While FEMA classifies 8.7 million properties as having substantial risk, or within Special Flood Hazard Areas (SFHAs), the First Street Foundation Flood Model identifies nearly 70% more, or 14.6 million properties with the same level of risk. This means nearly 6 million households and property owners have underestimated or been unaware of their current risk. This discrepancy exists because the Foundation uses current climate data, maps precipitation as a stand-alone risk, and includes areas that FEMA has not mapped.

When adjusting for future environmental factors like changing sea levels, warming sea surface and atmospheric temperatures and changing precipitation patterns, the Foundation's model finds the number of properties with substantial risk grows to 16.2 million by the year 2050. 

"In environmental engineering, there is a concept called stationarity, which assumes that today is going to be like yesterday, and tomorrow is going to be like yesterday," said Dr. Ed Kearns, First Street Foundation's chief data officer. "This concept used to work, but with a changing environment it's a poor assumption and no longer does. FEMA's method assumes stationarity, First Street's does not."

The model was developed by more than 80 hydrologists, researchers and data scientists from First Street Foundation; Columbia University; Fathom; George Mason University; Massachusetts Institute of Technology; Rhodium Group; Rutgers University; The University of California, Berkeley; and the University of Bristol. 

"Sophisticated investors have privately purchased flood risk information from for-profit firms for years," said Matthew Eby, executive director of First Street Foundation. "First Street Foundation has not only taken this kind of data to the next level, using peer-reviewed science, but is correcting an asymmetry of information by providing free access to everyday Americans."

The model identifies the likelihood of previous flooding by recreating 55 past hurricanes, tropical storms, nor'easters and major inland flooding events. A lack of disclosure laws in many states makes this information difficult or impossible to find. The model also calculates the current probability of tidal, storm surge, pluvial (rainfall) and fluvial (riverine) flooding for individual homes and properties.

In addition to current risk, future risk is calculated by incorporating anticipated environmental changes like sea-level rise, changing precipitation patterns, and warming sea surface and atmospheric temperatures. Technical documentation pertaining to the model development can be found here.