UK: Flood modelling for large and complex risks
By Ian Millinship
Large and complex risks, such as industrial sites, shopping centres and caravan sites, are common in UK (re)insurance portfolios. These sites are often prone to flood and yet can be particularly difficult to model as their geocoded coordinate locations tend to be a poor representation of both the actual location and the extent of the site. In order to provide the most effective loss assessments, JBA’s UK Flood Model has a new approach to provide the most accurate view of risk.
Why are these sites prone to flooding?
Large industrial and commercial sites can be prone to both river and surface water flooding as they are typically flat and/or close to watercourses. While newer commercial sites are built with sustainable urban drainage systems, they still have large impermeable surfaces, leaving sites prone to localised surface water flooding. Additionally, industrial sites are frequently located adjacent to the coast or rivers as water is essential for processing and cooling, leaving them vulnerable in the case of a large event.
Similarly, caravan sites can be a headache for portfolio managers, despite often being seen as an innocuous risk- they also tend to be located by the coast and adjacent to watercourses. For caravan sites, the scenic appeal masks the fact that permeant dwellings would be denied planning permission in these locations due to their high flood risk. There have been incidents of total loss and tragically loss of life at caravan sites in continental Europe over the last few years.
Why are these sites hard to model?
These sites can be difficult to model within a catastrophe model or risk pricing tool for several reasons. Typically, these sites are sprawling and the total insured value (TIV) is not evenly distributed across the site. Even when the risk is geocoded to a high degree of precision, often this point is the front gate or office of the site. This makes estimating the hazard at the critical part of the site challenging. Furthermore, no two sites are the same, with each having a different layout depending on their function and the site profile. For more robust sites, the material fabric of the buildings may not be significantly damaged by inundation, but production will be affected, and business interruption will occur. These factors can make estimation of loss in the event of flooding inaccurate.
How does the JBA UK flood model deal with these sites?
In addition to building footprint data for residential and commercial risks, which we explored in our previous blog, JBA’s UK Flood Model includes detailed site boundaries for industrial and caravan sites. Over 19,000 sites are included in the model. Specific site modelling means wherever the coordinate falls within the site, the uncertainty in location and variability in depth of water across the whole site will be accounted for.
What impact can this have on my losses?
Given the high values associated with these sites, inaccurate assessment of flood risk can bias portfolio level results. This could lead to unexpected surprises if a site thought to not be at risk is affected by a major event. This notably occurred in 2015 when the McVities factory in Carlisle was flooded during Storm Desmond, resulting in £12.7m in direct damages and a further £3.6m in business interruption claims (Global Reinsurer). Further afield, the Fukushima nuclear power station was thought to be safe behind its 5.7m sea wall, which was topped by a 10m high tsunami in 2011 (Munich Re).
Finally, as industrial sites tend to be clustered, a single event may affect multiple sites and lead to unexpectedly high accumulated losses. During the Thailand 2011 event, several sites along the Chao Phraya river were affected by flooding, resulting in an estimated $11.1bn in insured losses (Lloyds).
Accurately understanding the risk across the entire site and the likelihood of whether several sites being affected together is therefore key to assessing and managing the risk.
Is this page useful?Yes No Report an issue on this page
Thank you. If you have 2 minutes, we would benefit from additional feedback (link opens in a new window).