Last week we examined the impact of the recent U.K. floods on the re/insurance industry. This leads us to ask what we should expect in the future. The U.K. December floods provided a useful case study on the performance of modelling technology, and perhaps provide a touch point to help assess how reliable they are currently.
How Frequently May We Expect Such Extreme River Flows to Hit Urban Areas?
To help address this question let us turn to flood mapping and specifically the estimation of extreme river flows. If we assume that the return period of river flows in Carlisle in 2005 and in 2015 were both a 1-in-200 probability, the probability of getting two such flows in a 10-year period is around 0.11%.
This is a very small chance.
This raises the question over how robust the flow records are and whether we have enough historical data to effectively estimate such extremes. But growing scientific consensus suggests, given natural climate variability and urbanisation trends, such record river flows may occur more frequently in future.
This should be put into context. There are over 1,500 stream gauges in the U.K., measuring volumetric discharge (or ‘flow’). By making a few assumptions, the chances of getting two locations, such as Carlisle, experiencing extreme flows in a 10-year period is over 80% – so a very high chance. If extreme rainfall is more likely going forward, then it follows that we should not be surprised to see such extreme river flows every few years.
Figure 1 (above) illustrates the 2015 winter rainfall (on left-hand side) compared to the long-term average, highlighting the above-average rainfall impacting the northeast. On the right-hand side, the figure depicts, in red, the river-flow gauges where records were set and compares it with few representative gauges across England and Wales. The right-hand chart demonstrates that for many other parts of the country, the flow discharges were within the expected bands and that the December flooding was indeed very localised.
Are We in a Flood-Rich Period?
This extreme – yet localised – rainfall experience should not be surprising if, as seems likely, we consider that we are going through a ‘flood-rich’ period.
Looking at the winter rainfall from the early 1900’s, the interdecadal variability demonstrates periods where the moving average exhibits both cycles of reduced (‘flood-poor’) and elevated (‘flood-rich’) periods, respectively (blue dotted line in Figure 2).
During ‘flood-rich’ periods, the natural rainfall volatility increases and so, given that the U.K. appears to be in a ‘flood-rich’ period, the extreme rainfall recently observed should be expected more often.
The December 2015, events were driven by a phenomenon termed an ‘atmospheric river’, where a slow-moving front aligned with the atmospheric flow generated pulses of heavy rainfall that kept affecting the same area. Figure 3 illustrates that the December 2015 weather system was similar to ones generating flooding in 2009 and 2005.
In conclusion, although we experienced record-breaking localised flooding in December 2015, we should not be surprised to see such extreme river flows on any given year at a national level. In addition, the U.K. is most likely going through a ‘flood-rich’ period, which increases natural rainfall volatility, therefore increasing the likelihood of extreme rainfall events.
With extreme rainfall likely to occur more frequently, it is important that the insurance industry is prepared for these events by improving their flood risk quantification and employing the tools that can help them.
This blog is the second in a three part series: next week we will explore what needs to be done to make the industry more resilience, whether flood defences are enough to protect communities and how modelling can help to defend against these risks.
This post was originally published March 21, 2016.
Guest blogger Giorgis Hadzilacos is a Divisional Director for Willis Re, leading the Cat Modelling team for CMS Europe West, where he is responsible for delivering Willis Re’s analytical service offering to clients. Giorgis also leads the Analytics and Model Development team, specialising in developing bespoke analytical solutions for clients around the world in areas of cat modelling where models are scarce and/or uncertain. Giorgis, joined Willis Re in 2012 and is based in London.