicators published by the U.S. Census Bureau’s American Community Survey (ACS). Lastly, to calculate the characteristics of the housing and population in the floodplain, we weighted the ACS Census tract indicators with the floodplain weights.
We created floodplain weights for each Census tract by estimating the share of the housing units in the Census tract that fell within the floodplain. To estimate the location of housing units, we used 2010 decennial housing unit counts at the Census block level, the smallest geography available. To estimate the location of the 100-year and 500-year floodplain, we used the boundaries provided by the National Flood Hazard Layer (NFHL), which contains the Federal Emergency Management Agency’s (FEMA) Flood Insurance Rate Maps (FIRMs).
First, we overlaid the NFHL delineating the 100- and 500-year floodplains with 2010 Census block centroids. We then aggregated the total number of housing units in the Census blocks that intersected the 100- and 500-year floodplain. Because Census blocks nest within Census tracts, we aggregated the floodplain housing unit count up to the Census tract level.
To calculate the share of a Census tract’s housing units that were in the 100-year and 500-year floodplains, we divided the floodplain housing-unit count (described above) by the Census tract’s total housing units in 2010 (see Coverage note below).
For example, consider a Census tract consisting of 10 Census blocks with 100 housing units each. When we overlay the 100-year NFHL with the Census blocks, we find that four of the Census block centroids intersect the 100-year floodplain. Therefore, the share of the Census tract in the 100-year floodplain is four (blocks) multiplied by 100 (housing units) divided by 10 (blocks) multiplied by 100 (housing units), or 40 percent.
In order to describe the population and housing in the floodplain, we weighted indicators published by the ACS at the Census tract level by the floodplain weights (described above). Consider the Census tract in the previous example. If the ACS reported that the Census tract contained 500 housing units in multifamily buildings, we would estimate that 40 percent of those 500 multifamily units — 200 units — were in the 100-year floodplain. State and county estimates were derived from aggregating weighted Census tracts data (see Coverage note below).
For subsidized housing indicators (including Public Housing), we know the exact location of the properties. Therefore, estimates are not weighted because we are able to determine with precision whether a property is within the 100- or 500-year floodplain.
National, state, county, and Census tract level data are available for download. The files include separate indicators for the 100-year floodplain and combined floodplain, as well as a codebook with indicator definitions. Housing indicators include owner-occupied and renter-occupied housing units, size and age of residential buildings, and subsidized residential properties (privately-owned and public housing). Population indicators include racial and ethnic composition, household income, poverty status, households with children, and households with older adults.
We only derived estimates for Census tracts that fell within the approximately 2,200 counties that intersected the NFHL (71% of all U.S. counties, containing about 87% of the nation’s population as of 2011 - 2015 ACS 5-year estimates). Counties and Census tracts not covered by the NFHL, or having fewer than 100 housing units, are displayed as "No Data" on the map. For estimates at the state and national level, we compared the weighted floodplain estimates to total population and housing units, including counties not covered by the NFHL.
To estimate the subsidized housing stock in the floodplain, we used data from the National Housing Preservation Database, extracted in December 2016.
We used 2010 decennial Census data to estimate the share of housing units within a Census tract that are in the floodplain. All other indicators (with the exception of subsidized housing) are from the American Community Survey 2011-2015 5-year estimates. The state, county, and tract spatial files are from the U.S. Census Bureau.
We used the National Flood Hazard Layer (NFHL) last updated in May 2016. The NFHL covers most but not all of the U.S.; see Coverage note above.