A Flood of Information

Why is AIR’s new inland flood catastrophe model needed?

The risk of damage and loss from flood is very real, and currently there are no existing tools to help quantify those risks. If you look at river flooding across the U.S. and you capture overall FEMA disaster potential, you see that floods form a significant portion of the top FEMA losses across the country. To date, the tools to quantify flood risks have been things like flood hazard maps through FEMA where you can get concentrations of your risk in certain flood hazard zones. It doesn’t give you the ability to probabilistically correlate things like precipitation, river runoff, levees, and on- and off-floodplain flood risk. That’s really the gap the AIR model is looking to fill.

What does the new model do?

It’s a fully probabilistic flood model. We have a catalogue of simulated flood events, a 10,000-year stochastic catalogue that effectively propagates the entire lifecycle of precipitation. It looks at what we call the antecedent conditions, which is the continuous build up of moisture in the soil from precipitation. It looks at more than a million miles of river networks, streams, runoffs, lakes and reservoirs, and drainage catchments, and it simulates the water flow across this entire network of rivers and streams to estimate what the damage potential could be.

Why has it taken until now to develop such a model?

Not surprisingly, when you talk about millions of miles of rivers and 335,000 drainage catchments, there is a significant volume of data to process. It’s taken a lot of time to assimilate all of this climatological, precipitation and hazard data across the level of granular detail you need to model it. We use a 30-meter digitized terrain model. This leads to a massive volume of information to assimilate in a computer and come up with the potential risks, and that’s really what our model is packaging.

What data has to be collected and analyzed for this to work?

We have data from 18 hydrological regions, 3 million miles of river, 335,000 drainage catchments, 20,000 lakes and reservoirs, a significant volume of information on levees. We’ve analyzed data on roughly 14,000 miles of levee networks across the nation. All of these have been assimilated together to come up with the model footprint.

What’s it mean for the insurer and insurance brokerage industry?

It gives them anther tool to quantify risk they may already have on their books, and it gives them another tool they can use to develop new products. Maybe they want to write a certain piece of business they’re not able to fully quantify now. They can use this model in conjunction with things like flood hazard maps to assess the overall risk and figure ways it makes it profitable to write.