RMS’ Steffi Uhlemann-Elmer and Arno Hilberts considers how advanced data can help insurers tackle the complexities of flood on both sides of the Atlantic.
Insurers in the US and Europe both face similar fundamental issues when assessing flood risk, with an overreliance on national flood maps based on historical records, incomplete flood defense information and a separation between vulnerability and hazard – all of which has dented confidence.
But now there is a democratization of high-quality flood risk data which is helping to solve problems faced by insurers.
UK: Floods in Carlisle
Carlisle in northwest England had three 200-year return period flood events in ten years between 2005-2015. Storm Desmond in December 2015 – a 1,000-year event – overtopped the city’s new £38mn 200-year return period defenses, flooding 2,100 properties. This was a significant return period underestimate. Making investment decisions on return periods based just on river gauge flow records cannot capture tail-risk adequately.
Modeling the entire hydrological cycle using the widest variety of data sets possible, with stochastic simulation is required.
What about at a national level? RMS worked with the Environment Agency to examine current UK flood defenses. With 50,000 years of simulated events, the RMS® Europe Inland Flood HD Models accommodates extreme tail-risk, to show that current UK river flood defenses prevent close to 70 percent of losses for a 1-in-5 year inland flood event, but “only” 30 percent at the 500-year return period. As with Carlisle, flood defenses must be able to protect against higher return period events. Analyzing return periods in a more holistic way using probabilistic modeling to examine tail-risk will help guide these vital investment decisions.
US: Harvey disregards flood zones
In total, 204,000 homes and apartments were flooded across Houston after Hurricane Harvey in 2017, and almost three-quarters were outside of the FEMA 100-year return period flood zone. Flood in Houston is much more prevalent than the FEMA flood zones suggest.
RMS conducted a study for a selection of ZIP codes covering 60,000 properties in Harris County. Using the RMS® US Inland Flood HD Model, we established a higher figure. In total, 73 percent of properties were outside of the FEMA 500-year flood plain and our hazard data revealed the mismatch between FEMA and the reality of flooding – 52 percent of properties within the 100-year defended return period would be flooded and 27 percent of locations classified as “off floodplain” were at risk of flooding within our RMS 100-year return period. This highlights that defining insurability of a property around FEMA flood zones, even as these are being revised with FEMA Rating 2.0, will still result in many floodable properties not being covered, and that effective modeling flags up opportunities for insurers.
Europe: The perfect storm
A few years ago, RMS conducted an experiment entitled “The Perfect Storm,” to understand the effect of antecedent conditions, where soil is saturated from previous rainfall. Two major events across Central and Eastern Europe in August 2002 and June 2013, both hit the Elbe and Danube basins, affected multiple countries and caused both major economic losses and many fatalities. Similar amounts of precipitation triggered both floods, but the rainfall in 2002 occurred over just two days whereas in 2013 it lasted for four days.
Heavy rainfall preceding June 2013 meant soil moisture was at a high level. So, what would happen if you combined the high precipitation intensity of the 2002 event with the wet antecedent conditions of June 2013? The results were striking. This “perfect” flood scenario saw peak discharges rise 20 to 50 percent, leading to higher flood depths and more extended flooding areas. An increase of 50 percent in discharge along the Elbe or Danube river corresponds to an increase in flood extent of about 300 percent. An even stronger non-linearity exists between river discharge and total damage—losses could be four times higher than August 2002 under “perfect” flood circumstan