The National Research Council (NRC) published a report in June that explored how FEMA’s National Flood Insurance Program (NFIP) appraises flood insurance in the U.S. It is a very thorough and technical document (as one would expect), that reaches ten conclusions on how NFIP can improve its pricing strategies to become, essentially, more solvent. The conclusions focus on different aspects of how a policy price is determined, including the use of loss/claims data, the risk analysis, flood modeling techniques, damage and exposure estimations, replacement costs, and other factors. One conclusion stood out for me, and I wasn’t alone.
This month, both Insurance Business America (IBA) and Scientific American (SA) ran articles with headlines declaring that there is uncertainty about the flood risk for about 750,000 homes in high-risk areas due to a lack of dependable elevation data. Can this be right? Well, the report says so (on page 27):
“Of the approximately 1 million negatively elevated structures in the NFIP portfolio, only about 240,000 have structure elevation data and have been actuarially rated.”
But, does this translate well? Here is what IBA has to say:
“FEMA has accurate elevation data on just 240,000 properties that enjoy subsidized insurance rates – just one-quarter of the homes under its purview. Even more worrying is the fact that the most at-risk homes – low-lying older buildings near high-frequency flood plains – are most often those whose information is missing.”
And here is SA’s take:
“Elevation is also supposed to be used to calculate the price of insurance. The lower a home is, generally, the more its owner pays. Without an ‘Elevation Certificate,’ FEMA can't accurately calculate how much a policy should cost because it doesn't know the true risks to that home, researchers said.”
The NRC report is talking specifically about “structure elevation data,” which is basically using a tape measure to determine the height of the first floor above the ground, and (ideally) relating it to the height of a local base flood. This is what engineers do when they create Elevation Certificates.
The articles on the report, though, suggest it’s not just structure elevation data that’s missing on those 750,000 homes, but elevation data in general. Even though the NRC report does not state that, they are right.
There is absolutely no doubt about the central importance of elevation data for flood risk. Indeed, the NRC report does spend a lot of words on elevation data. Here are two points from the report that illustrate the importance of elevation information (I will paraphrase to spare you the technical jargon):
- Half a foot of water directly corresponds to 88% of the annual expected loss from a 10 year flood. If the risk assessment messes up that half foot, the expected loss estimate will be very wrong.
- For some houses located in high risk flood areas, half the annual expected loss can be attributed to the 2-year flood. Every second year, there is likely to be a loss-causing flood, so whoever is insuring that house had better have a solid understanding of the elevation of the river, the house, and the expected water level.
Elevations are only useful, though, if they are measured from a consistent datum. This consistency is important for Elevation Certificates, since they need to depict a vertical distance between the first floor of a house and an imaginary surface that is the theoretical water level of a hypothetical flood. It’s not a trivial thing to do, and in fact these certificates are frequently wrong. (As per NRC: "A Dewberry  study found a significant number of errors in Elevation Certificates, including grossly erroneous elevations.")
Based on the above, it’s clear that using a consistent and dependable elevation model (ideally a digital terrain model [DTM] that depicts bare-Earth elevations) improves flood risk assessment. In fact, if dependable elevation data is available for the whole country (which it is), it’s possible to create flood risk scores that add a more objective aspect to determining flood risk, beyond a model (whether it’s a FIRM, or other type of flood model), which is subjective…since it’s a model. This is a recommendation not included in the NRC report that could prevent future headlines about 750,000 homes being at unknown risk or having missing data.